Cogito

Revitalizing Electoral Geography
Edited by
Barney Warf and Jonathan Leib
Edited by
Barney Warf
University of Kansas, USA
and
Jonathan Leib
Old Dominion University, USA
© Barney Warf and Jonathan Leib 2011
All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system or transmitted in any form or by any means, electronic, mechanical, photocopying,
recording or otherwise without the prior permission of the publisher.
Barney Warf and Jonathan Leib have asserted their right under the Copyright, Designs and
Patents Act, 1988, to be identified as the editors of this work.
Published by
Ashgate Publishing Limited Ashgate Publishing Company
Wey Court East Suite 420
Union Road 101 Cherry Street
Farnham Burlington
Surrey, GU9 7PT VT 05401-4405
England USA
www.ashgate.com
British Library Cataloguing in Publication Data

Revitalizing electoral geography.
1. Electoral geography. 2. Electoral geography--Europe.
3. Electoral geography--United States.
I. Warf, Barney, 1956- II. Leib, Jonathan.
324.6-dc22
Library of Congress Cataloging-in-Publication Data

Warf, Barney, 1956-
Revitalizing electoral geography / by Barney Warf and Jonathan Leib.
p. cm.
Includes index.
ISBN 978-1-4094-1072-0 (hbk) -- ISBN 978-1-4094-1073-7 (ebk)
1. Electoral geography. 2. Electoral geography--Case studies. I. Leib, Jonathan. II. Title.
JF1001.W37 2011
324.9--dc22
2011011624
ISBN 9781409410720 (hbk)

ISBN 9781409410737 (ebk)
II and bound in Great Britain by

Printed
TJ International Ltd, Padstow, Cornwall.
Contents
List of Figures vii

List of Tables ix
List of Contributors xi
Part I Conceptual Foundations
1 Introduction 3
2 On the Shores of the “Moribund Backwater”?:

Trends in Electoral Geography Research Since 1990 9
Jonathan Leib and Nicholas Quinton
Part II Electoral Geography Outside of the US
3 Electoral Systems, Geography, and Political Behaviour:

United Kingdom Examples 31
Ron Johnston and Charles Pattie
4 Spatial Regression for Electoral Studies:

The Case of the Italian Lega Nord 59
Michael Shin and John Agnew
5 Elections and Cultural Political Economy:

The Political Geography of the Bloque Nacionalista Galego
in the Galicia Autonomous Community 75
Erinn P. Nicley
6 Place, Identity, and Taiwan’s Presidential Elections, 1996–2008 97

Daniel McGowin
7 The Power of Definition:

The Puerto Rican Diaspora and the Island’s Status Referenda 117
Luis D. Sánchez-Ayala
vi Revitalizing Electoral Geography
Part III Electoral Geography in the US
8 Class, Ethnicity, Religion and Place in the 2008

US Presidential Election 133
Barney Warf
9 The Emerging Geography of How Americans Cast their Ballots 157

Toby Moore
10 The 2008 US Presidential Primary Elections in Geographic

Perspective 175
Fred M. Shelley and Heather Hollen
11 Electoral Alignments and Place-Based Cleavages in Statewide

Votes in Alabama 195
Nicholas Quinton and Gerald R. Webster
12 Sexual Politics and the Sunshine State:

Electoral Geographies of Florida’s Marriage Protection Amendment 219
Thomas Chapman
Index 233
List of Figures
2.1 Number of Journal Articles on Electoral Geography Published,

1990–2007 11
2.2 Number of Articles on Electoral Geography Published by Journal,
1990–2007 12
2.3 Country of Study in Journal Articles on Electoral Geography,
1990–2007 13
4.1 Support Obtained in the 2008 National Election for the Partito
Democratico (PD), Popolo della Libertà (PdL) and Lega Nord (LN)
at the Provincial Level 61
4.2 “Hotspot Maps” Showing Local Morgan’s I Score for Each Party
in the 2008 Italian National Election 62
4.3 2008 Electoral Performance of the Lega Nord Across Lombardy 66
4.4 Geographically Weighted Regression Residuals Across Lombardy 69
4.5 Geographically Weighted Regression Map of Local Unemployment
Coefficients Across Lombardy 70
4.6 Geographically Weighted Regression for the Association Between
Foreigners and Lega Nord Support 71
4.7 Map of Local Coefficients for the Relationship Between
the Self-Employed and Lega Nord Support 72
5.1 BNG Election Support Map, 1993 (%) 83

5.2 BNG Election Support Map, 1997 (%) 83
5.3 Traditional Primary Sector Economic Activities in Galicia 86
6.1 Map of the Regions of Taiwan 111

6.2 Map of the 1996 Presidential Election 111
7.1 Status Referendums in Puerto Rico 120

7.2 Distribution of Puerto Ricans in the United States, 1990–2009 124
8.1 Percent of Votes for Obama by County, 2008 133

8.2 Percent of Votes for McCain by County, 2008 134
8.3 Household Income as a Percent of the National Average 144
viii Revitalizing Electoral Geography
8.4 Results of Bivariate GWR of Support for Obama and Average

Household Income without Controls for Ethnicity and Religion 144
8.5 Independent Effects of Household Income on Support for Obama,
Controlling for Ethnicity and Religion 145
8.6 Results of Bivariate GWR of Support for Obama and White %
of County Population without Controls for Income and Religion 146
8.7 Independent Effects of White Support for Obama, Controlling
for Other Ethnicities and Religion 147
8.8 Results of Bivariate GWR of Support for Obama and Black %
8.9 Results of Bivariate GWR of Support for Obama and Latino %
8.10 Catholics as % of Total Adherents 149
8.11 Independent Effects of Catholic Support for Obama, Controlling
for Income and Ethnicity 150
8.12 Baptists as % of Total Adherents 150
8.13 Independent Effects of Baptist Support for Obama, Controlling
for Income and Ethnicity 151
9.1 Percentage of Votes Cast In-person 162

9.2 Map of Early Voting 163
9.3 Map of Same Day/Election Day Voting Rates 165
9.4 Map of Voting Technology 168
10.1 Map Showing the Results of Missouri Republican Primary, 2008 180
10.2 Map Showing the Results of Oklahoma Republican Primary, 2008 181
10.3 Map Showing the Results of Virginia Democratic Primary, 2008 190
10.4 Map Showing the Results of Wisconsin Democratic Primary, 2008 191
11.1 Partisan Elections 198

11.2 Amendments 201
11.3 Results of the LISA Analysis for Partisan Elections 207
11.4 Results of the LISA Analysis for Referenda 208
11.5 Important Places in Alabama 211
11.6 Percentage of African-Americans in Alabama by County 213
12.1 The Statutory Landscape of Same Sex Marriage in 2010 222

12.2 States with Voter Approved Constitutional Language Prohibiting
Same Sex Marriage 222
12.3 Distribution of Votes for Florida’s Amendment 2 Banning Same
Sex Marriage (2008) 229
List of Tables
3.1 Results of the First Three Elections for the Mayor of London,
Indicating the Number (and Percentage) of the First Preference
Votes and the Final Number of Votes Going to the Candidates
of the Two Leading Parties 39
3.2 Allocation of Seats across the Nine English Regions for the European
Parliament Elections in 2009 44
3.3 Allocation of Seats in the European Parliament Election of 2007
for the Southwest Region of England According to the D’Hondt
and Sainte-Laguë Divisors 46
3.4 The Allocation of Seats in the European Parliament Election of 2007
from Britain According to the D’Hondt and Sainte-Laguë Divisors 46
3.5 Seats and Votes by Region at the 2007 Elections to the National
Assembly of Wales 52
4.1 Descriptive Statistics for Variables in Lombardy Data Set (N = 1,546) 67

4.2 Baseline OLS Model Estimates for Lega Nord Support in Lombardy 68
4.3 Spatial Lag Model Estimates for Lega Nord Support in Lombardy 68
7.1 Puerto Rican Population in the Island and the United States 122
8.1 Votes in 2010 Presidential Election by Household Income 135

8.2 Exit Polls of Voters by Ethnicity on Eve of 2008 Presidential Election 139
8.3 Votes in 2010 Presidential Election by White Percent of County
Population, 2008 139
8.4 Votes in 2008 Presidential Election by County Percent of Total
Adherents who are Catholic 141
8.5 Votes in 2008 Presidential Election by County Percent of Total
Adherents who are Baptist 141
8.6 Simple Correlation Coefficients between Percent of Votes for Obama
and McCain and Income, Ethnicity, and Religious Denomination 142
9.1 Top Provisional Ballot States, 2008 166
11.1 Principal Axis Factor Analysis with an Oblique Rotation (N=67) 209
12.1 Defense of Marriage Legislative and Court Actions 220

This page has been left blank intentionally
List of Contributors
John Agnew, Department of Geography, University of California at Los Angeles,

USA.
Thomas Chapman, Department of Political Science and Geography, Old

Dominion University, USA.
Heather Hollen, Department of Geography, University of Oklahoma, USA.
Ron Johnston, School of Geographical Sciences, University of Bristol, UK.
Jonathan Leib, Department of Political Science and Geography, Old Dominion

University, USA.
Daniel McGowin, Department of Humanities, Alabama State University, USA.
Toby Moore, Research Triangle Institute International, USA.
Erinn P. Nicley, Department of Geography, University of Illinois at Champaign-

Urbana, USA.
Charles Pattie, Department of Geography, University of Sheffield, UK.
Nicholas Quinton, Department of Geography, Florida State University, USA.
Luis D. Sánchez-Ayala, Department of History, Universidad de los Andes,

Colombia.
Fred M. Shelley, Department of Geography, University of Oklahoma, USA.
Michael Shin, Department of Geography, University of California at Los Angeles,

USA.
Barney Warf, Department of Geography, University of Kansas, USA.
Gerald R. Webster, Department of Geography, University of Wyoming, USA.

Part I
Conceptual Foundations
Chapter 1
Introduction
Electoral geography – the analysis of spatial patterns in voting – has a long history
and once possessed a distinguished status within the discipline of geography. From
the 1960s through the 1980s, electoral geographers were important practitioners
in the application of quantitative techniques and cartography to electoral
data at the local and national scales. This body of work, largely empiricist in
inspiration and focus, nonetheless generated a substantial literature that revealed
the deeply spatial nature of elections, including electoral redistricting, economic
and demographic factors, and the campaign strategies of parties and candidates
(Taylor 1973; Taylor and Johnston 1979; Gudgin and Taylor 1979; Swauger
1980; Morrill 1981; Archer 1988; Archer and Shelley 1986, 1988; Archer et al.
1985; Archer and Taylor 1981; Johnston 1982; Johnston et al. 1990; Martis 1982;
Martis 1989). Related works were concerned with the state, social relations, and
the socio-spatial context of ideology, in which elections are seen as arenas in
which subjects express their preferences within structural constraints ranging
from the local scale to the world system (Agnew 1996; Flint 2001; Johnston and
Pattie 2003). The vast majority of such projects focused on the US, with a few
notable examples drawn from Europe (Agnew 1995; Johnston and Pattie 2006;
Adams 2007). This literature has offered a rich, detailed portrait of the spatiality of
elections at the national and local levels; redistricting and gerrymandering; shifts
in voter preferences, turnout rates, and correlations with various socioeconomic
variables; and neighborhood effects on political behavior.
Over the last three decades, however, electoral geography has fallen into
senescence. As human geography became increasingly preoccupied with
conceptual matters approached through various forms of social theory and political
economy, electoral geographers’ obsession with techniques and data, reflecting
an unrepentant positivist or naïve empiricist outlook, and their studied neglect
of issues of theory and social context, left the subdiscipline unable to contribute
substantially to contemporary conceptual debates. Publications in this area,
accordingly, declined in number and status, and electoral geography degenerated
into its own “moribund backwater.” As well, the practitioners within the field
of electoral geography reflected an earlier time period within the discipline of
geography. While the discipline has witnessed a partial closing of its gender
inequality gap in recent decades (though not as fast and as thoroughly as we would
like), the overwhelming majority of electoral geography studies are still authored
by males.
4 Revitalizing Electoral Geography
More recently, however, a renaissance of sorts has begun to emerge in electoral

geography (Johnston 2005; Shin and Agnew 2008; Warf 2009; Webster, Chapman
and Leib 2010). Many political geographers, as well as some members in related
disciplines such as political science, have sought to bring new perspectives to the
field, countering the traditional positivism, engaging with conceptual debates, and
avoiding the overly American bias that once plagued the topic. Methodological
advances such as Geographic Information Systems have allowed electoral data
to be analyzed in novel and intriguing ways. New theoretical perspectives from
political economy to post-modernist and post-structuralist approaches have been
added to the electoral geography literature, accentuating the role of multi-scalar
explanations in the field, and injecting a concern for class, identity, ethnicity,
gender, and sexuality within electoral geography studies. At the same time, when
combined with more traditional spatial-analytic methodologies, these perspectives
have provided greater insight through newer “mixed methods” approaches to
the study of electoral geography. Moreover, the political environment itself has
become more conducive to the subdiscipline’s reconstruction, particularly in the
US, where, since the heated presidential election of 2000, academic work on
the spatiality of elections has intensified. Thus, theoretically, conceptually and
pragmatically, electoral geography is poised to re-establish itself as an important
part of the intersections between politics and geography.
This volume seeks to expedite the resurgence of electoral geography through
a series of chapters that integrate new conceptual and theoretical approaches with
a broad array of case studies from around the world. Key questions that tie the
theoretical and empirical sections together include:
1. How has electoral geography changed since the 1980s, when the last wave
of works in this subdiscipline appeared?
2. In what ways does contemporary scholarship in social theory inform the
analysis of elections and their spatial patterns?
3. How has electoral geography been reconfigured by social and technological
changes such as new requirements for voter identification?
4. As electoral geography distances itself from its previous obsession with
the United States, how has the comparative analysis of elections informed
the field?
Summary of Chapters
The chapters in this volume fall into three major categories. The first set deals
conceptually with theoretical and conceptual topics, including the evolution of
electoral geography, lines of debate, and silences and omissions. The second group
consists of case studies drawn from the UK, Italy, Spain, and Taiwan. The third
focuses on the United States; some are concerned with elections at the national
scale, while others focus on more local contexts. These chapters also deal with
Introduction 5
different types of electoral geography topics, including general and primary

candidate elections, referenda, voter eligibility issues, as well as the varied impacts
of different types of voting and registration procedures and systems.
Jonathan Leib and Nicholas Quinton begin by tracking electoral geography’s
trajectory over the past 20 years. Analyzing 224 articles published over the last
two decades, they trace major topic areas of electoral geography research, such
as voting studies, electoral systems, redistricting, and pedagogic issues, as well as
the diverse theoretical perspectives that have been brought to bear on the subject,
such as spatial analysis, political economy, and poststructuralist interpretations.
The latter, in particular, has led to a thorough reconceptualization of power within
electoral geography and studied attempts to overcome the chasm that separated
the analysis of elections from wider issues of class, gender, ethnicity, sexuality
and inequality.
One of the signs of electoral geography’s maturation is the growing number
of case studies not located in the United States. Britain, of course, has long been
the subject of attention from electoral geographers. Ron Johnston, one of human
geography’s most well known and prolific contributors, together with Charles
Pattie use the British context to examine the impacts of different voting systems,
including first-past-the-post, supplementary vote, and alternative vote, on national
and local elections there, as well as comparisons with similar systems in other
countries. Each of these approaches has strengths and weaknesses, as no voting
system is perfect, but their adoption inevitably entails careful attention to the
context and spatial specifics in which they are implemented.
In an indication of electoral geography’s growing methodological maturity,
Michael Shin and John Agnew use geographic information systems and
geographically weighted regression (GWR) to explore the electoral geographies
of Italy’s Lega Nord, a secessionist and anti-immigrant group that has recently
enjoyed rising popularity. Their case study reveals the necessity to take spatial
structures into account when conducting statistical analysis of electoral patterns.
Erinn P. Nicley introduces a cultural political economy approach to examine
the rise and fall of the regionalist political party, the Bloque Nacionalista Galego
(BNG), in regional elections in Spain’s northwest province of Galicia from 1985
to 2005. In his analysis, Nicley uses this cultural political economy perspective to
bring political economy, cultural studies, and semiotics into the study of electoral
geography, and, combined with GIS analysis, seeing these elections as moments
in time that both are influenced by and influence changing political, economic and
social life in northwestern Spain.
Traditionally, English-language studies of electoral geography have focused
almost exclusively on elections in North America, Europe, and Australia/New
Zealand. Daniel McGowin works to rectify this imbalance through his study of
Taiwan/Republic of China’s four direct presidential elections from 1996 to 2008.
He traces the hegemony of the KMT, viewed by many Taiwanese as a foreign party
that arrived from the mainland in the chaos of the 1940s. In the tussles between
the KMT and rival parties in these elections, issues of place, identity, ideology and
Taiwan’s future status become hopelessly intertwined.
Puerto Rico, caught between its roles as US Commonwealth and an autonomous
island with its own political character, is analyzed by Luis D. Sánchez-Ayala.
In this chapter, Sánchez-Ayala focuses on the multiple attempts to resolve the
island’s final status through electoral referenda: whether as a Commonwealth, a
US state, or an independent country. He focuses on a key issue within this debate
over the island’s future status: who would be considered eligible to vote in a future
referendum. Would the electoral universe for a future referendum be limited to
only those who live on the island? Would those who were born on the island but
have moved to the mainland be included? Or would those born and raised on the
mainland, but who can trace their ancestry back to the island, be also allowed to
vote? Because there are more Puerto Ricans living on the US mainland than on the
island itself, this issue is intertwined with complex questions about the construction
of Puerto Rican identity/identities both on the island and the mainland.
In the US context, Barney Warf describes the impacts of class, ethnicity, and
religion in the historic 2008 presidential election, which led to the first African-
American president, Barack Obama. Contrary to stereotypes that portray Democrats
as the party of the working class and Republicans as the party of the wealthy, he
demonstrates that Obama’s support tended to rise steadily with average household
income level. Using GWR, he concludes that, holding other influences constant,
many voters in each group tended to support Obama more strongly when they
lived outside of regions in which their ethnicity or religion dominated, perhaps
indicating a subtle geography of appreciation for difference.
Toby Moore summarizes the spatiality of several dimensions of American
voting in the 2008 elections, including same day versus election day registration,
early voting, provisional voting, voter identification laws, and voting technologies,
all of which combine to shape local and national outcomes in complex ways,
as well as emerging issues such as the impacts of votes cast by citizens located
overseas.
While numerous studies exist, both historical and contemporary, of the
geography of US presidential general elections, Fred M. Shelley and Heather
Hollen examine an understudied yet vitally important aspect of these elections, the
Presidential primary election process. Using the 2008 presidential primaries as a
case study, Shelley and Hollen examine the important geographical aspects of the
American primary election system that helped lead to the nomination of Barack
Obama and John McCain by the country’s two main political parties.
Using measures of spatial autocorrelation, Nicholas Quinton and Gerald
R. Webster analyze spatial patterns of voting patterns in multiple elections in
Alabama, both partisan contests and “culture war” referenda, demonstrating how
quantitative approaches can be combined with contextually-sensitive studies in a
“conciliatory” manner. As the culture wars flared across this deeply conservative
state, they exposed deep and persistent geographic divides among the state’s
regions in which race, ideology, and place were inseparably fused.
Introduction 7
Finally, Thomas Chapman analyzes the geographical contours of Florida’s

2008 marriage amendment referendum, in which proponents sought to place a ban
on same-sex marriage within the state’s constitution. Taking a cue from the recent
subgenre of work that invokes a critical and post-positivist approach towards
electoral outcomes and the political geographies of sexuality, Chapman examines
the electoral geographies of Florida’s referendum results through the lens of the
cultural politics of sexuality.
While diverse in numerous ways, the overall goal of the 11 chapters that follow
is to outline the myriad theoretical, conceptual and methodological perspectives and
applications that together are ushering in electoral geography’s revitalization.
References
Adams, P. 2007. Atlantic Reverberations: French Representations of an American

Presidential Election. London: Ashgate.
Agnew, J. 1995. The rhetoric of regionalism: The Northern League in Italian
politics. Transactions of the Institute of British Geographers 20: 156–72.
Agnew, J. 1996. Mapping politics: How context counts in electoral geography.
Political Geography 15: 129–46.
Archer, J. 1988. Macrogeographical versus microgeographical cleavages in
American Presidential elections: 1940–1984. Political Geography Quarterly
7: 111–26.
Archer, J. and F. Shelley. 1986. American Electoral Mosaics. Washington, DC:
Association of American Geographers.
Archer, J. and F. Shelley. 1988. The geography of US presidential elections.
Scientific American 259: 44–53.
Archer, J., G. Murauskas and F. Shelley. 1985. Counties, states, sections, and
parties in the 1984 presidential election. Professional Geographer 37: 279–87.
Archer, J. and P. Taylor. 1981. Section and Party: A Political Geography of
American Presidential Elections. New York: John Wiley, Research Studies
Press.
Gudgin, G. and P. Taylor. 1979. Seats, Votes and the Spatial Organisation of
Elections. London: Pion.
Johnson, R. 1982. The changing geography of voting in the United States: 1946–
1980. Transactions of the Institute of British Geographers 7: 187–204.
Johnston, R. 2005. Anglo-American electoral geography: Same roots and same
goals, but different means and ends? Professional Geographer 57: 580–87.
Johnston, R. and C. Pattie. 2003. Representative democracy and electoral
geography. In J. Agnew, K. Mitchell and G. Toal (eds) A Companion to
Political Geography. Oxford: Blackwell, 337–55.
Johnston, R. and C. Pattie. 2006. Putting Voters in Their Place: Geography and
Elections in Great Britain. New York: Oxford University Press.
Johnston, R., F. Shelley and P. Taylor (eds) 1990. Developments in Electoral

Geography. London and New York: Routledge.
Martis, K. 1982. The Historical Atlas of United States Congressional Districts:
1789–1983. New York: The Free Press.
Martis, K. 1989. The Historical Atlas of Political Parties in the United States
Congress: 1789–1989. New York: Macmillan.
Morrill, R. 1981. Political Redistricting and Geographic Theory. Washington,
DC: Association of American Geographers.
Shin, M. and J. Agnew. 2008. Berlusconi’s Italy: Mapping Contemporary Italian
Politics. Philadelphia: Temple University Press.
Swauger, J. 1980. Regionalism in the 1976 presidential election. Geographical
Review 70: 157–66.
Taylor, P. 1973. Some implications of the spatial organization of elections.
Transactions of the Institute of British Geographers 60: 121–36.
Taylor, P. and R.J. Johnston. 1979. Geography of Elections. London: Penguin.
Warf, B. 2009. The US electoral college and spatial biases in voter power. Annals
of the Association of American Geographers 99(1): 184–204.
Webster, G., T. Chapman and J. Leib. 2010. Sustaining the “societal and scriptural
fence”: Cultural, social, and political topographies of same-sex marriage in
Alabama. Professional Geographer 62: 211–29.
Chapter 2
On the Shores of the “Moribund Backwater”?:
Trends in Electoral Geography Research
Since 1990
Jonathan Leib and Nicholas Quinton
In the 1970s and 1980s, political geography emerged from its “moribund
backwater” (quoted in Berry 1969). In the forefront of this resurgence was the
subfield of electoral geography, as many of the leading political geographers of
the time helped to advance the subdiscipline (e.g., Taylor and Johnston 1979).
The rise of electoral geography culminated in a 1988 conference in Los Angeles
and the subsequent 1990 publication of papers from that conference in the book,
Developments in Electoral Geography, edited by three prominent political
geographers: Ron Johnston, Fred M. Shelley and Peter Taylor. However, criticisms
of major strains of work in electoral geography were also on the rise at the time,
with such charges of “methodological obsession” (Agnew 1990, 15), “rampant
empiricism” (Shelley, Johnston and Taylor 1990, 1), and a lack of concern for
social theory leveled against much of the work in electoral geography (see also
Agnew 1987, Reynolds 1990).
The purpose of this chapter is to analyze recent trends in electoral geography
research since the publication of Developments in Electoral Geography in 1990.
First, we examine whether electoral geography has declined as an important
subfield of political geography. Second, we assess trends in the topics studied
and theoretical approaches employed within electoral geography. We conclude by
assessing to what extent electoral geographers have responded to the criticisms
and critiques of the subdiscipline issued 20 years ago.
To address these issues, we have created a database of 224 electoral geography
journal articles published between 1990 and 2007. Before we discuss our survey
of the subdiscipline through these articles, we need to address how we decided
what to include and what not to include within this database.
First, we chose to examine electoral geography articles published in major
geography journals, given the ability of databases that allow us to perform a
fairly comprehensive search of these journals. We have included in our survey
human geography journals indexed in the ISI-Journal Citation Reports database,
a standard source for assessing major journals in the field. To this group, we
have added additional non-ISI journals that we know have published a variety
of electoral geography articles (such as the Southeastern Geographer). However,
we recognize that it is nearly impossible to track down every non-ISI journal

published that might contain electoral geography articles. As a result, we know we
have missed some electoral geography articles published in non-ISI journals over
the past 20 years (e.g., Ingalls and Strickland 1992). In addition, we have limited
ourselves to English-language human geography journals. We recognize that, as a
result, we have missed some electoral geography articles from journals published
in other languages (e.g., Caviedes 1993). As well, our emphasis on journal articles
means that we have not included books published in electoral geography, books
that contain electoral geography analyses, chapters in edited volumes and essays in
encyclopedias (for example, Agnew 2002b, Secor 2004, Leib and Webster 2009).
Has Electoral Geography Declined as an Important Subfield?
The first issue we address is whether electoral geography has declined as an

important subfield within political geography. Our anecdotal observations gleaned
from conferences and journals suggest that the field of electoral geography and
the number of electoral geographers has been in decline over the past 20 years.
As John Agnew (2003, 613) put it, by the late 1980s the focus on “studies of
voting in electoral geography … seemed passé to many geographers.” Writing in
2003, Agnew (2003, 613) suggested that while “much effort in political science
and political sociology is still put into electoral studies … with a few notable
exceptions, leading political geographers have moved on to other, presumably
greater, things.” It is telling that one of the leading current undergraduate political
geography textbooks, Painter and Jeffrey’s 2009 Political Geography, directs
readers to Developments in Electoral Geography (Johnston, Shelley and Taylor
1990) “for a broader survey of the field of electoral geography” (Painter and
Jeffrey 2009, 102), thereby suggesting the paucity of work conducted in this area
of study over the past two decades.
So how has electoral geography fared? By our count, between 1990 and 2007
some 224 journal articles were published on topics that we considered to be
electoral geography, or an average of about 12 per year (Figure 2.1). In terms of the
trends over time, we see a spike in the number of electoral geography articles from
1995 to 2002. Though speculative, we attribute this spike to several factors. First
are factors in terms of the publishing of the journals themselves, including guest
edited issues of journals on electoral geography topics, as well as plenary lectures
and commentaries (such as John Agnew’s [1996] and Ron Johnston’s [2002] in
Political Geography), which increase the number of electoral geography articles
in these years (though an increase in such articles and guest-edited issues may also
be reflective of an increased interest in electoral geography). Secondly, electoral
geography analyses of important political events of the mid- to late-1990s and
early 2000s also helped lead to this spike: such as post-Cold War elections in
former communist states, the US Supreme Court rulings of the mid- to late-1990s
on the use of race in redistricting, and the disputed 2000 US presidential election.
Figure 2.1 Number of Journal Articles on Electoral Geography Published,

1990–2007
In terms of journals, we found at least one electoral geography article in 28

different journals (Figure 2.2). Of these 28 journals, it is not surprising that about
half of the 224 articles were published in one journal, Political Geography, the
field’s flagship journal. The only other two journals containing at least 5 percent
of the electoral geography articles were Post-Soviet/Eurasian Geography and
Economics (with 21 articles) and the Southeastern Geographer (with 13 articles).
When we considered who was publishing electoral geography research, these
224 articles were published by 157 different authors (we assigned multiple
authored articles to the lead author) from multiple disciplines. With that said,
the disciplinary homes of these authors were overwhelmingly geography (59
percent) and political science (34 percent).
We should also note that the vast majority of geographers publishing articles
in electoral geography during this time period were male. Only five female
geographers published a single-authored/lead-authored article in the journals we
surveyed from 1990 to 2007. Therefore, only 3.9 percent of all such electoral
geography articles were single- or first-authored by a female geographer, a smaller
percentage than that of female political geographers as a whole, and far below the
36.2 percent of all geographers who are women (as measured by 2009 membership
figures from the Association of American Geographers) (AAG 2009).
Figure 2.2 Number of Articles on Electoral Geography Published by

Journal, 1990–2007
When we map out the sites of electoral geography research, the United States and
the United Kingdom dominate: 39 percent of all articles focus on the US, whether at
the local, state, regional or national levels, while 20 percent focus on the UK (Figure
2.3). This is probably not surprising given that we are looking at English-language
journals and that many of the authors publishing in these journals are either from the
US or the UK. Third, behind the US and the UK, are studies of elections in Russia,
reflecting an interest in the electoral process there since the breakup of the USSR. In
addition, starting with Martis, Kovacs, Kovacs, and Peter’s 1992 article on elections
in Hungary, we have seen an additional eight articles on elections in other former
Soviet states or their former Eastern European satellites.
However, the lack of coverage of elections in the rest of the world is a bit
startling. We counted only eight articles since 1990 dealing with elections in Asia
(three of these were about Israeli elections), and only four articles on Latin America
(with three of these on Mexico). Over this time period, we can only find three
articles about elections in Africa. Certainly there is much room for English-language
electoral geography research on the non-North American and European world.
As discussed above, about half of all electoral geography articles appear in
the journal, Political Geography. The percentage of geographers and political
scientists publishing electoral geography articles in Political Geography is
similar to the percentage of articles published in journals as a whole: 62 percent
Figure 2.3 Country of Study in Journal Articles on Electoral Geography,

1990–2007
of articles in Political Geography are published by geographers, with 33 percent

published by political scientists. We suggest there are two explanations for
the relatively high percentage of political scientists publishing articles about
elections in Political Geography: first, political scientists are still carrying out
many studies of elections (and some are taking geographic perspectives more
seriously [e.g., Gimpel, Karnes, McTague and Pearson-Merkowitz 2008]), and
second, the journal Political Geography is highly ranked in the ISI Journal
Citation Reports for political science. For 2009, the journal ranked sixth among
112 political science journals ranked by ISI. Hence for political scientists looking
for a “top 10” publication, Political Geography is a good outlet.
Topics of Electoral Geography Research
What topics have been covered under the heading of electoral geography over
the past 20 years? While the obvious answer is elections, the specific topics
covered have been far ranging: from analyzing spatial dimensions of post-Soviet
Russian election outcomes, to theoretical and empirical studies of redistricting and
gerrymandering, to suggesting methods for teaching about the American Electoral
College in middle- and high-school classrooms. Given this range, we have grouped
electoral geography studies into four broad categories which we discuss below in
more detail: vote studies, comparisons amongst and critiques of electoral systems,
discussions and examinations of the (re)districting process, and methods to bring
electoral studies into the classroom.
Before we go further, we should also mention that analysis in electoral
geography is not only carried out on a number of topics, but also across a variety of
political scales, including supranational (e.g., Laponce 2004), national (e.g., Shin
2001), and subnational (e.g., Chapman, Leib and Webster 2007). As well, electoral
geography studies are carried out over a variety of electoral contexts. While most
involve popular votes for candidates, parties, and issues, voting behavior within
elected legislatures have also been analyzed through the study of legislative roll
call voting (e.g., Webster 1992).
Vote Studies
Vote studies are likely the most recognizable form of electoral geography, and this
is the category that includes the largest volume of work. Dissecting the intricacies
of popular votes in a particular election or elections has been a common practice
for electoral geographers since the 1950s and before.
These studies focused on the votes cast for a candidate, votes cast for a party,
those on single-issue/referendum, or roll-call legislative votes. In general, vote
studies examined either the spatial distribution of compositional factors (such as
race or income) associated with electoral outcomes (Clem and Craumer 1995a,
1995b, 1996, 2000 for examples, or King 1996 for advocacy of this type of work),
or place-based contextual factors (Agnew 1994, 1995, 1996; Rasmussen 2006;
Shin 2001; Shin and Agnew 2002) and their relationship to voting behavior.
Further, vote studies, like electoral geography as a whole, have covered a
variety of topics. For example, there have been studies that highlighted regional
polarization within states or growing sentiments of nationalism and how that
related to votes cast (Agnew 1995; Giordano 1999, 2001a, 2001b). Other studies
have highlighted the difference between urban and rural voters (Clem and Craumer
1995b), or central city and suburban voters (Hodge and Staeheli 1992; Ley 1994).
Still other geographers have studied turnout (Kohfeld and Sprague 2002; Lee and
Brunn 1996; Pattie and Johnston 1998), tied economic factors to voting behavior
(O’Loughlin, Shin, and Talbot 1996; Pattie, Dorling, and Johnston 1997; West
2005), and explored the effects of elections on migration (Lo and Teixeira 1998).
This type of work has also included a series of studies that related debates over
identity including race/ethnicity, gender, and sexual orientation to electoral
outcomes (Brown, Knopp, and Morrill 2005; Chapman, Leib, and Webster 2007;
O’Reilly and Webster 1998; Webster 2000b). Thus, vote studies cover a number of
disparate topics in different ways, but all of them offered analysis of a particular
vote or series of votes.
Electoral Systems
Electoral systems, or perhaps more appropriately voting systems, are critical to

election outcomes. Voting systems are the framework through which a person’s vote
is transferred to the candidate, party, or issue of their choice. Thus, understanding
vote systems is important for electoral geography.
Electoral system research has approached the issue of voting systems from
several directions. One group of articles assessed how effective a voting system is
at translating votes into representation. So, for example, there have been studies
comparing various systems in an effort to determine which provided the highest
quality of representation, such as Arrington and Ingalls (1998), who offered an
example of this approach with an analysis of six voting systems at use in the
US based upon nine measures of feasibility, and Johnston (2007), who sought to
outline Australian exceptionalism in terms of voting practices in a similar broad
based study of voting systems. In a slightly different approach, some authors have
assessed bias between the proportion of votes cast and the proportion of seats
awarded. Johnston et al. (2002) explored this issue in Great Britain with attention
to shifting bias from the Conservatives to the Labor party. Yet another related topic
was offered by Johnston et al. (2006) in their attempt to replay the 2000 and 2004
US presidential elections under the Maine/Nebraska Electoral College system (in
these two states, rather than electoral votes being awarded on a “winner-take-all”
basis, their electoral votes are allocated based both on a “winner-take-all” basis
and by congressional district). So, the effectiveness of representation has been one
avenue of research about electoral systems.
Another avenue of research dealt with legitimacy. Vote systems must attain a
level of legitimacy within an electorate in order to gain popular acceptance and
be implemented (Arrington and Ingalls 1998). Included in our cross-section of
vote systems are those studies that focused on implementation, including the study
of public debates over changing the electoral systems in Israel (Waterman and
Zefadia 1992). In a similar vein of research, Hazan (1999) followed the process of
introducing primaries into the 14th Israeli Knesset (1996–1999), their ramifications
and geographical significance. But some authors have gone beyond implementation
to assess the legitimacy of the mechanics of the voting process itself. These authors
have studied the means through which votes are initially cast and are then counted.
Much of this work occurred in the aftermath of the disputed 2000 US presidential
election, including such topics as residual votes in Florida (Warf 2006b), in terms
of impact on the minority community (Warf 2006a), and in reference to vote tally
irregularities associated with voting technology (Leib and Dittmer 2002).
(Re)Districting
Topics of vote studies and electoral systems are closely related to that of (re)
districting. Redistricting, the “process of drawing electoral district boundaries for
the purpose of electing members to a legislative body” (Leib 2006, 404), has been
a staple of electoral geography research going back as far as the early work of Carl
Sauer (1918). Though most associated with US redistricting, overviews of the
redistricting process have not been limited to studies in the US. Broad overviews
have been published on the districting process in the UK (see Johnston 1992),
India (see Singh 2000), and Canada (see Belanger 2003).
In the US, such studies have been important in terms of examining the partisan
impact of redistricting, the impact of redistricting for effective representation
of racial and ethnic minority groups, and ways of creating and measuring fair
and effective redistricting plans. Given the role of the courts in the redistricting
process over the past 20 years, a number of these studies have examined the impact
of court decisions and case law on the political-geographic division of space.
Of these, a series of articles have provided broad overviews of the precedents
established in case law and how recent court rulings have impacted existing
ways of regulating, implementing and evaluating the redistricting process (e.g.,
Forest 1995; Forest 1996; Forest 2004b; Leib 1998; Webster 1997, 2000a, 2002).
Other US-based studies have examined specific dimensions of the redistricting
process, such as gerrymandering (e.g., Ingalls and Moore 1995; Lennertz 2000)
and malapportionment (Rallings, Johnston, and Thrasher 2004).
More than in other areas of electoral geography research, redistricting studies
have been produced by a wide array of authors, as authors from across the discipline
of geography, outside of geography, and outside of academe have contributed.
Along with political geographers and political scientists, redistricting issues attract
GIS experts, statisticians, Department of Justice professionals, and social theorists
alike (though, of course, there is the potential for overlap among these categories).
Such studies have included the use of quantitative applications to meet legal
standards for (re)districting (Cirincione, Darling, and O’Rourke 2000; Eagles,
Katz, and Mark 1999; Thill 1998), such as an evaluation of compactness (Belanger
and Eagles 2001; Flaherty and Crumplin 1992; Horn, Hampton, and Vandenberg
1993), and the use of Bayesian statistics to detect electoral manipulation (Altman
2002). In addition, there are (re)districting studies that offered a critical perspective,
such as affirmative (re)districting (Webster 1993) in establishing/maintaining
voting rights (Webster 1995, 2007). And, in a similar vein, post-structuralist ideas
of hybridity and multiplicity have emerged in research discussing communities of
interest (e.g., Forest 2004a; Leib 1998). Thus, the body of work on (re)districting
is produced by a unique web of authors within and outside of electoral geography.
Education/Classroom Methods
Finally, a small, but critical, component of electoral geography relates to praxis

in education. We have several articles that have been published in geography
journals that provide approaches to relate issues of elections and representation
to students in a classroom setting. Depoali (1993) offered various strategies to
aide in teaching middle- and high-school students about the American Electoral
College. Jones (1997) highlighted geographical perspectives on a range of issues
that include delineating electoral districts in an effort to help inform public policy
debate amongst students. Finally, Webster (2004) provided examples of the (re)
districting process to help incorporate social justice issues into the classroom.
Theoretical Perspectives within Electoral Geography Research
Topics studied within electoral geography over the past 20 years have been
disparate to say the least, ranging from poll taxes in Scotland (McCormick 1996)
to grassroots movements in India (Currie 1998). Each study, however, fits within
at least one of the three major theoretical approaches, as outlined by John Agnew
(2002a), which have dominated political geography in its post-1970 renaissance.
These theoretical frameworks are the spatial-analytic tradition, political economy,
and post-structural approaches. We outline these below.
Spatial Analysis
The first and largest set of electoral geography studies are those that are firmly
situated within the spatial-analytic tradition. Spatial analysis is typically
conceptualized as the application of statistical methodologies to quantifiable
data, though this is not necessarily always the case (Agnew 2002a). However, the
dominant manifestation from this theoretical outlook in electoral geography has
undoubtedly been quantitative in nature. The spatial analysis tradition dominated
electoral geography in the 1970s. This is not surprising, in that the rise of electoral
geography in that decade, leading the way in the resurgence of political geography,
coincided with the high point of spatial analytic methodologies within human
geography. Elections provided a “natural” area of study for geographers within
this tradition, as elections provide vast amounts of quantitative data (i.e., votes)
that can be mapped, manipulated and analyzed through sophisticated statistical
techniques.
Since 1990, electoral geography work from a spatial-analytic perspective has
included analyses of the channels of information that political discussion travels,
including friends and neighbors effects (Johnston, Jones, Sarker, Burgess, and
Bolster 2004; Kramer 1990; Rallings, Johnston, and Thrasher 2004; Sui and
Hugill 2002). There have also been explorations of a variety of political cleavages
that divide selected electorates under the spatial-analytic umbrella (Hodge and
Staeheli 1992; Perepechko, Kolossov, and ZumBrunnen 2007). One study has been
published expounding on the cube rule in search of fractality (Maloney, Pearson,
and Pickering 2003). Further, there are studies that have sought to deepen the
level of quantitative sophistication in terms of eigenfunction analysis (Thill 1998),
combating the ecological fallacy (Johnston and Pattie 2003; Owen and Grofman
1997), using Bayesian statistics to detect electoral manipulation (Altman 2002),

evaluating the significance of temporal and spatial data (Fernandez-Duran, Poire,
and Rojas-Nandayapa 2004), and forwarding the use of local area statistics (LISA)
in electoral analysis (Darmofal 2006).
Finally, there are a series of studies that employ statistical methodologies
to describe electoral outcomes. Some catalog demographic trends as they vary
through space over particular elections or groups of elections (Archer 2002;
Clem and Craumer 1995a; Clem and Craumer 1995b). Others include discussions
of normal and deviating elections as parts of broader electoral trends (Forrest,
Johnston, and Pattie 1999). These studies have also been extended into discussions
of the mechanics of large scale electoral changes, including flow-of-vote as
opposed to the uniform swing hypothesis (see Pattie and Johnston 1993).
Political Economy
Over the past 20 years, political economy approaches have been less popular within
electoral geography than studies those from a spatial analytic perspective. There
are at least three notable studies that incorporated political economy perspectives
during this time period (though having said this, there were electoral geography
studies from a political economy approach in the 1980s, when political economy
perspectives were more in favor within political geography (Oseikwame and
Taylor 1984). First, Duncan (1991) explored the geography of support and power
in Fijian elections. Next, Flint (2001) employed a world systems approach to
explain the rise of the Nazi party in Germany. Finally, Varsanyi (2005) delves into
the intersection between labor unions, undocumented immigrants, and political
control. Despite these studies, it is apparent that this perspective has received
less attention within electoral geography than it has within political geography in
general over the past two decades.
Post-Structuralism
Lastly, there are studies from post-structuralist viewpoints. These have grown in
popularity in recent years and now include studies utilizing feminist perspectives
and those explicitly dealing with the politics of identity. Pieces of note include
Webster and Leib’s (2001) study of the display and meaning of the Confederate flag
in South Carolina with consideration to the contestation of identity and “place.”
Forest’s (2004b) study of racial identity in Shaw v. Reno and Easley v. Cromartie
is another example. The formation of regional political identities in Mexico from
a feminist perspective also illustrates the point (Nelson 2006). Finally, Rasmussen
(2006) offers a critique of the 2000 Nebraska anti-gay marriage referenda in terms
of creating a decidedly non-gay “place” known as Nebraska.
Answering the Call?: Electoral Geography Research since 1990
To what extent, then, have electoral geographers answered the call for change made
in Developments in Electoral Geography? We cannot provide an answer without
contextualizing the initial call. Accompanying the growing critiques of electoral
geography in Developments and those criticisms from other sources within the
late 1980s and early 1990s was the growing engagement with post-structural
perspectives across geography. Though the critiques levied against electoral
geography were not necessarily made from explicitly post-structural perspectives,
the critiques themselves can not be separated from the social milieu of academic
geography at that juncture in time and space. Discipline-wide criticisms included
feminist and post-colonial interjections into the meta-narratives produced by both
positivist approaches and critical strands of geography. These post-structural
critiques were followed by others concerned with identity, including queer theory,
and post-structuralisms detailing heterogeneous associations, such as actor-
network theory. At present the uses of post-structuralist ideas have become a
significant trajectory of geographic inquiry in their own right, and are no longer
simply challenging positivist epistemologies at each turn.
Despite criticism, electoral geographers did participate in these broader
changes within academic geography. A critical component of geographic post-
structuralisms is that space and place is produced and power relations are
central to its production. In the sense that power is critical to spatial relations,
each of the political economy studies embodies an answer to the call for change
in electoral geographies. Spaces of capital accumulation are at the forefront of
the political debates included in these studies and each involved the exercise of
power throughout. Another body of work focused on power relations and space
and place involved in redistricting(s) in the United States. These studies focused
on redistricting designed to provide fair and effective representation for African-
American citizens. Thus, issues of political representation based upon racial and
ethnic identity were drawn to the forefront of these debates concerning issues of
delineating spaces of representation.
Another component of post-structural geographies is the multiplicity of
channels through which power travels. Thus, some of the most recognizable
examples of changing epistemologies manifested themselves in the studies of
identity and electoral politics listed above. Race, gender, and sexuality were all
brought into focus as contestations of spatialized identities as played out within
electoral processes. The diversity in these cleavages of power and their usage
to produce spaces and places leaves little doubt that post-structuralisms have
influenced the current shape of electoral geography.
Electoral geography studies that dealt with political economy, redistricting, and
identity politics each engaged another axis by which post-structural geographies
can be differentiated; contextualized and contingent power relations. The most
sustained engagement with this particular idea has been found in the work of John
Agnew (e.g., 1987, 1996, 2002b). Through the context of peoples’ lives these
works on place and its political implications have highlighted the contingent
results of electoral processes. In the same sense that the outcomes of political
debates over identity are not preordained in electoral processes, the outcomes of
place-based political debates are contingent on the application of power in varied
and unpredictable ways.
Surprisingly (or possibly not surprising), however, the most numerous
engagements with post-structural ideas by electoral geographers has been in
conjunction with the spatial analytic tradition. While post-structuralist studies
throughout the broader discipline have by and large jettisoned quantitative
methods, electoral geographers’ engagements with post-structural ideas have
embraced quantitative methods. Recent works have taken the relationship
between the post-structural ideas of power, difference, and contingency and
coupled them with quantitative methodologies. In particular recent work by John
Agnew and Michael Shin (Shin and Agnew 2002; Shin and Agnew 2008), Richard
Morrill, Larry Knopp, and Michael Brown (Brown, Knopp, and Morrill 2005;
Morrill, Knopp, and Brown 2007), as well as Gerald R. Webster, Jonathan Leib,
Thomas Chapman and Nicholas Quinton (Chapman, Leib, and Webster 2007;
Webster, Chapman, and Leib 2010; Webster and Leib 2001; Webster and Leib
2002; Webster and Quinton 2010), have made inroads in coupling post-structural
and quantitative geographies. All have deployed contexts and the capillaries that
power traveled through within these places alongside quantitative measures.
Thus, in these ways, attempts have been made to fuse post-structural ideas with
the dominant positivist forms of geographic inquiry in electoral geography,
quantitative methodologies.
Thus, has the call for change in electoral geography issued in 1990, that
the field address its rampant empiricism, methodological obsession, and lack
of concern for social theory, been answered? The answer appears to be that it
has, at least to a limited extent. Though any claims to significant change must
be tempered with the broader findings of this chapter. In actuality, the number of
articles employing post-structural viewpoints has been small compared to the total
number of articles we count as electoral geography. The paucity of engagement
with post-structural approaches should not be surprising given the penchant for
empiricism and quantitative methods shown by electoral geographers. Indeed, the
overwhelming number of articles we identified that fit into the spatial analytic
tradition shows that electoral geographers have by and large not answered the call
for change, post-structural or otherwise. Still, approaches have changed and post-
structural ideas have been employed in some accounts of electoral geographies,
even though they have occasionally come in unlikely shapes.
To what extent, then, has electoral geography declined as an important subfield
of political geography? The uses of post-structural ideas have undoubtedly been
a part of the changes in the broader practice of political geography over the
20-year time period since the 1990 publication of Developments in Electoral
Geography. At the same time, the maintenance of the quantitative tradition in
electoral geography has distanced it from recent trends in political geography
more generally. The answer then is mixed. As Murdoch (2006, 2) notes, “post-
structuralism’s endless variety means it becomes hard to draw any clear line or
boundary around post-structuralist theory or its influence.” In this sense we can
say clearly that while on the one hand electoral geography has contributed to
the discussion of post-structuralism more generally (though in its own way) and
thus remains a relevant subfield of political geography, on the other hand further
engagement with emerging trends within the discipline are necessary for it to
regain its former prominent place within political geography.
References
AAG (Association of American Geographers) 2009. Profile of the AAG

Membership, 2009. Available at http://www.aag.org/galleries/project-programs
-files/2009_Membershipprofile.pdf [last accessed: October 27, 2010].
Agnew, J.A. 1987. Place and Politics: The Geographical Mediation of State and
Society. Boston: Allen & Unwin.
Agnew, J.A. 1990. “From political methodology to geographical social theory?
A critical review of electoral geography, 1960–87.” In R.J. Johnston, F.M.
Shelley, and P.J. Taylor (eds), Developments in Electoral Geography. London:
Routledge, 15–21.
Agnew, J.A. 1994. “The national versus the contextual – The controversy over
measuring electoral change in Italy using Goodman Flow-of-Vote estimates.”
Agnew, J.A. 1995. “The rhetoric of regionalism – The Northern League in Italian
politics, 1983–94.” Transactions of the Institute of British Geographers 20:
156–72.
Agnew, J.A. 1996. “Mapping politics: How context counts in electoral geography.”
Agnew, J.A. 2002a. Making Political Geography. London: Arnold.
Agnew, J.A. 2002b. Place and Politics in Modern Italy. Chicago: University of
Chicago Press.
Agnew, J.A. 2003. Classics in Human Geography Revisited: Agnew, J.A. 1987:
Place and Politics: The Geographic Mediation of State and Society. Author’s
Response. Progress in Human Geography 27: 609–14.
Altman, M. 2002. “Discussion – A Bayesian approach to detecting electoral
manipulation.” Political Geography 21: 39–48.
Archer, J.C. 2002. “The geography of an interminable election: Bush v. Gore,
2000.” Political Geography 21: 71–7.
Arrington, T.S. and G.L. Ingalls. 1998. “The limited vote alternative to affirmative
districting.” Political Geography 17: 701–28.
Belanger, P. 2003. “Commissioned ridings: Designing Canada’s electoral districts.”
Belanger, P. and M. Eagles. 2001. “The compactness of federal electoral districts

in Canada in the 1980s and 1990s: An exploratory analysis.” Canadian
Geographer-Geographe Canadien 45: 450–60.
Berry, B.J. 1969. Review of International Regions and the International System
(by B.M. Russett). Geographical Review 59: 450–51.
Brown, M., L. Knopp and R. Morrill 2005. “The culture wars and urban electoral
politics: Sexuality, race, and class in Tacoma, Washington.” Political
Geography 24: 267–91.
Caviedes, C.N. 1993. “Estudios sobre la variabilidad espacial del voto en America
Latina.” Inter-American Review of Bibliography 43: 257–67.
Chapman, T., J.I. Leib and G.R. Webster 2007. “Race, the creative class, and
political geographies of same sex marriage in Georgia.” Southeastern
Geographer 47: 27–54.
Cirincione, C., T.A. Darling, and T.G. O’Rourke. 2000. “Assessing South
Carolina’s 1990s congressional districting.” Political Geography 19: 189–211.
Clem, R.S. and P.R. Craumer 1995a. “The geography of the Russian 1995
parliamentary election: Continuity, change, and correlates.” Post-Soviet
Clem, R.S. and P.R. Craumer 1995b. “Rayon-level analysis of the Russian election
and constitutional plebiscite of December 1993.” Post-Soviet Geography 36:
459–75.
Clem, R.S. and P.R. Craumer 1996. “Roadmap to victory: Boris Yel’tsin and
the Russian presidential elections of 1996.” Post-Soviet Geography and
Economics 37: 335–54.
Clem, R.S. and P.R. Craumer 2000. “Spatial patterns of political choice in the post-
Yel’tsin era: The electoral geography of Russia’s 2000 presidential election.”
Post-Soviet Geography and Economic 41, 465–82.
Currie, B. 1998. “Public action and its limits: Re-examining the politics of hunger
alleviation in eastern India.” Third World Quarterly 19: 873–92.
Darmofal, D. 2006. “The political geography of macro-level turnout in American
political development.” Political Geography 25: 123–50.
Depoali, K. 1993. “The Electoral-College – not just every 4 years – some exercises
in political-geography.” Journal of Geography 92: 180–93.
Duncan, C.J. 1991. “Ethnicity, election and emergency – The 1987 Fiji general-
election in the context of contemporary political-geography.” Political
Geography Quarterly 10: 221–39.
Eagles, M., R.S. Katz, and D. Mark. 1999. “GIS and redistricting – Emergent
technologies, social geography, and political sensibilities.” Social Science
Computer Review 17: 5–9.
Fernandez-Duran, J.J., A. Poire, and L. Rojas-Nandayapa. 2004. “Spatial and
temporal effects in Mexican direct elections for the chamber of deputies.”
Flaherty, M.S. and W.W. Crumplin. 1992. “Compactness and electoral boundary
adjustment – An assessment of alternative measures.” Canadian Geographer-
Geographe Canadien 36: 159–71.
Flint, C. 2001. “A TimeSpace for electoral geography: Economic restructuring,
political agency and the rise of the Nazi party.” Political Geography 20: 301–
29.
Forest, B. 1995. “Taming race – The role of space in voting-rights litigation.”
Urban Geography 16: 98–111.
Forest, B. 1996. “Regionalism in election district jurisprudence.” Urban Geography
17: 572–8.
Forest, B. 2004a. “Information sovereignty and GIS: The evolution of “communities
of interest” in political redistricting.” Political Geography 23: 425–51.
Forest, B. 2004b. “The legal (de)construction of geography: Race and political
community in Supreme Court redistricting decisions.” Social & Cultural
Forrest, J., R.J. Johnston, and C.J. Pattie. 1999. “The effectiveness of constituency
campaign spending in Australian state elections during times of electoral
volatility: The New South Wales case, 1988–95.” Environment and Planning
A 31: 1119–28.
Gimpel, J.G., K.A. Karnes, J. McTague, and S. Pearson-Merkowitz. 2008.
“Distance-decay in the political geography of friends-and-neighbors voting.”
Giordano, B. 1999. “A place called Padania? The Lega Nord and the political
representation of Northern Italy.” European Urban and Regional Studies 6:
215–30.
Giordano, B. 2001a. “The contrasting geographies of ‘Padania’: The case of the
Lega Nord in Northern Italy.” Area 33: 27–37.
Giordano, B. 2001b. “‘Institutional thickness’, political sub-culture and the
resurgence of (the ‘new’) regionalism in Italy – A case study of the Northern
League in the province of Varese.” Transactions of the Institute of British
Geographers 26: 25–41.
Hazan, R.Y. 1999. “Constituency interests without constituencies: The geographical
impact of candidate selection on party organization and legislative behavior in
the 14th Israeli Knesset, 1996–99.” Political Geography 18: 791–811.
Hodge, D.C. and L.A. Staeheli. 1992. “Social-relations and geographic patterns of
urban electoral-behavior.” Urban Geography 13: 307–33.
Horn, D.L., C.R. Hampton, and A.J. Vandenberg. 1993. “Practical application of
district compactness.” Political Geography 12: 103–20.
Ingalls, G.L. and J.L. Strickland. 1992. “No blurred edges; No crowded middle:
Votes for Jesse Helms in 1984 and 1990.” North Carolina Geographer 1: 46–
54.
Ingalls, G.L. and T. Moore. 1995. “The present and future of racial gerrymandering:
Evidence from North Carolina’s 12th Congressional District.” Southeastern
Geographer 35: 58-74.
Johnston, R.J. 2002. “Manipulating maps and winning elections: Measuring the
impact of malapportionment and gerrymandering.” Political Geography 21:
1–32.
Johnston, R.J., F.M. Shelley, and P.J. Taylor, eds. 1990. Developments in Electoral
Geography. London: Routledge.
Johnston, R.J. 2007. “Understanding Australian exceptionalism with electoral
systems-and bringing geography in.” Australian Geographer 38: 367–74.
Johnston, R.J. 1992. “Political-Geography of Contemporary Events: Redistricting
in England Revisited.” Political Geography 11: 579–87.
Johnston, R.J., K. Jones, R. Sarker, S. Burgess, and A. Bolster. 2004. “Party support
and the neighbourhood effect: Spatial polarisation of the British electorate,
1991–2001.” Political Geography 23: 367–402.
Johnston, R.J. and C. Pattie. 2003. “Evaluating an entropy-maximizing solution to
the ecological inference problem: Split-ticket voting in New Zealand, 1999.”
Geographical Analysis 35: 1–23.
Johnston, R.J., C. Rallings, and M. Thrasher. 2002. “Electoral success, electoral
bias, and Labour hegemony: Electoral system effects in English metropolitan
boroughs.” Environment and Planning A 34: 1303–17.
Johnston, R.J., D. Rossiter, and C. Pattie. 2006. “Changing the scale and changing
the result: Evaluating the impact of an electoral reform on the 2000 and 2004
US Presidential elections.” Political Geography 25: 557–69.
Jones, D.D. 1997. “The geography for life standards and American political
geography.” Journal of Geography 96: 55–60.
King, G. 1996. “Why context should not count.” Political Geography 15: 159–64.
Kohfeld, C.W. and J. Sprague 2002. “Race, space, and turnout.” Political
Kramer, D.C. 1990. “Those people across the water just don’t know our problems
– An analysis of friends and neighbors voting in a geographically-split
legislative district.” Political Geography Quarterly 9: 189–96.
Laponce, J.A. 2004. “Turning votes into territories: Boundary referendums in
theory and practice.” Political Geography 23: 169–83.
Lee, D.O. and S.D. Brunn 1996. “Politics and regions in Korea: An analysis of the
recent presidential election.” Political Geography 15: 99–119.
Leib, J.I. 2006. “Redistricting.” In B. Warf, ed., Encyclopedia of Human
Geography. Thousand Oaks, CA: Sage, 404–5
Leib, J. I. 1998. “Communities of interest and minority districting after Miller v.
Johnson.” Political Geography 17: 683–99.
Leib, J. I. and J. Dittmer. 2002. “Florida’s residual votes, voting technology, and
the 2000 election.” Political Geography 21: 91–8.
Leib, J.I. and G.R. Webster. 2009. “Electoral Districts.” In R. Kitchin and N.
Thrift, eds., International Encyclopedia of Human Geography, Volume 3.
Oxford: Elsevier, 399–404.
Lennertz, J.E. 2000. “Back in their proper place: Racial gerrymandering in
Georgia.” Political Geography 19: 163–88.
Ley, D. 1994. “Gentrification and the politics of the new middle-class.”

Environment and Planning D-Society & Space 12: 53–74.
Lo, L. and C. Teixeira 1998. “If Quebec goes … the ‘exodus’ impact?” Professional
Maloney, J., B. Pearson, and A. Pickering. 2003. “Behind the cube rule:
Implications of and evidence against a fractal electoral geography.”
Environment and Planning A 35: 1405–14.
Martis, K., Z. Kovacs, D. Kovacs, and S. Peter. 1992. “The geography of the 1990
Hungarian parliamentary elections.” Political Geography 11: 283–305.
McCormick, J. 1996. “Taxes and turnouts: The poll tax and the Scottish regional
elections.” Scottish Geographical Magazine 112: 147–57.
Morrill, R., L. Knopp and M. Brown 2007. “Anomalies in red and blue:
Exceptionalism in American electoral geography.” Political Geography 26
525–553.
Murdoch, J. 2006. Post-structuralist Geography: A Guide to Relational Space.
London; Thousand Oaks, CA: Sage.
Nelson, L.K. 2006. “Geographies of state power, protest, and women’s political
identity formation in Michoacan, Mexico.” Annals of the Association of
American Geographers 96: 366–89.
O’Loughlin, J., M. Shin and P. Talbot 1996. “Political geographies and cleavages in
the Russian parliamentary elections.” Post-Soviet Geography and Economics
37: 355–85.
O’Reilly, K. and G.R. Webster 1998. “A sociodemographic and partisan analysis of
voting in three anti-gay rights referenda in Oregon.” Professional Geographer
50: 498–515.
Oseikwame, P. and P.J. Taylor 1984. “A Politics of Failure – the Political-
Geography of Ghanaian Elections, 1954–1979.” Annals of the Association of
American Geographers 74: 574–89.
Owen, G. and B. Grofman. 1997. “Estimating the likelihood of fallacious
ecological inference: linear ecological regression in the presence of context
effects.” Political Geography 16: 675–90.
Painter, J. and A. Jeffrey. 2009. Political Geography (Second Edition). London:
Sage.
Pattie, C., D. Dorling and R. Johnston 1997. “The electoral geography of
recession: Local economic conditions, public perceptions and the economic
vote in the 1992 British general election.” Transactions of the Institute of
British Geographers 22: 147–61.
Pattie, C.J. and R.J. Johnston 1998. “Voter turnout and constituency marginality:
geography and rational choice.” Area 30: 38–48.
Perepechko, A.S., V.A. Kolossov, and C. ZumBrunnen. 2007. “Remeasuring and
rethinking social cleavages in Russia: Continuity and changes in electoral
geography 1917–1995.” Political Geography 26: 179–208.
Rallings, C., R. Johnston, and M. Thrasher. 2004. “Equalising votes but enabling
bias: The electoral impact of the 1977 and 1999 ward boundary reviews in
London.” Urban Studies 41: 1367–93.
Rasmussen, C.E. 2006. “We’re no metrosexuals: Identity, place and sexuality in
the struggle over gay marriage.” Social & Cultural Geography 7: 807–25.
Reynolds, D.R. 1990. “Whither electoral geography? A critique.” In R.J. Johnston,
F.M. Shelley, and P.J. Taylor (eds), Developments in Electoral Geography.
London: Routledge, 22–35.
Sauer, C.O. 1918. “Geography and the Gerrymander.” American Political Science
Review 12: 403–26
Secor, A.J. 2004. “Feminizing electoral geography.” In L.A. Staeheli, E. Kofman,
and L.J. Peake (eds) Mapping Women, Mapping Politics: Feminist Perspectives
on Political Geography. New York: Routledge, 261–72.
Shelley, F.M., R.J. Johnston, and P.J. Taylor 1990. “Developments in electoral
geography.” In R.J. Johnston, F.M. Shelley, and P.J. Taylor (eds), Developments
in Electoral Geography. London: Routledge, 1–11.
Shin, M.E. 2001. “The politicization of place in Italy.” Political Geography 20:
331–52.
Shin, M.E. and J. Agnew 2002. “The geography of party replacement in Italy,
1987–1996.” Political Geography 21: 221–42.
Shin, M.E. and J.A. Agnew. 2008. Berlusconi’s Italy: Mapping Contemporary
Italian Politics. Philadelphia: Temple University Press.
Singh, C.P. 2000. “A century of constituency delimitation in India.” Political
Sui, D.Z. and P.J. Hugill. 2002. “A GIS-based spatial analysis on neighborhood
effects and voter turn-out: A case study in College Station, Texas.” Political
Taylor, P.J. and R.J. Johnston. 1979. Geography of Elections. London:
Harmondsworth.
Thill, J.C. 1998. “A note on the matrix formulation of gerrymanders.” Environment
and Planning B-Planning & Design 25: 495–505.
Varsanyi, M.W. 2005. “The paradox of contemporary immigrant political
mobilization: Organized labor, undocumented migrants, and electoral
participation in Los Angeles.” Antipode 37: 775–95.
Warf, B. 2006a. “Do voting technologies discriminate across the urban hierarchy?
Evidence from two presidential elections.” Urban Geography 27: 664–77.
Warf, B. 2006b. “Voting technologies and residual ballots in the 2000 and 2004
presidential elections.” Political Geography 25: 530–56.
Waterman, S. and E. Zefadia. 1992. “Political-geography in practice II: Israeli
electoral reforms in action.” Political Geography 11: 563–78.
Webster, G.R. 1992. “Geography of a Senate confirmation vote.” Geographical
Review 82: 154–65.
Webster, G.R. 1993. “Congressional redistricting and African-American representation
in the 1990s – An Example from Alabama.” Political Geography 12: 549–64.
Webster, G.R. 1995. “Review of Quiet Revolution in the South – The Impact of
the Voting-Rights Act, 1965–1990 – edited by C. Davidson and B. Grofman.”
Social Science Quarterly 76: 474–75.
Webster, G.R. 1997. “The potential impact of recent Supreme Court decisions on
the use of race and ethnicity in the redistricting process.” Cities 14: 13–19.
Webster, G.R. 2000a. “Playing a game with changing rules: Geography, politics
and redistricting in the 1990s.” Political Geography 19: 141–61.
Webster, G.R. 2000b. “Women, politics, elections, and citizenship.” Journal of
Webster, G.R. 2002. “The US presidential election and the Bush v. Gore Supreme
Court decision.” Political Geography 21: 99–104.
Webster, G.R. 2004. “Representation, geographic districting, and social justice.”
Journal of Geography 103: 111–26.
Webster, G.R. 2007. “Social justice and voting rights in the South.” Southeastern
Webster, G.R., T. Chapman, and J. Leib. 2010. “Sustaining the ‘societal and
scriptural fence’: Cultural, social, and political topographies of same-sex
marriage in Alabama.” Professional Geographer 62: 211–29.
Webster, G.R. and J.I. Leib. 2001. “Whose South is it anyway? Race and the
Confederate battle flag in South Carolina.” Political Geography 20: 271–99.
Webster, G.R. and J.I. Leib. 2002. “Political culture, religion, and the Confederate
battle flag debate in Alabama.” Journal of Cultural Geography 20: 1–26.
Webster, G.R. and N. Quinton 2010. “The electoral geographies of two
segregationist (‘Jim Crow’) referenda in Alabama.” Political Geography 29:
370–80
West, W.J. 2005. “Regional cleavages in Turkish politics: An electoral geography
of the 1999 and 2002 national elections.” Political Geography 24: 499–523.
Part II
Electoral Geography Outside
of the US
Chapter 3
Electoral Systems, Geography, and Political
Behaviour: United Kingdom Examples
Ron Johnston and Charles Pattie
Geography is at the core of most systems used to elect legislators to represent

territorially-defined constituencies. Defining and then operating within such
constituencies poses challenges to both the political parties and the electorate not
only because they structure the choices that voters make but also because their
spatial configuration can strongly influence the electoral outcome – how votes
are translated into seats. These characteristics are not always realized, except in
the case of single-member, plurality constituency systems (often termed first-
past-the-post or FPTP) where the potential for gerrymandering is great – not only
intentionally, as in the USA, but also unintentionally because of the rules that non-
partisan districting bodies have to apply, as Gudgin and Taylor (1979) showed in
a classic contribution to electoral geography and political science. Constituency
definition can also have major impacts on the operation and outcome of contests
using other types of electoral system, however.
The United Kingdom offers a fascinating case study for appreciating these
geographical components of electoral operation, because of the wide variety of
systems deployed: indeed, examples of most of the main types of system can be
found there (Electoral Reform Society, 2007c; Ministry of Justice, 2008). The
basic system, used for elections to the House of Commons (the lower house of
the UK Parliament), is first-past-the-post; this is also used in elections to local
governments in many parts of England and Wales, though in some areas (such
as the English metropolitan boroughs) each ward returns three councillors at
an election, with voters having three votes. For the election of city mayors in
England – including the Mayor of London – a preferential voting method termed
the supplementary vote in single member constituencies is used, and a similar
system, the Alternative Vote (known in North America as instant run-off), has
been proposed for electing the House of Commons – if approved in a national
referendum to be held in May 2011.
In addition to those using single-member constituencies, all of the three other
main types of electoral systems are also used, at least in some parts of the United
Kingdom. List systems, designed to produce proportional representation, are
widely used in many parts of Europe, and one – the closed party list in multi-
member constituencies – was adopted by the UK Parliament, in order to meet
European Union requirements, for elections to the European Parliament from
1989 on, in England, Scotland and Wales, but not Northern Ireland. (Oelbermann
et al., 2010, illustrate the considerable variety of systems used across the 27
member states for elections to that Parliament, consistent with the proportional
representation requirement.) Another multi-member constituency system
that involves preferential voting – the Single Transferrable Vote – that can, in
many circumstances, produce a result broadly consistent with the principle of
proportional representation, is used for elections to the Northern Ireland Assembly,
to the European Parliament from Northern Ireland, and to local governments
in Northern Ireland and Scotland. Finally, versions of the hybrid system often
known as Multi-Member Proportional or Additional Member Proportional, in
which electors have two votes, one for legislators to represent single-member
constituencies (elected by FPTP) and another to elect members of a party list in
multi-member constituencies, are used for elections to the Scottish Parliament,
the National Assembly of Wales, and the London Assembly.
Examples of all of these systems are used here to illustrate the role of geography
in their construction and operation. Having introduced their core features, we focus
on the issues that they raise for both parties and voters, plus the consequences of
how they translate votes into seats.
The Issues
Three main sets of issues are addressed here:
1. The geography of the system – basically how the constituencies are defined
and their boundaries demarcated;
2. Issues that the parties have to address, such as how many candidates to
field in some circumstances, how to spatially structure their campaigns,
and what sort of advice they offer to the electors – which may be much
more than ‘vote for us’;
3. Issues that the electors have to address, such as whether to vote tactically
(Americans call this voting strategically) and what information they need
to make such decisions.
One feature common to all systems is the minimum requirements principle,

which is the smallest number of votes that will guarantee election in a particular
constituency. This number is generally known as the quota, and it varies according
to the magnitude of the constituency (i.e., the number of legislators that it elects,
not the number of voters there or some other measure of its size). It is derived by
the formula Q = [(total number of valid votes cast/(total number of legislators to
be elected + 1)] + 1.
Thus, in a constituency returning one member in which 100,000 votes are cast,
the quota is [(100,000)/(1 + 1)] + 1 = 50,001 which is, of course, a bare majority:
Electoral Systems, Geography, and Political Behaviour 33
any candidate with that number of votes (and there can only be one) must be
elected. If two legislators are to be elected, then the quota is [(100,000)/(2 +1)]
+ 1 = 33,334 and again any candidate with that number of votes is guaranteed
election (since if there were two other candidates, only one could get more than
33,334 – if she got 33,335 votes, for instance, then the third candidate could only
get a maximum of 33,331).
Basically the larger the magnitude of the constituency, the smaller the Q. It can
alternatively be expressed as PCQuota = [100/(n + 1)] where n is the constituency
magnitude (number of legislators to be elected). Thus in a two-member
constituency, 33.3 percent of the votes will guarantee election for a candidate,
in a four-member constituency 20 percent are sufficient, and in a nine-member
constituency 10 percent.
Single-Member Constituency Systems
Single-member constituency systems are a minority in the world, being

especially popular only in the English-speaking countries, most of them
formerly part of the British Empire. Among those countries there have been
moves away from that arrangement: in 1994, for example, South Africa’s first
fully democratic general election – and all subsequent contests – have been held
using a proportional representation (PR) list system and two years later New
Zealand’s general election for the first time used a hybrid system, introduced
after two popular referendums rejected the British system that had previously
been used there (Renwick, 2010).
First-Past-The-Post (FPTP) – or Plurality – Elections
FPTP elections have been the sole means of electing members of the House of
Commons only since 1951. In the nineteenth century multi-member constituencies
were the norm, with electors having the same number of votes as there were seats
to be filled, so that if there were two seats for a constituency the two candidates
with the largest number of votes were elected. The Reform Acts of that century
(1832, 1867, 1885) and of 1918 slowly replaced those seats with single-member
constituencies, but it was not until 1949 that the last multi-member constituencies
were removed (Rossiter et al., 1999).
FPTP elections are often promoted as straightforward, especially for the
elector; the act of voting is ostensibly simple – voters choose their preferred
candidate/party – and identification of the winner is similarly clear – the candidate
with most votes. But the system is open to many distortions and biases, and can –
in most circumstances – present the voters with difficult choices.
The results of UK general elections since 1950 have been characterized by
two main features. The first is disproportionality: parties’ percentages of the
votes cast are not commensurate with their percentages of the legislative seats
won. Some get a higher percentage of seats than votes; others get the reverse. In
general, the largest party in terms of votes cast gets an even larger share of the
seats and the smallest parties get lower shares of the seats than of the votes – there
is in effect a ‘winner’s bonus’, which usually means that the winning party gets
a majority of seats in the House of Commons (even though no party has won a
majority of votes at any election during that entire period). But this is not always
the case. In 1951, for example, Labour won 48.8 percent of the votes cast (its
largest share ever) and the Conservative-Unionists 48.0 percent, but the latter got
321 seats (a Parliamentary majority) to Labour’s 295. At the first election held
in 1974, the situation was reversed: the Conservatives won 37.9 percent of the
votes compared to Labour’s 37.2, but Labour got four more seats than its main
opponent (but not a Parliamentary majority). And then in 2010, the Conservatives
were again the largest party, with 37.4 percent of the votes, but they obtained only
47.2 percent of the seats, resulting for the first time since 1945 in the creation of
a formal coalition government.
The main reason for this disproportionality is geography, specifically the
geography of where each party wins its votes within the matrix of constituencies
– as laid out in one of the classic works of electoral geography (Gudgin and Taylor,
1979). Take the 1951 election as the example. If Labour had won 48.8 percent
of the votes in all of the 625 constituencies and the Conservatives 48.0 percent,
then Labour would have won all of the seats. It didn’t because too many of its
votes were in the wrong places. As Gudgin and Taylor (1979) demonstrated, the
key parameter in the translation of votes into seats is the frequency distribution
of a party’s share of the votes across all constituencies. If it wins a very large
share of the votes in a small minority of the seats (65 percent in 100 of the 625,
say) but a smaller than average share in a much larger set (35 percent in 350,
say) then it will win all of the former but lose all of the latter. Its opponent, on
the other hand, may have its votes much more efficiently distributed, winning
perhaps 52 percent of the votes in 400 of the seats. The latter will then get a
much better return for its share of the votes cast than the former – hence the
disproportionality. And of course a small party, winning three percent of the
votes, say, will get no seats at all – unless most of them are concentrated in a
very small number of constituencies where it obtains a plurality (which was the
case for the UK Liberal party in 1951, that gained 2.5 percent of the votes but
only six MPs out of 623).
To some observers, the disproportionality inherent in the FPTP system is among
its strengths – despite the occasional anomalies when the winning party in terms
of votes is not also the largest in terms of seats. (Such anomalies also occurred in
New Zealand at successive elections in 1978 and 1981.) The largest party is able
to form a government with a workable majority in most cases but because the
system tends also to magnify changes in support – a small change in percentage
of the votes won is usually associated with a much larger change in share of the
seats, again a function of the frequency distribution of the former (especially if it
is leptokurtic: Gudgin and Taylor, 1979) – a governing party that loses popular
support is thus likely to be replaced. Small parties are disadvantaged, however, but
this is accepted as the price to pay for clarity in the election results. In effect, as
Duverger (1954) argued in his classic study of electoral systems, FPTP is not only
most appropriate for a country where two parties predominate but also ensures in
most cases that this is the case.
The United Kingdom, however, no longer has a two-party system – if it ever
did. It has always had some smaller parties that were strong in some parts of
the country only – Nationalists and Unionists in Northern Ireland; the Scottish
National Party (SNP) in Scotland; and Plaid Cymru (PC) in Wales – but only in
Northern Ireland have these come to dominate and create a different party system.
(Separate party systems in the different Provinces are common in Canada, of
course.) Within Britain, the SNP and PC increased their support substantially after
1970, alongside resurgence in voter support for the Liberal party (which after a
merger with the Social Democrats in 1988 became the current Liberal Democrat
Party). As a result of their combined success, a multi-party system began to
emerge: the Conservative and Labour combined share of the votes cast fell from
over 90 percent in the 1950s to only 67 percent in 2010, but they still obtained
the great majority of the seats – 86 percent in 2010. The Liberals were especially
disadvantaged because their votes were too widely dispersed – at elections since
1992 they have averaged around 20 percent of the votes but less than 8 percent of
the seats.
The growth of a multi-party system (with a number of other smaller parties,
some of which – like the United Kingdom Independence Party (UKIP), the
British National Party (BNP) and the Green Party – contest many constituencies
now) means that disproportionality is even greater than in the earlier post-World
War II decades. Alongside it a further feature has developed – bias. The system
does not treat the two largest parties equally in the translation of votes into seats.
At recent elections, Labour has benefited very substantially from this. If we take
a ‘fair disproportional’ election system as one that translates votes into seats in
the same ratio whichever party is the victor, it would be unbiased. But if one
party is advantaged over the other the electoral system is biased. As a baseline,
we can argue that if two parties got the same share of the votes cast, they should
get the same number of seats. At the last four general elections this was not so.
In 1997, if the Conservatives and Labour had been equal in vote shares, Labour
would have won 82 more seats than its opponent; in 2001, it would have won
142 more; in 2005, 112; and in 2010, 54 (Johnston et al., 2006; Johnston and
Pattie, 2011a).
The reasons for this bias are, in effect though not in means, the same as the
classic causes of electoral abuse in the United States – malapportionment and
gerrymandering (Johnston et al., 2001). Constituency boundaries in the UK are
reviewed approximately every decade by non-partisan commissions operating
rules laid down by Parliament. The boundaries are not put in place to advantage
one party over another (although because the parties can challenge, though not
determine, the boundaries during the public consultation process there are some
attempts at the equivalent of gerrymandering, which are of minor importance only:
Rossiter et al., 1999). But, again as demonstrated by Gudgin and Taylor (1979),
because those constituencies are superimposed upon maps of support for the parties,
and all constituencies are – for a variety of reasons (Borisyuk et al., 2010) – not the
same size, the equivalent of gerrymandering and malapportionment emerges; and
since 1992 this has been substantially to Labour’s advantage (Johnston, 2002).1
This superimposition of the constituency map on that of voter preferences
creates a geography that has become of increasing relevance to the country’s
political parties and voters. Of the two largest parties, the Conservatives have
their greatest support among the middle classes, concentrated in the suburban
and rural areas, whereas Labour’s heartlands are in the inner cities and the
industrial areas. For each, therefore, there are areas of the country where their
support is so strong that they are almost certain to win – their ‘safe seats’ – and
others where they are sure to lose – their ‘hopeless seats’. The contest between
them is thus compressed into those constituencies where neither’s support base
predominates – the ‘marginal seats’, comprising constituencies that incorporate
a range of socio-economic environments. These are where the election is won
and lost, and are the parts of the country where the parties and their candidates
campaign most intensively. Voters in the safe and hopeless seats get relatively
little attention from the parties in the months and weeks before an election
whereas those in the marginals are the subject of a great deal of interest as the
local party machines (increasingly carefully managed and resourced from central
party headquarters) invest much time and energy into identifying their potential
supporters and encouraging them to vote and, potentially, make a difference to
the result. And it works; the more intensively a party campaigns in a constituency
(especially a party other than one currently in government) the better the result
for it there (Johnston and Pattie, 2008; Pattie and Johnston, 2009; Johnston and
Pattie, 2011b).
Voters respond to this geography of campaigning, of course. Some, who feel
that their vote is not needed, abstain. In particular, electors in ‘hopeless seats’
are markedly less likely to vote (even to use the increasingly-promoted postal
voting facility introduced to increase turnout – which has fallen nationally by
some 20 percentage points in recent decades) than those in the marginals. Turnout
is generally lower in such constituencies, especially those that are Labour-held;
potential Labour voters decide that their vote is not needed – especially if they
haven’t been canvassed by the party – whereas supporters of other parties decide
their votes too will have no impact and so abstain. (The former is less likely to
1 In 2010 the coalition government that took office after the election in May that year
introduced a Parliamentary Bill that would both reduce the number of MPs (from 650 to
600) and require the new constituencies to be drawn so that all their electorates (with two
exceptions in Scotland) were within 5 percent of the national quota (some 78,000). (On
predecessors to that proposal, see Johnston et al., 2009.)
be the case in Conservative areas; that party’s voters are more likely to turn out
whatever the local situation.)
Abstaining is only one response to the geography of the election and its
campaigns. Rather than abstain in the knowledge that their vote may have no
impact, some electors decide to vote for another party. Indeed, they do that in
considerable numbers. The 2005 British Election Study survey, for example,
found that 15 percent of all respondents who voted did not cast their ballot for
their most preferred party. Of those who preferred a party other than either the
Conservatives or Labour, 45 percent voted for a party other than their first choice;
and of those whose first preference was neither one of those two parties nor
the Liberal Democrats, fully 79 percent of them voted for other than their most
preferred candidate. To some extent, this was because their preferred party did
not present a candidate in the constituency, but that was not the major cause. The
Conservatives, Labour and the Liberal Democrats fielded a candidate in every
British constituency, for example, as did the SNP and PC in Scotland and Wales
respectively. And the ‘minor parties’ also offered candidates in most places: of the
632 constituencies in Britain, 557 had a UKIP candidate, 337 one from the BNP,
and 332 from the Greens.
The main reason for voting for a candidate other than one’s first choice at
a British election is what is termed tactical voting there (strategic voting in
North America: for an introduction to the concept, see Cox, 1997). Many vote
for their second preference because that person has a better chance of victory
locally than their first choice. Thus, for example, in 1997 the two left-of-centre
parties – Labour and the Liberal Democrats – wanted to end the Conservative
sequence of governments that started in 1979 and so in constituencies where
Labour candidates has the best chance of unseating a Conservative incumbent
many Liberal Democrats voted Labour to achieve that, whereas in constituencies
where the Liberal Democrats were best placed for victory the reverse was the
case. Since 1997, such tactical voting has focused on preventing a Conservative
return to power. In many cases, the parties themselves have promoted – implicitly
if not explicitly – such tactical voting: many Liberal Democrat candidates have
published leaflets with electoral data (some of it fairly spurious) claiming that
only they could beat the Conservatives in their constituency (a tactic supported
by some of the media, and by senior members of both the Liberal Democrat
and the Labour parties – with the latter promoting what they termed ‘intelligent
voting’). In 2010, when all three parties believed they could win at some stage
of the election campaign, as well as the promotion of tactical voting there was
also ‘anti-tactical-voting’ campaigning: Conservatives in some areas claimed
that a Liberal Democrat vote would assist the return of a Labour government, for
example, whereas in other areas Labour campaigned that a Liberal Democrat vote
could ensure a Conservative administration. Tactical voting is now an established
feature of British elections, with some 15 percent engaged in it in 2010. As a
result, instead of England having a three-party system, it has in effect three two-
party systems: in about half of the constituencies, the contest is between Labour
and the Conservatives, with the Liberal Democrats a poor third; and in about one-
quarter each there are Conservative-Liberal Democrat contests (mainly in the non-
metropolitan south of England) and, in the urban north, Labour-Liberal Democrat
contests (Johnston and Pattie, 2011a)
First-past-the-post is thus a system that is open to manipulation through the
cartography of its constituency map (Johnston, 2002) and through tactical voting,
and election results are likely to display both disproportionality and bias. It meets
what, to many, is a basic criterion for a successful electoral system – the elected
legislators have a clear and individual link to a geographical segment of the relevant
territory – but it fails to meet many others, such as PR. The same conclusions can
be drawn about other systems that are based on single-member constituencies.
Preferential Voting in Single-member Constituencies
Whereas in FPTP electoral systems voters pick one candidate/party only to receive
their support, in preferential systems they place two or more of the candidates in
order of preference. The goal is to ensure that the elected candidate has majority
support; FPTP in multi-party systems can result in many of the legislators being
elected with a minority of the votes only (as was the case with two-thirds of the MPs
elected at the 2010 general election). These preferential systems do not eliminate
the disproportionality and bias characteristic of FPTP – indeed, they probably
accentuate them, because of the issues involved in constituency delimitation – but
they also introduce new issues for both parties and electors.
The Supplementary Vote
The supplementary vote (SV) was introduced in the late 1990s for the election
of the Mayor of London and also mayors in a small number of other cities and
towns where such a form of government was supported by the local population
in a referendum. (It was devised by a Labour MP when his party was considering
electoral reform after its third successive defeat by the Conservatives in 1987,
was commended by the Plant Commission – Plant, 1993 – and subsequently
recommended for the London Mayoral election: Dunleavy and Margetts, 1998;
van der Kolk, Rallings and Thrasher, 2004; Rallings, Thrasher and Cowling,
2002.) Voters are presented with a ballot paper in which they indicate both their
first and their second preferences. The first preferences are counted, and if one
candidate has secured more than 50 percent (i.e., has achieved the quota) he/
she is declared elected. If there is no outright winner, there is an instant run-off
between the top two candidates. All of the others are eliminated and their second
preferences allocated: if they are for one of the first two candidates, they are added
to the relevant total; if they are for one of the eliminated candidates, they are
discarded, and the second preferences of electors who voted for one of the run-off
candidates are ignored. Of the candidates in the run-off, the one with most votes
Table 3.1 Results of the First Three Elections for the Mayor of London,
Indicating the Number (and Percentage) of the First Preference
Votes and the Final Number of Votes Going to the Candidates of
the Two Leading Parties
First Preferences Final Total Votes Cast

2000
Labour 667,877 776,427 1,714,072
(39.0)
Conservative 464,434 564,137
(27.1)
2004
Labour 685,541 828,380 1,863,671
(35.7)
Conservative 542,423 667,178
(28.2)
2008
Labour 893,877 1,028,966 2,415,958
(37.0)
Conservative 1,043,761 1,168,738
(43.2)
(first plus second preferences) is elected: this may be with a clear majority – if
most of the supporters of the eliminated candidates gave their second preferences
to one of the candidates in the run-off.
Table 3.1 shows the results for the three elections of the Mayor of London
in 2000, 2004 and 2008. On each occasion the leading candidate failed to gain a
majority of the first preference votes, but sustained his position after the allocation
of second preferences. In 2000, the winning candidate (Ken Livingstone) had 69
percent of all of the votes after preferences had been allocated, but in each of the
subsequent elections the victor failed to get an overall majority (Livingstone got
43 percent of the 1.9 million cast in 2004 and his opponent – Boris Johnson – 46
percent in 2008, indicating that a substantial number of second preferences in each
case were wasted).
Indeed, the wasting of votes can be a major issue. In the first election for Mayor
of Stoke-on-Trent, the leading two candidates – who entered the run-off stage –
each got about one-fifth of the first preference votes (there were 12 candidates). The
second-placed candidate was the eventual victor, but although 24,866 votes were
cast for the ten eliminated candidates, only 5,964 were allocated to one of the two
run-off candidates – so that the winner was elected despite 75 percent of the voters
giving him neither their first nor their second preferences (van der Kolk, Rallings
and Thrasher, 2004). Such extensive wasting can occur for one of three reasons:
people only indicate their first preference; people mistakenly give their first and
second preferences to the same party; or people give their second preferences
to an eliminated candidate. (The last of these is especially likely to occur when
voters are uncertain which two candidates will be in the run-off.) Data for the first
two London Mayor elections allow a decomposition of the wasted votes. Thus in
2000, of the 581,761 first preferences given to candidates other than the two front-
runners, 16 percent had no stated second preference, a further 7.7 percent had the
second preference allocated to the same candidate as the first preference and so
were eliminated, and 40.5 percent went to candidates outside the first two; thus
only 35.8 percent of the potential second preferences were allocated to one of the
two leading candidates. Four years later, when the same two candidates occupied
the run-off positions, 42.1 percent only of the potential second preferences was
allocated to one of them (van der Kolk et al., 2006). Thus whereas in each case
the winner (Ken Livingstone) obtained a majority of the ‘effective votes’, after the
allocation of second preferences he was elected by only a minority of all of those
who voted (45.2 and 44.4 percent in 2000 and 2004 respectively).
This system clearly involves electors having to make more than the single
decision involved under FPTP, selecting their second as well as their first preference
candidate. In doing so, they have to be aware that their second preference will not
count if it is given to a candidate other than one of the two who are in the run-
off. How will they know who those two will be?2 In some high profile contests
opinion poll data might provide the needed information – but if it is a fairly close
three-way contest that might not help. Otherwise, electors will have to rely on
either their own general appreciation of the local political situation or what they
are told by either the parties (in electioneering materials) or the local media –
which may not be reliable. Parties and/or candidates may campaign for second
preferences if they are confident of making the run-off, whereas others who are
not may suggest how their first preference voters should use their second choice –
which would also be the case if the supplementary vote were used to replace FPTP
in Parliamentary elections.3 In general, however, many electors may ‘waste’ their
second preference in the absence of any definitive information/advice – which of
course may be erroneous if those offering it wrongly identify the two candidates
who will make the run-off.
The SV system has not been used in the United Kingdom to elect a representative
body rather than a single executive, such as a mayor. If it were, it would clearly
carry many of the same disadvantages as FPTP: indeed, it could well exacerbate
2 In ‘true’ runoff elections, where a second round of voting is held at a later date, with
candidates getting less than a specified percentage of the first round votes being eliminated
– as in Presidential and Legislative elections in France – this problem is absent.
3 Of course, to suggest how supporters might use their second preferences could be
seen as an admission of defeat, unless it was deployed in a positive way to show how that
could influence the winning candidate’s policies.
them. It is designed to ensure that the victor has a majority mandate – by the
allocation of second preferences if no candidate gets above the quota (50 percent
of the votes cast plus one) – but by restricting the allocation of preferences only to
the top two candidates on the first preference votes it prevents a relatively popular
‘third party’ candidate from overtaking them. Disproportionality – measured using
each party’s percentage of the first preference votes and of the seats – is thus
likely to be as large if not larger than with FPTP, and the outcome could also be
biased for the same reasons that FPTP results are. Further, because the electors
are to some extent operating under a ‘veil of ignorance’, in that their decisions on
who to support in the run-off must be made without knowing who the candidates
in that run-off will be (though they may be able to estimate that – or be given
an estimate by the parties and their candidates), it may well be that many of the
second preferences – and hence the first preferences with which they are linked –
are ineffective (or ‘wasted’).
The Alternative Vote
In this system (AV), rather than indicate first and second preferences only among
the candidates, electors are asked to rank order them all. If a candidate gets a
majority of the first preference votes, he/she is elected. If no candidate achieves
the quota, however, the one with fewest first preferences is eliminated, and her/his
second preferences allocated among the others; if one then exceeds the quota, the
election is over. If no candidate still reaches the majority, the remaining one with
fewest first preferences is eliminated and second preferences allocated – or third,
if the second preference is for an eliminated candidate. The procedure continues
until either one candidate reaches the quota or only two remain, in which case the
one with most votes is elected. (This may not be a majority because some electors
do not rank order all of the candidates.)
The alternative vote system was pioneered in Australia in 1919 as a way
of combining the votes for several right-of-centre parties that otherwise might
lose an FPTP election to a single left-of-centre party’s candidate who failed to
obtain a majority of the votes cast (McLean, 1996); it has been used since then
for elections to the country’s House of Representatives, as well as to several of
the state legislatures (Farrell and McAllister, 2006). It places considerably more
demands on the voter than even SV, especially where there is a large number of
candidates (in some Australian elections, electors have been asked to rank order
more than 20 candidates in a single constituency). Although voters do not need to
be prescient as to which the leading two candidates on the first preferences will
be, some sophisticated decision-making may be needed to ensure that their votes
are not wasted, especially if their first preference is for a party very unlikely to
be elected in the constituency. Tactical voting can be very effective, especially
at preventing the election of a candidate with substantial, but not majority, first
preference support but, unlike FPTP, the system does not discourage people
from giving their first preference to a candidate who will almost certainly lose.
If the second and lower preferences are allocated, however, the order in which
candidates are placed can be crucial – which is why Australian political parties
distribute ‘how to vote’ cards to voters, suggesting a rank-ordering that is in their
own candidates’ best interests by reducing the chances of their main contenders.
The production of these ‘how to vote’ cards will almost certainly be preceded
by political bargaining among the parties. If parties A and B both want party C’s
voters’ second preferences to ensure victory in as many constituencies as possible,
they may well have to concede legislative concessions to party C (perhaps without
divulging that they have), and failure to fulfil their promises could make it more
difficult to bargain for such recommendations from party C to its voters at future
contest. Furthermore, a sophisticated electorate could soon realize that a party that
has little or no chance of victory should still get its supporters’ first preference
votes. Indeed, the more first preference votes that such parties obtain, the greater
the probability that no party will win an outright majority on the first preferences:
not only will such minor party supporters be more likely to influence the electoral
outcome in a considerable number of constituencies but they are also more likely
to have an impact on policy development than if they gave their first preferences
to the ultimate winner (Jaensch, 1983, calls this ‘second-preference government’).
In addition, the more first preference votes that go to ‘hopeless’ parties the greater
the probability that the margin between the two (or three) leading parties will
be small, putting greater power into the choice of second preferences by the
eliminated candidates’ supporters
AV, like SV, does not produce election outcomes consistent with proportional
representation. Indeed, because it favours the largest parties it is likely to generate
outcomes that are even more disproportional and biased than those associated with
FPTP (Johnston and Forrest, 2009a, 2009b). Nevertheless, AV was proposed as
a replacement for FPTP by the UK Labour party in 2010, in an attempt to win
support from the Liberal Democrat party (which favours PR using the STV system
discussed below) in a possible coalition agreement after the general election then.
Their argument was that because AV ensured that all MPs were elected with a
majority mandate from their constituency this would enhance their legitimacy –
although the realpolitik was that the main losers, should it be adopted, would be the
Conservatives if, as was probably the case until then, Liberal Democrat candidates
were more likely to get second preferences from Labour candidates in third place
than were the Conservatives, and also that Labour candidates would pick up more
second preference votes from third-placed Liberal Democrat candidates than the
Conservatives.
Some Liberal Democrat politicians saw AV as a valuable first step on the route
to PR, recognizing its advantage in eliminating some of the pressure for tactical
voting; more voters would be able to indicate their ‘sincere (first) preference’
for the Liberal Democrats under AV without wasting their vote if their other
preferences were to be taken into account. But the potential for tactical voting
is present in an AV election. If, of three remaining candidates in a constituency,
B is likely to lose to A after the final preferences are counted but to beat C – on
the assumption that B will be in the final run-off – then B’s supporters should try
to ensure that C is not eliminated before the final distribution of preferences; if
enough of B’s supporters vote for C so that it enters the run-off but that B is not
eliminated – leaving A and C in the run-off – then B’s chances of ultimate victory
are enhanced; you want to run-off against the weaker of your two main opponents
if you are sure of being at that stage yourself. Such a tactic of course depends
upon detailed information being available to the voters – either from independent
sources that they can employ to make their own tactical decisions or, more likely,
from the parties who stand to gain from such behaviour. Alternatively, a party that
knows it cannot win in a constituency may prefer to come third – so that its second
preferences are then crucial in ensuring that a less-preferred candidate wins – than
coming second and then being beaten.4
The outcome of the 2010 general election was a coalition of the Conservatives
– who were strongly in favour of retaining FPTP – and the Liberal Democrats. As
part of the agreement, the Conservatives agreed to support a Parliamentary Bill
to hold a referendum on a change from FPTP to AV in 2011 – but reserved their
right to campaign against the change when it was put to the electorate. Labour
was prepared to legislate for the change without a referendum – having previously
promised one if re-elected – if the Liberal Democrats entered a coalition agreement
with them rather than the Conservatives, but the negotiations broke down on a
number of issues and a Conservative-Liberal Democrat coalition was formed (on
which see Mandelson, 2010).
Multi-Member Proportional Representation List Systems
There are two basic types of list system. In the first – known as closed-list systems
– the political parties nominate a list of candidates for each constituency in rank
order. Electors vote for their preferred list, and parties are then allocated a number
of legislators consistent with their share of the votes. Thus in a nine-member
constituency a party with 30 percent of the votes would be guaranteed three
legislators (the PCQuota is 10), and the first three candidates on its list would be
declared elected. The other type is known as the open-list system, in which instead
of just plumping for a party the electors can, if they wish, indicate their first (and
sometimes lower) preferences among their chosen party’s candidates – they can
only vote for one party’s candidates. In these cases, if a party gets 30 percent of
the votes in a nine-member constituency, the three most popular of its candidates
are elected. (If an elector does not indicate her/his preferred candidate, then her/
his vote is given to the top candidate on the party’s list.)
4 If C’s supporters know that they cannot win in a constituency against A, then they
might prefer that their second preferences ensure that B defeats A than take the chance that
if B came third, its supporters would give insufficient second preferences to C to prevent
A’s victory.
Table 3.2 Allocation of Seats across the Nine English Regions for the
European Parliament Elections in 2009
Northeast 3
Northwest 8
Yorkshire and the Humber 6
East Midlands 5
West Midlands 6
Eastern 7
London 8
Southeast 10
Southwest 6
As open list systems are not currently used in the UK, we will not address any
particular issues that they raise. Closed list systems raise three. Basically they
are fairly straightforward to operate. Parties present a full field of candidates –
certainly at least one more than they expect to get elected and no problems are
raised if they nominate more – and they campaign for the maximum number of
votes.
There are two apportionment issues, however. The first relates to the definition
of constituencies. Two options are available to the system designers: a bespoke
set of constituencies for the task, defined by a Boundary Commission or similar
body; or using an existing set of administrative areas. The latter option was taken
in the UK for elections to the European Parliament. Each of the four constituent
countries was allocated a number of seats from the total – in the latest allocation,
Northern Ireland was allocated three, Wales four, Scotland six and England the
remaining 59. Within England, the total was then subdivided across the country’s
nine administrative regions, as shown in Table 3.2.
This task may seem straightforward, but there is a variety of ways in which the
number of seats can be allocated – ways that have been debated for more than a
century and have informed the decennial allocations of House of Representatives’
seats in the United States where some important paradoxical aspects the procedures
explored were generated (Balinski and Young, 2001). In the UK case, the goal was
to allocate seats to regions consistent with the number of registered electors there,
and the most proportional formula – the Sainte-Laguë (or Webster) was used by
the UK Electoral Commission: use of another formula could have changed the
allocation slightly, with the Southeast getting one more seat.5
This allocation immediately identifies an issue for the parties and electors. The
different number of seats per region means that the quota varies spatially. In the
extreme cases within England, in the Northeast region a party needs at least 25
5 An accessible, non-technical introduction to these issues is Szpiro (2010).

percent of the votes to be guaranteed one seat, whereas in the Southeast it needs
only 9 percent. (Many countries have a threshold below which parties will not be
allocated seats. This is usually no more than 5 percent, although in Turkey it is
10 percent; at the 2007 elections there only three parties crossed that threshold
and the largest – with 46.6 percent of the votes – won 341 of the 550 seats in the
legislature. On various systems, see Carey and Hix, 2011.) Smaller parties are
better able to win representation in some parts of the country than others, therefore.
In those where their chances are low they may decide not to present a list, or at
least not to campaign hard there if they do so. Supporters of such parties, if they
are aware (or made aware, presumably by other parties) of the issue may decide
not to ‘waste’ their vote by casting it for their preferred party and instead vote for
their second preference, which has a much greater chance of success. Tactical
voting is sensible in some proportional representation situations, therefore.
The second allocation issue relates to the distribution of seats within each
region once the votes are counted. A party will rarely get precisely the quota and
thus an integer allocation of seats and so some formula has to be deployed to make
the most proportional allocation. As with the allocation of seats to region, Sainte-
Laguë/Webster achieves that, but in many cases, including the UK’s elections
to the European Parliament, the less proportional D’Hondt/Jefferson formula is
used. This can create significant differences (McLean and Johnston, 2009). With
D’Hondt, the total number of votes each party receives is divided by, successively,
1, 2, 3, 4 …, whereas with Sainte-Laguë the sequence is 1, 3, 5 ,7 ....6 Seats are then
allocated to the parties achieving the highest n ratios (where n is the constituency
magnitude – the number of members to be elected from that constituency), as
illustrated in Table 3.3 for the 2009 European elections for the Southwest England
constituency (only the six largest parties are shown).
The seats are allocated sequentially according to the size of the ratios. Thus
under the D’Hondt formula the first seat goes to the Conservatives, the second
to UKIP, the third to the Liberal Democrats and so on. Only three parties win
seats, whereas under Sainte-Laguë five do, which is a substantial difference for
a constituency returning only six legislators. Indeed, across the whole of Great
Britain (excluding Northern Ireland, which elects its three MEPs by a different
system), the difference between the two systems in the overall outcome would
have been quite substantial, as shown by Table 3.4. In effect, the D’Hondt
formula favours the largest parties (the Conservatives and UKIP won 27.7 and
16.5 percent of the votes respectively, but 36.2 and 18.8 percent of the seats)
whereas the smaller parties suffer (the Greens won 8.6 percent of the votes but
2.9 percent of the seats; the BNP got 6.2 and 2.9). The Sainte-Laguë formula
substantially reduces that disproportionality. (For more details, see McLean and
Johnston, 2009.)
6 In some Scandinavian countries 1.4 is used as the first Sainte-Laguë denominator

rather than 1, which slightly disadvantages smaller parties.
Table 3.3 Allocation of Seats in the European Parliament Election of 2007

for the Southwest Region of England According to the D’Hondt
and Sainte-Laguë Divisors
Divisor
D’Hondt 1 2 3 4
Conservative 468,742 253,741 156,247 117,186
Labour 118,716 59,538 39,752 29,679
Liberal Democrat 266,253 133,127 88,751 66,563
Green 144,179 72,090 48,060 28,544
UK Independence 341,845 170,923 113,948 78,711
British National 60,899 30,445 20,296 15,225
Sainte-Laguë 1 2 3 4
Conservative 468,742 156,247 93,748 66,963
Labour 118,716 39,752 23,743 16,959
Liberal Democrat 266,253 88,751 53,521 38,036
Green 144,179 48,060 28,836 20,597
UK Independence 341,845 113.948 68,369 48,835
British National 60,899 20,296 12,178 8,700
Note: Allocated seats are shown in bold; the figures shown in the first column are the total
number of votes won by each party.
Table 3.4 The Allocation of Seats in the European Parliament Election of

2007 from Britain According to the D’Hondt and Sainte-Laguë
Divisors
D’Hondt Sainte-Laguë
Conservative 25 21
Labour 13 13
Liberal Democrat 11 11
Green 2 7
UK Independence 13 11
British National 2 3
Scottish National 2 3
Plaid Cymru 1 2
Preferential Voting in Multi-Member Constituencies: Single Transferable Vote
One system promoted by both the Liberal Democrat party and other bodies – such
as the long-established Electoral Reform Society – is the Single Transferrable Vote
(STV), used for elections to the lower house of the Irish Parliament (the Dáil
Éireann) since 1919 and to the Australian Senate since 1949. It is an extension
of the AV system to multi-member constituencies. Electors rank order all of the

candidates and a quota is determined (as with the above formula). Any candidate
with more first preference votes than the quota is declared elected, and her/his
surplus second preferences (i.e., the first preference votes additional to those
needed to meet the quota) are distributed to the other candidates. When no candidate
exceeds the quota, the one with fewest first preferences is eliminated, as in AV, and
second (or lower) preferences distributed. The process of successive elimination is
continued until sufficient candidates exceed the quota and the constituency’s full
complement of legislators is determined.
One feature of STV that commends it to many proselyts is that it allows voters
choice of candidates within parties as well as between them (assuming that a party
fields more than one!) This to some extent removes the selection of legislators
from the party machines: in single-member systems and also closed list PR
systems, the choice of candidates by the parties ensures that the electorate has
little influence over who can represent them (unless the parties run open primary
contests for selection of candidates prior to the election itself). Voters can thus, for
example, punish a party’s incumbent legislator and replace her/him by a preferred
alternative representing the same party.
In some circumstances, the outcome of an STV contest approaches PR,
depending on two aspects of the geography of the constituencies deployed. First, as
in list systems, the greater the constituency magnitude (n – the number of members
to be elected) the smaller the quota and the greater the electoral prospects of small
parties. Secondly, whether the constituency returns an odd or even number of
members can be very influential, especially in small-magnitude constituencies.
For example, in a four-member constituency the quota is 20 percent, whereas with
three members it is 25 percent. Where two parties predominate in an area, if one
of them has less than 46 percent of the votes there, it would prefer it to have a
four-member constituency, as it is likely to win two of the four seats, whereas
if it has more than 46 percent it would prefer three members, of which it would
gain two (Gallagher, 1974). Of course, it is not easy in most circumstances to
gerrymander constituency magnitudes so as to have even-numbers where a party
is relatively weak and odd-numbers where it is strong, but the potential is there and
the outcome may eventuate without it being intended (a nonpartisan gerrymander)
– as has been the case in Dáil elections (Paddison, 1976; Parker, 1986).
A further issue that parties have to address in STV systems is the number of
candidates to field in each constituency as well how to campaign for their support
locally. A party may be under-represented if it fields too many candidates: for
example, if it expects to get 36 percent of the first preferences in a five-member
constituency (quota, 16.7 percent), it would anticipate winning two seats. However,
if it fielded four candidates several of them might be eliminated early because they
got relatively few first preferences, and it could finish with only one being elected.
In general parties should field only one candidate more than the number of seats
they expect to win.
Even with a sensible selection of candidate numbers, however, a party can still
fail to win a number of seats commensurate to its total number of first preferences.
It is for this reason that parties seek to segment their first preference votes, by
encouraging different groups of their supporters to give their first preferences
to different candidates from among the slate. One common measure of such
segmentation – notably in elections in Ireland, where STV has been in use for
nearly a century – is geographical; voters for party A in different parts of a multi-
member constituency are encouraged to give their first preferences to a particular
candidate. Indeed, candidate selection may also be geographically-based, with A
selecting candidates with homes in different parts of the constituency in order to
promote a ‘friends-and-neighbours’ effect, with each candidate developing her/his
own bailiwick within the territory (Parker, 1982, 1983). This practice is clearly
illustrated by a leaflet used by the Democratic Unionist Party (DUP) in one six-
member constituency for the 2007 Northern Ireland Assembly elections. The DUP
fielded three candidates: it asked its supporters in one group of 10 polling booths
to rank order them a, b and c (in the first three places, of course); in another 7 to
use the order c, b, a; and in a third group of nine to put them in the order b, a, c.
Such vote management strategies may involve asking supporters to transfer their
first preference from a much-admired local candidate who would otherwise get a
large personal vote to the party’s other candidates. Sinn Féin did this successfully
in the West Belfast constituency at that election. Gerry Adams, the party leader,
got 6,029 of the party’s 23,631 first preference vote total there, and Sinn Féin’s
other five candidates each got between 4,250 and 4,750 first preferences. Overall,
Sinn Féin got 70 percent of the first preference votes (the PCQuota in a six-
member constituency is 14.3) but a careful strategy ensured that all five candidates
were elected, when only four were guaranteed. Similar strategies were successful
elsewhere, with the DUP winning four of the six seats in another constituency
with only 50 percent of the first preferences – in part because too many of a Sinn
Féin candidate’s second preferences were non-transferable (i.e., his supporters did
not give a full list of preferences), preventing a candidate of the other nationalist
party winning the sixth seat when preferences were allocated (Electoral Reform
Society, 2007a).
That STV produces outcomes that are often consistent with PR also means that
it is unlikely in most cases that a single party will have a majority in the relevant
legislature. This is illustrated by the case of Scottish local government, where the
method of election was changed from FPTP between 2003 and 2007. In the last
FPTP elections, a single party obtained a majority in 20 of the 32 local authorities,
but in the first STV election this outcome occurred in just five (the majority ‘party’
in three was others/independents; Labour won a majority in Glasgow, with 43.3
percent of the first preference votes, and in North Lanarkshire, with 49.3 percent:
Electoral Reform Society, 2007b). Overall, in 2007 STV ensured that 74 percent
of voters saw their first choice candidate elected, compared to 52 percent who
voted for the winning candidate in 2003 under FPTP.
The Hybrids: MMP/AMP and their Variants
A hybrid electoral system, combining first-past-the-post in single member

constituencies with party list PR in multi-member constituencies, was introduced
in (then West) Germany in 1949 to produce outcomes that included both
representation from small, identifiable territories with a top-up system that moves
the overall result towards PR. It was adopted in New Zealand in 1993 in its ‘purest’
form and has since been adopted in a number of other places in a variety of forms
(Shugart and Wattenberg, 2001; Colomer, 2004; Lundell, 2009).
At the most recent New Zealand general election in 2008, the country was
divided into 70 single-member constituencies where FPTP was used. A further 52
members of its House of Representatives were elected from a single (nation-wide)
list constituency. Each party’s share of the total number of seats in the House was
determined by the latter contest, so that a party with 31 percent of the votes would
be allocated 39 seats. If it won 28 seats in the FPTP contests, they would count
towards the 39 and a further 11 would then be allocated from the list presented by
the party in the other component of the contest. (If it won more seats in the FPTP
contests than the list vote allocation entitled it to, the party would retain those
extra seats and the size of the House of Representatives would be expanded until
the next election. According to this practice two extra seats were allocated after
the 2008 election, the Maori party getting five seats rather than three.) The result
of MMP elections is seen by some as the ‘best of both worlds’ – an overall PR
outcome but with a substantial number of legislators representing single-member
constituencies.7
The London Assembly
This system was adopted in 1999 for election of the London Assembly, which
has 25 members: 14 are elected by FPTP from single-member constituencies
and the remaining 11 from a London-wide closed list PR contest. There is no
überhangmandat, however, so the overall pattern of representation in the
Assembly may deviate somewhat from PR. That was indeed the case after the
2008 elections, when with 37.4 and 28.0 percent of the FPTP votes respectively
the Conservatives and Labour won all 14 single-member constituency seats (8 and
6 respectively). In the list contest, those two parties won 34.5 and 27.1 percent of
7 There is one major quirk to the system. If a party wins more first-past-the-post seats
than its proportional entitlement, as in Germany it retains them and the relevant number of
MPs is added to the overall Parliament, which increases in size until the next election: this is
known as the überhangmandat. In New Zealand, parties only qualify for seats in the House
if they either win at least 5 percent of the list votes or win at least one first-past-the-post
seat, which militates against the smaller parties gaining representation and so fragmenting
the Parliament – which might happen under ‘pure’ PR (as in Israel).
the votes. Parties getting less than 5 percent of the votes were not entitled to seats
and the Conservatives and Labour were allocated a further three and two seats
respectively, with the remaining six going to: the Liberal Democrats (11.2 percent
of the votes, three seats); the Greens (8.3 and 2); and the BNP (5.3 and 1) The
overall outcome was thus close to proportionality.
That London Assembly election illustrates the issues that parties have to face
under such an electoral system – especially the smaller parties. Should they contest
the single-member constituencies (which in London are very large, averaging
360,000 voters), knowing they are unlikely to win any of those seats, or should
they focus their attention on the list contest? If they choose the latter option, then
do they advise their supporters how to vote in the FPTP contests, where most
of their potential voters will presumably have to vote tactically for their second-
preference given that their first is not available? If they don’t, then the electors
will have to make the decision without any such advice or information: tactical
voting may be forced upon them – and may be especially difficult if their second
preference party is also unlikely to win in their single-member constituency (a
situation that could face a Green supporter whose second preference is the Liberal
Democrats).8
The National Assembly of Wales
As part of an agreement between several of the political parties – but not the
Conservatives – prior to the enactment of a Scottish Parliament in 1998, a form of
MMP/AMP was accepted as the electoral system to be deployed there – and it was
then also adopted for elections to the National Assembly of Wales. Because some
of the parties involved in those discussions were committed to PR as the basis for
elections, the Labour party – which expected to be the largest in each of the new
institutions, and some of whose leading members involved in the negotiations
were not averse to PR – agreed to the new system, believing that if it did not gain
a majority in the new legislatures it would at least be the largest party and able to
dominate a coalition government of the centre-left.9
8 The Liberal Democrats, the Greens and UKIP contested all 14 constituencies
in 2008, so that their supporters had to decide whether to vote for them in the almost
certain knowledge it would be a ‘wasted’ vote or to vote tactically for their second
choice. The BNP contested only one constituency, so its supporters had to vote tactically
in the other 13.
9 Labour formed a coalition administration with the Liberal Democrats after the first
election to the National Assembly of Wales in 1999; it was able to form a single-party
administration after the 2003 elections, having just 30 of the 60 seats, but it gained only 26
seats at the 2007 elections and entered a coalition with Plaid Cymru. In Scotland, Labour
was the largest party in coalitions with the Liberal Democrats after the 1999 and 2003
elections, but in 2007 the Scottish National Party was the largest party by one seat, and
formed a minority administration.
The system adopted varied from that used in New Zealand and London,
however. Instead of having a single PR list election that would determine the
overall composition of the legislature, Scotland and Wales were divided into a
number of regions, each of which had a separate list election. Thus in Wales, for
example, the 40 constituencies employed for election to the House of Commons
were used to elect the FPTP members of the National Assembly (AMs). These
were grouped into four regions, each of which elected four further members
from a list contest. The list votes suggested the overall party composition of the
region’s complement of AMs. Thus if a party won 33 percent of the list votes
in a region that was to return 12 AMs, it would be entitled to 4 members. If it
had five of its candidates elected from the FPTP contests, however, there was
no überghangmandat, and so it was over-represented, with one of its opponents
consequently under-represented.
This operation of this system is illustrated by the results of the 2007 Assembly
election, for each of the five regions (Table 3.5). The Mid and West Wales region
returned 12 AMs, so the PCQuota was 7.7. Four parties qualified for seats:
Plaid Cymru (which won 31 percent of all List votes, but 36.2 percent of the
total won by the four – 85.6 percent – that had quota places) was entitled to five;
the Conservatives three; and Labour and the Liberal Democrats two each. Plaid
Cymru had won four of the eight FPTP seats, and so was allocated one further
from its list of candidates; the Conservatives won two FPTP contests, and so were
also allocated a list seat; the Liberal Democrats won in two constituencies and so
gained no list AMs; with no FPTP victories, Labour got two list seats.
Whereas the result in that region was consistent with the role of the top-up
region element (as the list component is generally referred to) allocating seats
to parties that perform well in the list contest but win no FPTP seats, that was
not the case in the South Wales West region. There Labour won 35.8 percent of
the list votes, which meant a theoretical entitlement to four seats (the quota in
that 11-seat region was 8.3, and the four parties with quotas together obtained
82.0 percent of the list votes). But it won all seven FPTP seats, and so was over-
represented by three AMs, meaning that the other three parties together obtained
four of the seats rather than seven. Disproportionality was a consequence of the
distribution of votes and the way that these were translated into seats by the
system’s geography.
This example shows not only that the Scottish and Welsh version of MMP/
AMP does not necessarily produce an outcome consistent with PR – which the
New Zealand version does, because of the much larger, national list component
– but also that it can create both a situation of voter uncertainty and the potential
for tactical voting. In New Zealand electors can vote for their preferred party of
government in the list contest, knowing that because the composition of the House
of Representatives will reflect the distribution of list votes, then no vote will be
wasted (unless their preferred party is not going to meet the threshold of 5 percent
of the votes to be entitled to Parliamentary representation); they can then give
their FPTP vote to the candidate they think best able to represent the constituency,
Table 3.5 Seats and Votes by Region at the 2007 Elections to the National
Assembly of Wales
Party C PC LD L BNP UKIP G ∑
North Wales
Const. Seats 1 3 0 5 0 0 0 9
List Votes (%) 25.6 25.7 7.8 26.4 5.1 4.1 2.9
List Seats 2 1 1 0 0 0 0 4
Total Seats 3 4 1 5 0 0 0 13
Mid and West Wales
Const. Seats 2 4 2 0 0 0 0 8
List Votes (%) 22.9 31.0 13.3 18.4 2.9 3.8 4.0
List Seats 1 1 0 2 0 0 0 4
Total Seats 3 5 2 2 0 0 0 12
South Wales East
Const. Seats* 1 0 0 6 0 0 0 8*
List Votes (%) 20.0 13.6 11.0 35.8 4.7 4.6 2.8
List Seats 1 2 1 0 0 0 0 4
Total Seats 2 2 1 6 0 0 0 12*
South Wales Central
Const. Seats 1 0 1 6 0 0 0 8
List Votes (%) 21.7 15.5 14.0 34.0 3.8 3.7 3.8
List Seats 2 2 0 0 0 0 0 4
Total Seats 3 2 1 6 0 0 0 12
South Wales West
Const. Seats 0 0 0 7 0 0 0 7
List Votes (%) 16.1 17.7 12.4 35.8 5.5 3.6 3.8
List Seats 1 2 1 0 0 0 0 4
Total Seats 1 2 1 7 0 0 0 11
TOTAL
Const. Seats* 5 7 3 24 0 0 0 40*
List Votes (%) 21.5 21.0 11.7 29.6 4.3 4.0 3.5
List Seats 7 8 3 2 0 0 0 20
Total Seats 12 15 6 26 0 0 0 60*
Note: * an Independent candidate won one of the constituency seats. Key: Const. Seats – the
number of constituency seats won by each party; C – Conservative; PC – Plaid Cymru; LD
– Liberal Democrat; L – Labour; BNP – British National Party; UKIP – United Kingdom
Independence Party; G – Green; ∑ – Total.
knowing that whoever wins there will not alter the overall composition of the
legislature. They can vote sincerely in the list contest.
That is not the case in the British version, as illustrated by South Wales West.
There all of the list votes cast for Labour are wasted, because it can get no seats
additional to the seven won in the FPTP contests, unless it gets some 80 percent
of the list votes. As that is extremely unlikely, in order not to waste their list votes,
those who would otherwise have supported Labour should vote for their second
preference party – the one they would most like to see in the Assembly alongside
Labour (which would probably boost the representation of either or both of the
Liberal Democrats and Plaid Cymru rather than the Conservatives.). Of course,
they should only do that if they know that Labour is going to win all – or most –
of the FPTP contests, and they cannot be certain of that even if the other parties
canvassing for their second preference list votes suggested that they will. The
rational thing to do in such a circumstance is to vote a ‘split ticket’ – voting for
their first preference (Labour) in the FPTP contest but their second in the List.
(On split ticket voting at those contests, see Johnston and Pattie, 2002, and Johns
et al. 2010.) But such rational behaviour depends on perfect information and a
sophisticated understanding of how the system works in such circumstances: less
than a third had the latter (none could be absolutely certain of the former) and so
voted a split-ticket, with the remainder in effect wasting their list vote. Supporters
of the other parties, on the other hand, should have voted sincerely in the list
contest and tactically in the FPTP contests – supporting the candidate most likely
to defeat Labour (again calling for information and sophistication).
Although voters in the three South Wales regions might have been fairly sure
that Labour was going to win virtually all of the FPTP contests there, and so they
should structure their choices in the two contests accordingly, the likely outcome
at the regional level may have been less clear-cut in the other regions – so different
strategies might have been called for there. In this modified version of MMP/AMP,
it is very difficult for the voters to make rational decisions that will maximise the
efficacy of their two votes.
Alternative Voting (AV)
A further version of MMP/AMP has been debated in the UK over the last 15
years. In its 1997 general election manifesto, the Labour party promised to hold a
referendum on the electoral system – part of its uncertainty as to whether it would
win an outright victory and so in preparation for a potential coalition deal with the
Liberal Democrats. After the election (which it won with a very large majority)
it established an independent commission to recommend an alternative system
that met certain criteria (‘broad proportionality; the need for stable government;
an extension of voter choice; and the maintenance of a link between MPs and
geographical constituencies’ – which the Commission recognised in its report
were ‘not entirely compatible’). The Commission recommended a version of
MMP/AMP, but no referendum was ever held (Jenkins, 1998).
The system was similar to that adopted for Scotland and Wales with one major
difference – the ‘top-up’ regions were even smaller. The country was to be divided
into 80 regions (eight in Scotland, five in Wales, two in Northern Ireland, and
the remaining 65 in England): each would comprise a number of constituencies
returning MPs via FPTP and would also return no more than one or two further
MPs based on a list vote. This small list component meant that the system would
almost certainly not return a House of Commons whose composition was consistent
with PR, and the large number of FPTP constituencies would almost certainly
see the continuation of the disproportionality and bias typical of the system it
was supposed to replace. The list votes would provide some compensation to the
parties that came second or perhaps third in most regions but gained no (or very
few) FPTP seats there – and voters would be placed in the same tactical voting
uncertainty as those in some Welsh regions as discussed above.
The system has recently been revived, however, in a slightly revised – and
even more complex – form; some senior members of the Labour party suggested
its adoption, with the addition that AV would be used rather than FPTP in
the constituency contests, on the grounds that so doing would ensure that the
constituency MPs would have a majority mandate while the list component
would be a slight concession to PR supporters (especially the Liberal Democrats,
with whom Labour might enter a coalition). The elector would be faced with
an even more sophisticated set of choices to make – in which, in the absence
of the needed information, rational behaviour would be very difficult. On the
assumption that their preferred party would get their first rank in the AV contest,
they would then have to decide tactically which party should get their second
preference – and then whether their preferred, their second preference, or another
party still should get their list vote. In certain circumstances giving their three
votes to three different parties may be the rational decision – but whether that
could be reached (even by a party advising its first preference supporters) before
the event is very unlikely.
Conclusions
This chapter has not been structured to present an evaluation of the merits and
disadvantages of various electoral systems, let alone suggest that any one is
superior to the others. As has been stressed by many authors, there is no perfect
system, none that performs satisfactorily against the major criteria against which
they can be assessed (see, for example, Hix et al., 2010). Choosing an electoral
system thus involves a careful selection of the one that best meets the needs of the
particular case.
One feature that virtually every electoral system shares is that its geography
is open to manipulation in one or more ways – in the drawing of constituency
boundaries, for example, and in the practice of tactical voting (which may be
discouraged by the patterning of constituencies). In this chapter we have focused
on that aspect of various electoral systems currently either being deployed
or proposed for deployment within the United Kingdom. In so doing we have
emphasized various aspects of the crucial role that geography plays in the creation,
operation and electoral consequences of those electoral systems.
References
Balinski, M.L. and H.P. Young. 2001. Fair Representation: Meeting the Ideal of
One Man, One Vote. Washington, DC: The Brookings Institution.
Borisyuk, G., C. Rallings, R.J. Johnston, and M. Thrasher. 2010. Parliamentary
constituency boundary reviews and electoral bias: How important are variations
in constituency size? Parliamentary Affairs, 63: 4–21.
Carey, J. and S. Hix. 2011. The electoral sweet spot: Low magnitude proportional
electoral systems. American Journal of Political Science, 55: 383–97.
Colomer, J.M. (ed.) 2004. Handbook of Electoral System Choice. Basingstoke:
Palgrave Macmillan.
Cox, G.W. 1997. Making Votes Count: Strategic Coordination in the World’s
Electoral Systems. Cambridge: Cambridge University Press.
Dunleavy, P. and H. Margetts. 1998. Report to the Government Office for London:
Electing the London Mayor and the London Assembly. London: London
School of Economics, Public Policy Group.
Duverger, M. 1954. Political Parties. London: Methuen.
Electoral Reform Society. 2007a. STV in practice: A Briefing on the Northern
Ireland Assembly Election 21 March 2007. London: Electoral Reform Society.
Electoral Reform Society. 2007b. Local Authority Elections in Scotland. London:
Electoral Reform Society.
Electoral Reform Society. 2007c. Britain’s Experience of Electoral Systems.
London: Electoral Reform Society.
Farrell, D.M. and I. McAllister. 2006. The Australian Electoral System: Origins,
Variations and Consequences. Sydney: University of New South Wales Press.
Gallagher, M. 1974. Disproportionality in a proportional representation system.
Political Studies, 23: 501–13.
Gudgin, G. and P.J. Taylor. 1979. Seats, Votes and the Spatial Organisation of
Elections. London: Pion.
Hix, S., R.J. Johnston, and I. McLean. 2010. Choosing an Electoral System.
London: The British Academy.
Jaensch, D.M. 1983. The Australian Party System. Sydney: Allen & Unwin.
Jenkins, Lord. 1998. Report of the Independent Commission on the Electoral
System. London: HMSO, CM-4090I.
Johnston, R.J. 2002. Manipulating maps and winning elections: Measuring the
impact of malapportionment and gerrymandering. Political Geography, 21:
1–32.
Johnston, R.J. and J. Forrest. 2009a. Electoral disproportionality and bias under the
alternative vote: Elections to Australia’s House of Representatives. Australian
Journal of Political Science, 44: 521–28.
Johnston, R.J. and J. Forrest. 2009b. Geography and election results:
Disproportionality and bias at the 1993–2004 elections to the Australian
House of Representatives. Geographical Research, 47: 95–108.
Johnston, R.J., I. McLean., C.J. Pattie, and D.J. Rossiter. 2009. Can the boundary
commissions help the Conservative party? Constituency size and electoral bias
in the United Kingdom. The Political Quarterly, 80: 479–94.
Johnston, R.J. and C.J. Pattie. 2002. Campaigning and split-ticket voting in new
electoral systems: The first MMP elections in New Zealand, Scotland and
Wales. Electoral Studies, 21: 583–600.
Johnston, R.J. and C.J. Pattie. 2008. How much does a vote cost? Incumbency
and the impact of campaign spending at English general elections. Journal of
Elections, Public Opinion and Parties, 18: 129–52.
Johnston, R.J. and C.J. Pattie. 2011a. The British general election of 2010: A three-
party contest or three two-party contests? Geographical Journal, 177: 17–26.
Johnston, R.J. and C.J. Patti. 2011b. The local campaigns and the outcome. In
N. Allen and J. Bartle, eds, Britain at the Polls, 2010. London: Sage
Publications.
Johnston, R.J., C.J. Pattie, D. Dorling, and D.J. Rossiter. 2001. From Votes to
Seats: The Operation of the UK Electoral System since 1945. Manchester:
Manchester University Press.
Johnston, R.J., D.J. Rossiter, and C.J. Pattie. 2006. Disproportionality and bias
in the results of the 2005 general election in Great Britain: Evaluating the
electoral system’s impact. Journal of Elections, Public Opinion and Parties,
16: 37–54.
Johns, R., D. Denver, J.K. Mitchell, and C.J. Pattie. 2010. Voting for a Scottish
Government: The Scottish Parliament Election of 2007. Manchester: Manchester
University Press.
Lundell, K. 2009. The Origin of Electoral Systems in the Postwar Era: A Worldwide
Approach. London: Routledge.
McLean, I. 1996. E.J. Nanson, social choice and electoral reform. Australian
Journal of Political Science, 31: 369–85.
McLean, I. and R.J. Johnston. 2009. When is proportional not proportional? Great
Britain’s 2009 elections to the European Parliament. Representation, 45: 349–
55.
Mandelson, P. 2010. The Third Man. London: Harper Collins.
Ministry of Justice, 2008. The Governance of Britain – Review of Voting Systems:
The Experience of New Voting Systems in the United Kingdom since 1997.
London: HMSO, CM-7304.
Oelbermann, K.-F., A. Palomares and F. Pukelsheim. 2010. The 2009 European
Parliament elections: From votes to seats in 27 ways. European Electoral
Studies 5: 148–82.
Paddison, R. 1976. Spatial bias and redistricting in proportional representation
electoral systems: A case study of the Republic of Ireland. Tijdschrift voor
Economische en Sociale Geografie, 67: 230–40.
Parker, A.J. 1982. The ‘friends and neighbours’ effect in the Galway West
constituency. Political Geography Quarterly, 1: 243–62.
Parker, A.J. 1983. Localism and bailiwicks: The Galway West constituency in the
1977 general election. Proceedings of the Royal Irish Academy, Section C, 83:
17–36.
Parker, A.J. 1986. Geography and the Irish electoral system. Irish Geography, 19:
1–14.
Pattie, C.J. and R.J. Johnston. 2009. Still talking, but is anybody listening? The
changing face of constituency campaigning in Britain, 1997–2005. Party
Politics, 15: 411–34.
Plant, R. 1993. Report of the Working Party on the Electoral System. London: The
Labour Party.
Rallings, C., M. Thrasher and D. Cowling. 2002. Mayoral referendums and
elections. Local Government Studies, 28 (4): 67–90.
Renwick, A. 2010. The Politics of Electoral Reform. Cambridge: Cambridge
University Press.
Rossiter, D.J., R.J. Johnston and C.J. Pattie. 1999. The Boundary Commissions:
Redrawing the UK’s Map of Parliamentary Constituencies. Manchester:
Manchester University Press.
Shugart, M.S. and M. Wattenberg (eds) 2001. Mixed-Member Electoral Systems:
The Best of Both Worlds? Oxford: Oxford University Press.
Szpiro, G. 2010. Numbers Rule: The Vexing Mathematics of Democracy from
Plato to the Present. Princeton, NJ: Princeton University Press.
van der Kolk, H., C. Rallings and M. Thrasher. 2004. Electing mayors: A
comparison of different electoral procedures. Local Government Studies, 30:
589–608.
van der Kolk, H., C. Rallings and M. Thrasher. 2006. The effective use of
the supplementary vote in Mayoral elections: London 2000 and 2004.
Representation, 42: 91–102.
Chapter 4
Spatial Regression for Electoral Studies:
The Case of the Italian Lega Nord
Michael Shin and John Agnew
Spatial regression, and more generally, spatial analysis, are premised upon the
fact that spatially referenced data are “special” (O’Sullivan and Unwin 2003).
Spatially or geographically referenced data refer to data that are collected on the
basis of identifiable locations or places. What makes spatial data special is that they
tend to be biased by the spatial interdependence (or linkages) of places for which
data are collected, and frequently, because the values of individual observations
are not independent, but are correlated with each other. Such biases are usually
manifest as clusters or concentrations of similar values at the local scale, which
in turn can lead to notable regional differences and variations at a wider scale of
analysis. Detecting and evaluating such geographic biases, referred to as spatial
dependence or spatial autocorrelation at the local level, and spatial heterogeneity
at the regional scale, are the primary foci of contemporary spatial analysis.
This chapter outlines and discusses how spatial analysis and spatial
regression techniques can inform and extend ecological (i.e., aggregate data)
analyses of voting behavior. The first section provides an overview of the key
methodological issues surrounding the evaluation of spatial autocorrelation
and spatial heterogeneity in geographic data sets. Next, a worked example
that focuses on the recent success of the Lega Nord is presented. This party,
claiming to represent the interests and identities of northern Italians in a polity
that the party’s leaders see as biased in favor of the Italian South, has achieved
considerable success in recent Italian national and local elections. In coalition
with the main national center-right party, the PdL, the Lega has formed part of
Italian national governments in 2001–6 and since 2008. In recent years the Lega
has become obsessed with seeing Italy adopt some sort of fiscal federalism,
with local governments becoming much more important than the national one
in raising revenues and making expenditures across a range of functions, even
though in the late 1990s it had been avowedly separatist. It balances precariously
between being a party of government, a protest movement akin to the Tea Party
movement in the US, and a regional party similar to those in Catalonia and
Scotland. The chapter concludes with a summary discussion of various issues
that arise when using the spatial analytic and geo-statistical techniques to
examine the geography of support for the Lega Nord.
Geographic Information Systems, Spatial Analysis, and Spatial Regression
The influences of geography and geographic context upon both natural and
social processes and phenomena, ranging from precipitation to voting behavior,
are topics of increasing interest. Thanks to recent advances and innovations in
information technology (e.g., Google Earth), levels of geographical awareness
and curiosity have increased dramatically in recent years. One special class of
information technology that has been central to this shift towards geographic
thinking is geographic information systems (GIS). Briefly, GIS is a special type
of software that integrates data from various sources, and permits such data and
the relationships between variables to be visualized as maps. GIS also serves as
a platform for more robust analytical techniques, such as exploratory spatial data
analysis (ESDA) and spatial regression analysis.
The key requirement for using GIS for spatial analysis is access to geo-
referenced data. The terms ‘geocoded’, ‘spatially referenced’ and ‘geospatially
referenced’ are interchangeable with the term ‘geo-referenced’. Geo-referenced
data refer to data that can be associated with identifiable locations (e.g., geographic
coordinates, addresses, postal codes, regions, comuni). As noted above, the
locations where data are collected often introduce, shape or produce place-specific
biases. Moreover, data that are collected from proximate locations often share
certain traits, for instance, similarity in type, size or magnitude. In statistical terms,
the value of a variable at one location is often correlated with nearby values, and
is not independent as classical probability theory necessitates.
When geo-referenced data are mapped, such biases are usually manifested as
spatial clusters or ‘hotspots’ of similar values at the local scale, and can also result
in notable regional-level variations. Detecting and evaluating such geographic
biases, referred to formally as spatial dependence or spatial autocorrelation at the
local level, and spatial heterogeneity or spatial variation at the regional scale, are
the objectives of spatial analysis. Moreover, the results from such analyses serve
to inform and refine subsequent analyses, models and avenues of inquiry.
For example, consider the series of quartile maps of support obtained in the
2008 national election for the Partito Democratico (PD), Popolo della Libertà
(PdL) and Lega Nord (LN) at the provincial level (Figure 4.1). The PD is the main
center-left party, the PdL is the main center-right party, and the LN is the northern
regional party. The PD (and its political ancestors) is strong in central Italy, the
PdL in the metropolitan north and the South, and the LN, not surprisingly, in the
North. Levels of both high and low support for each respective party tend to be
concentrated across certain areas of the Italian peninsula. Put another way, the
geographic distribution of party support for the PD, PdL and LN is not random.
What is more interesting, however, is how localized and/or regionalized these
patterns truly are. This cannot be determined from simply “eye balling” or staring
at the maps.
To test whether this visual clustering of support for each party is statistically
significant, the local Moran’s I statistic is calculated. Moran’s I provides a formal
Spatial Regression for Electoral Studies 61
Figure 4.1 Support Obtained in the 2008 National Election for the Partito
Democratico (PD), Popolo della Libertà (PdL) and Lega Nord
(LN) at the Provincial Level
Source: Authors.
method for identifying and evaluating spatial autocorrelation (Anselin 1995a).

The localized Moran’s I is formally expressed as:
zi
Ii = ∑ wij zj
∑ zi2 j
where z are standardized levels of support for a given party, and w is an element in a
row-standardized spatial weights matrix, W, that summarizes the spatial relationships
between the Italian provinces (see Anselin 1995a, 1995b; Shin and Agnew 2008). The
component, Σjwijzj , is called the spatial lag of z and is in effect the weighted average
of values that surround any given observation. In this example, W summarizes the
relationship between each province and its immediate (i.e., contiguous) neighbors.
Used in many statistical indicators of spatial association, spatially lagged variables
provide a simple way to capture and to summarize geographic relationships in
quantitative data. Inference for the local Moran’s I is based upon a conditional
randomization method (see Anselin 1995a; Anselin 1995b).
Note that the mean of all local Moran’s I values is equal to Moran’s I, the statistic
most commonly used to detect the presence of spatial autocorrelation across a data set.
Values of Moran’s I fall between negative one and positive one, with a positive value
indicating the presence of clustering and a negative value indicating a chessboard-like
pattern of geographic difference (for details, see O’Sullivan and Unwin 2003). The
significance of Moran’s I, reported as a z-score, is based upon the assumption that
values are dispersed randomly across the study area (Cliff and Ord 1981).
The values of Moran’s I for the PD (I = 0.67), PdL (I = 0.53) and LN (I = 0.92)
in the 2008 Italian general election are all high and statistically significant. In other
Figure 4.2 “Hotspot Maps” Showing Local Morgan’s I Score for Each
Party in the 2008 Italian National Election
Note: No spatial outliers were detected.
Source: Author.
words, voter support for a given party in one province is indeed associated to voter
support in neighboring provinces. Since a local Moran’s I score is calculated for
each province, significant I values can be mapped to identify and locate “hotspots”
of positive spatial autocorrelation (i.e., high values surrounded by similarly high
values, or low values surrounded by similarly low values), or spatial outliers (i.e.,
high values surrounded by low values, or low values surrounded by high values).
“Hotspot maps” for each party are presented in Figure 4.2.
Statistically significant clusters of high support for the PD, PdL and LN are
shaded in dark, and clusters of low support are shaded more lightly. No spatial
outliers of low or high support surrounded by dissimilarly high or low support
were detected. The ability to identify the location of such hotspots and outliers
provides important insights about which places may merit further exploration
and explanation. Though the territorial dimensions to Italian politics are well
recognized (e.g., Galli and Prandi 1970; Cartocci 1990; Shin and Agnew 2008),
the spatial analytic techniques presented above serve to complement and extend
such explanations of the political geography of Italy. Such methods can also be
used to inform strategies for pre- and post-election surveys and interviews.
The detection of spatial autocorrelation in a data set also has implications for
other statistical techniques, and in particular regression analysis. For instance,
both the independent and dependent variables used in a regression may exhibit
varying levels of spatial dependence. Furthermore, it is relatively common for the
residuals from a regression that uses spatially referenced data to be contaminated
by spatial autocorrelation, thus violating the assumption of the independence
of errors. Mapping the residuals from a regression, and checking for clusters of
similar values and directionality (i.e., +/-), is a relatively simple and straightforward
method to check for the presence of spatial autocorrelation.
The presence of spatial autocorrelation within a data set has several implications
for the parameter estimates from an ordinary least squares (OLS) regression (i.e.,
y = α + βX + ε). When the value of a dependent variable in a given location, yi ,
is influenced by or associated with spatially proximate values for the dependent
variable (i.e., yj ) or independent variables (e.g., xj ), parameter estimates will be
biased and inefficient (i.e., parameter estimates for the unknown true value, β, are
either too high or too low). When this occurs, it is called spatial structure in the data.
If the error term from a regression is contaminated with spatial autocorrelation,
parameter estimates will be inefficient (i.e., inflated standard errors for parameter
estimates). Subsequently, this can lead to incorrect inferences because possibly
significant parameter estimates will not be evaluated as such. In light of recent
interest and attention surrounding spatial analytic methods, formal specification
diagnostics that test for the presence of spatial autocorrelation, for example, based
on Moran’s I, have been developed and are used widely.
To combat spatial structure in a data set, as well as the contamination of the
error term with spatial autocorrelation, two spatial econometric strategies can be
employed (see Anselin 1988). In the latter case, a spatial error regression model can
be estimated where the spatial lag of the error term (i.e., Wε) is included on the right
hand side of each regression (i.e., y = α + βX + Wε + μ). The spatial lag of the error
term, Wε, is obtained by multiplying a spatial weights matrix, W, and the residuals, ε.
The spatial lag of the error returns the weighted average of geographically proximate
residual values for each observation. The inclusion of the spatial lag, Wε, controls
for spatial autocorrelation in the error term, ε, and returns a residual term, u, that is
distributed normal and independent. More importantly, the spatial error model yields
more efficient parameter estimates and more reliable significance tests.
To mitigate the effects of spatial structure in a regression data set, a spatial
autoregressive or spatial lag model can be estimated, in which the spatially lagged
dependent variable, Wy, is included on the right hand side of the equation (i.e., y
= Wy+α+βX +μ). As with the spatial error model, the spatially lagged dependent
variable, Wy, is in essence the weighted average of values (e.g., vote share) from
surrounding observations (e.g., provinces) and uses the same spatial weights
matrix, W. Maximum likelihood estimation is used to fit such models and returns
unbiased parameter estimates.
One of the assumptions underlying the spatial econometric techniques described
above is that parameter estimates are homogenous, or that the relationships
between variables are uniform and consistent across the data set. Depending upon
the geographic context and spatial resolution of the data used in the analysis, this
assumption may be untenable. For instance, it is well-established that Italian politics,
economics and society exhibits a significant amount of geographic variation. Such
variation, formally termed spatial heterogeneity, is arguably a function of how
various processes operate differently in different places. In the context of regression
analysis, several methods have been developed to account or to control for spatial
heterogeneity, such as the inclusion of regional dummy variables, trend-surface
analysis, and the estimation of separate models for subsets of a data set.
One innovative technique used to model processes that vary spatially

is geographically weighted regression (GWR) (see Brundson et al. 1996;
Fotheringham et al. 2002). Generally, GWR is concerned with estimating
parameters anywhere across a geo-referenced data set. Premised upon the
argument that spatially proximate observations are more likely to be similar with
each other than with more distant observations, the parameter estimation in GWR
gives precedence to proximate observations over more distant ones. The GWR
model can be expressed as:
yi = β (u) + β (u) x1i + β2i (u) x + … + βmi (u) x
0i 1i 2i mi
where β0i(u) is the parameter estimate around location u and is specific to that
particular location. Parameter estimates can be obtained for each and every
location for which there are values for the dependent (i.e., y) and independent
(i.e., x) variables.
In order to obtain local parameter estimates for each and every observation
in a data set, a geographical weighting scheme that describes and summarizes
the spatial relations between observations must be specified. Generally, GWR
uses a distance-decay function, with proximate observations having more local
influence than more distant observations. Within the context of GWR, this
involves specifying a bandwidth and a kernel. The bandwidth can be considered
a smoothing parameter, with larger bandwidths resulting in a smoother surface of
estimates. The kernel refers to whether the bandwidth is constant across the study
area (i.e., fixed kernel) or allowed to vary (i.e., variable kernel). In case of the
former, a Gaussian function is used to weight all observations in a data set, but in
the latter, a bi-square function is implemented in conjunction with the bandwidth
that determines the number of observations to be used to estimate local parameters.
Unlike OLS regression that returns a single parameter estimate for each
modeled relationship, GWR returns a set of estimates and standard errors that are
associated with the location of each observation that can and should be mapped. The
interpretation of the GWR estimates is identical to that used in OLS, except for the
fact that there are far more estimates to be interpreted. Changes in the magnitude
and direction of the estimates across a data set indicate that the relationships
between variables in a model vary geographically (i.e., spatial heterogeneity).
Measures of fit for GWR rely upon the Akaike Information Criterion (AIC)
which provides a method to compare GWR models that use different mixes of
independent variables to each other, as well as to the global OLS model. It should
be noted that GWR presents several methodological challenges with regard to
significance tests and analysis of variance and are the subject of ongoing research
(Charlton and Fotheringham 2009).
Election data, and in particular, Italian election data, are well suited for
analysis with GWR in light of the geographic dimensions to Italian politics. The
next section provides a worked example of GWR, and examines recent support for
the Lega Nord within one of its regional strongholds: Lombardy.
A Worked Example: The Resurgence of the Lega Nord
One of the unexpected results of the Italian general election of 2008 was the
resurgence in popular support for the Lega Nord. Famous for its rhetoric in defense
of local and regional interests in northern Italy, hostility to foreign immigrants,
and governmental alignment with Silvio Berlusconi and his various parties (first
Forza Italia and now PdL), support for the Lega Nord nearly doubled since the last
election of 2006. GWR is used to examine in detail this recent change in electoral
fortunes for the Lega Nord across the party’s regional stronghold of Lombardy.
Lombardy is a good setting to examine the recent success of the Lega Nord
with GWR because it is the recognized stronghold for the party, it is economically
diverse, home to small, medium, and large scale enterprises, and serves as a beacon
to new immigrants (Mainardi 1998). There are also a relatively large number
of comuni (municipalities) across the region, which arguably contributes to
administrative fragmentation. Figure 4.3 maps the 2008 electoral performance of
the Lega Nord across Lombardy and highlights the geographic concentrations and
variations in support for the party. At the national level, the Lega Nord obtained
8.3 percent of the vote in 2008, yet the party averaged over 25 percent within its
regional stronghold. High levels of support for the Lega tend to be concentrated in
the northern reaches of Lombardy, to the north of Bergamo and around Sondrio.
Support for the Lega Nord decreases considerably across the south of the region, as
well as in the urban centers of Lombardy (e.g., Milano, Pavia, Cremona, Brescia).
The recent election also permits us to test some general hypotheses about issues
important to supporters of the Lega Nord, and in particular, questions concerning
immigration and immigrants.
Despite the adages that globalization and Europeanization are good and that
many countries have been built upon the work of immigrants, in many parts of
the world, recent immigrants tend to be viewed more as a threat than as an asset.
Furthermore, there are several aspects to this perceived threat that immigrants pose
that political parties and politicians integrate into their own platforms, campaigns
and rhetoric. Perhaps the most common anti-immigrant argument is based upon
economics. Following simple (but flawed) logic is the fear that immigrants take
jobs away from local workers. Moreover, since recent immigrants are surely poor
and willing to work for less, their employment undercuts wages. Paralleling this
argument is concern and consternation over the provision and abuse of local social
services to immigrants and especially undocumented immigrants.
Exacerbating such economic arguments against immigration and immigrants
are those based on cultural fear and social insecurity. By way of example,
criminal acts occur with similar relative frequency across all societies, but those
committed or believed to be committed by immigrants tend to resonate more with
local populations for several plausible reasons. For instance, local identities and
traditions are often considered to be under threat or displaced by the arrival of
immigrants and foreigners are seen as more likely than locals to violate accepted
social norms as defined in criminal law.
Figure 4.3 2008 Electoral Performance of the Lega Nord Across Lombardy
Source: Authors.
The Lega Nord is renowned for its continuing negative stance on immigration,
as well as for its more positive but equally divisive views on political and economic
issues such as devolution and federalism (Agnew 1995; Biorcio 1997; 1999; Diamanti
1993; 1996). In light of such political positions, and the discussion above about how
immigration can be seen as a particularly important source of voter support for the
Lega, three elementary but general hypotheses can be articulated to explore and
possibly explain the resurgence in support for the Lega Nord using GWR:
• H1: League support may be positively related to the presence of immigrants

(immigrant threat)
• H2: League support may be positively related to unemployment (immigrants
take jobs)
• H3: League support may be positively related to self-employed (defense of
the local)
The above hypotheses draw from Lega rhetoric during recent election campaigns,
and its historical anti-immigrant position. For instance, if indeed immigrants were
taking jobs away, perceived as a threat to local traditions or involved in criminal
activities, a positive relationship between Lega support and the relative size of
the immigrant population is plausible. There is the possibility that such perceived
threats occur in the local absence of migrants and are the result of second-hand
Table 4.1 Descriptive Statistics for Variables in Lombardy Data Set

(N = 1,546)
Variable Mean Std. Dev. Minimum Maximum

Lega Nord 2008 25.8% 8.7% 5.3% 64.7%
Log (population) 3.71 0.54 1.81 6.41
Unemployed 4.4% 1.6% 0.0% 16.2%
Self-employed 17.0% 5.0% 4.6% 50.0%
Foreigners 6.7% 3.8% 0.0% 23.0%
Table 4.2 Baseline OLS Model Estimates for Lega Nord Support in
Lombardy
Variable Coefficient Std. Error t-value
Constant 0.480 0.022
log (population) -0.039 0.004 -8.65
Unemployed -0.731 0.133 5.50
Self-employed -0.144 0.047 -3.05
Foreigners -0.272 0.058 -4.70
information (e.g., via media reports or anti-immigrant rhetoric and propaganda),

a general fear of the unknown and not direct experience. Keeping this in mind,
the hypotheses presented above are used primarily for illustrative purposes. Using
aggregate data and GWR, the above arguments as they relate to the recent success
of the Lega Nord are tested.
The following analyses are based on data obtained from the Cattaneo Institute
in Bologna and the Italian national statistics bureau ISTAT for the 1,546 comuni in
Lombardy. Descriptive statistics for the relevant variables are provided in Table 4.1.
Across Lombardy, the mean level of unemployment is relatively low, and
average levels of self-employed and foreigners are moderately high, especially
when compared to the rest of Italy. Each of the variables however exhibits a
considerable amount of variation across the region. Using the above variables, a
baseline, global OLS regression is estimated. Results from the OLS are reported
in Table 4.2. The goodness of fit measures indicate that the model explains a
relatively small amount of the total variance (i.e., R2 = 0.09). Despite this relatively
small value, parameter estimates from the model are statistically significant.
However, the direction of the relationships between support for the Lega Nord
and unemployment, self-employment and the number of foreigners are opposite
of what was hypothesized. This may be the result of model (or theoretical)
misspecification and merits further investigation.
Spatial diagnostic tests (unreported) indicate that a spatial lag model is
preferred over a spatial error model. Estimates for the spatial lag model are
reported in Table 4.3. As a measure of fit in the spatial lag model, a pseudo-R2
index is calculated that compares the predicted variance to the observed variance
in the model (see Anselin 1992). This value is 0.77, but perhaps more interesting is
Table 4.3 Spatial Lag Model Estimates for Lega Nord Support in
Lombardy
Variable Coefficient Std. Error t-value
Constant 0.092 0.013
spatial lag (Wy) 0.858 0.014 62.17
log (population) -0.010 0.002 -4.32
Unemployed -0.334 0.070 -4.95
Self-employed 0.026 0.024 1.09
Foreigners -0.103 0.029 -3.48
a comparison of the parameter estimates between the spatial lag and baseline OLS
models. Specifically, the estimates for population, unemployed and foreigners
remain significant but decrease in magnitude in the spatial lag model, and the self-
employed variable becomes positive, though it is no longer statistically significant.
As with the OLS model, the direction of the predicted relationships are opposite of
what was expected. The spatial lag of y is remarkably significant and indicates that
the spatial structure within the data cannot be ignored.
Recall that the spatial lag model assumes that parameter estimates are
homogeneous or stable across the data set. To test this assumption and to explore
further the spatial structure of the Lombardy data set, geographically weighted
regression (GWR) is used. This analysis was carried out with the ArcGIS 9.3. An
adaptive kernel was used, and bandwidth was selected on the basis of minimizing
the AIC value (see Fotheringham et al. 2002 for details). The procedure for
determining bandwidth is automated and involves assessing difference between
observed and fitted values, as well as the complexity of the model. As noted
earlier, estimates for each and every observation are returned by GWR, and are
best visualized through maps.
The first map from the GWR is for the standardized residuals (Figure 4.4).
Darker shades indicate areas where the model under-predicted Lega support and
lighter shades indicate areas where the model over-estimated Lega support. This
map shows clearly that support for the Lega Nord is underestimated consistently
in the urban centers of Lombardy (e.g., Milano, Brescia) Moreover, in some areas,
such as around Milano and Brescia, and to the southeast of Cremona, there are
pockets or clusters of residuals of similar magnitude. Such patterns reinforce the
need to consider carefully the geography of elections, and even call into question
the utility of a single parameter estimate for a region such as Lombardy.
Maps of the local coefficients for each independent variable (except population)
clearly show the spatial structure that underlies the data, as well as the spatial
heterogeneity in the nature of relationships for a given variable, as well as between
variables. It is useful to keep in mind the actual estimates obtained from both the
OLS (Table 4.2) and spatial lag (Table 4.3) models for comparison and as a starting
point for interpretation. Recall that the estimates for unemployment returned by the
OLS and spatial lag models were, -0.73, and, -0.33, respectively. The GWR map
(i.e., surface) of local unemployment coefficients shows where in Lombardy such
Figure 4.4 Geographically Weighted Regression Residuals Across Lombardy

Source: Authors.
estimates, or in other words, the relationship between support for the Lega Nord
and unemployment, are both positive and negative (Figure 4.5). For instance, the
relationship is positive in areas to the north of Brescia, the west of Sondrio and
the east of Mantova. Conversely, there is a strong inverse relationship between
unemployment and Lega Nord support to the east of Milano. To put into context
the geographic variation of this relationship, the local GWR estimates vary from a
minimum of -3.17 around Milano to a maximum of +0.76 in a few of the comuni
in the northern and eastern extremities of Lombardy. Hence, in some areas the
results from the GWR support our initial hypotheses regarding the linkage between
unemployment and support for the Lega, yet in other areas our hypothesis is rejected.
The map of GWR estimates for the association between foreigners and Lega
support also reveals some areas where the relationship is negative and one area in
particular where it is positive (Figure 4.6). With regard to the latter, the cluster to
the east of Milano and south of Bergamo contains GWR estimates between +0.37
and +0.58. For comparison, OLS returned an estimate of -0.27 and the spatial
lag model returned an estimate of -0.10. Though this result (i.e., cluster of GWR
estimates) supports our hypothesis that the presence of foreigners is linked to
higher levels of support for the Lega Nord, this is only the case in a certain area
of Lombardy. The map also identifies areas of Lombardy, and in particular to the
east and west of Sondrio, where the association between number of foreigners
and Lega support is negative and more in line with the estimates returned from
Figure 4.5 Geographically Weighted Regression Map of Local

Unemployment Coefficients Across Lombardy
Source: Authors.
the OLS and spatial lag models. Such discrepancies in both parameter estimates
across models, as well as returned by GWR, may be a result of theoretical or model
misspecification. Regardless the cause of such differences, such discrepancies and
geographic variations merit careful interpretation and consideration.
The map of local coefficients for the relationship between the self-employed
and Lega support again shows how the association is positive in some areas of
Lombardy, but negative in others (Figure 4.7). The most conspicuous cluster
of positive estimates is situated between Milano and Bergamo, in the heart of
Lombardy. The magnitude of the association between self-employment and Lega
Nord support tends to decrease as distance increases from this central cluster,
with the exception of a local concentration of positive local estimates around
Mantova. The pockets of negative estimates appear on the western, northern and
eastern borders of Lombardy. Like the previous maps, the results from the GWR
highlight the geographic instability in the relationship, in this case, between the
self-employed and Lega support. Again, depending upon where GWR coefficients
are obtained, our hypothesis could either be accepted or rejected at the local level.
The above worked example illustrates the potential that GWR has within the
context of electoral studies and highlights how it differs from both OLS and spatial
econometric techniques. The exact choice of regression technique to use when
examining a given election is perhaps less clear, and is likely to reflect exposure and
Figure 4.6 Geographically Weighted Regression for the Association Between

Foreigners and Lega Nord Support
Source: Authors.
familiarity with geographic concepts, methods and issues (e.g., scale, spatial analysis,
spatial autocorrelation) in general. For instance, OLS is used and understood widely,
which is arguably its greatest strength. Despite recent innovations and diffusion of
geo-statistical and spatial analytic techniques, the simplicity and accessibility of
OLS is its greatest advantage. As noted earlier, however, is the fact that many data
sets are spatially biased, which in turn can bias or influence the results from OLS.
Spatial econometric methods, such as the spatial lag model, introduce a level
of complexity to regression that in some cases may be necessary in order to obtain
unbiased and efficient parameter estimates. The trade-off when using geo-statistical
techniques versus OLS is that much more additional information about a particular
data set must be obtained or generated. For instance, the geographic locations
of each observation, and the geographic relations between observations, need to
conceptualized and recorded. Though this is a trivial exercise within the context of
geographic information systems (GIS), it is a non-trivial issue for the uninitiated.
What is more is that the interpretation of spatial econometric models is not
necessarily straightforward, but depends upon whether geography (i.e., the spatial
bias within a data set) is conceived and understood as merely a nuisance that needs
to be controlled or a substantive effect that needs to be explained
Furthermore, conceptualizing geography as simply an additional parameter within
a regression model is in many ways problematic and unsatisfying because geography
Figure 4.7 Map of Local Coefficients for the Relationship Between the Self-
Employed and Lega Nord Support
Source: Authors.
is not necessarily an additive effect; rather, geography contextualizes the variables

and relations of interest differently in different places. In other words, unemployment
in and around Milano may have different connotations and impacts upon the local
population when compared to unemployment in a more rural setting such as Sondrio.
As an exploratory technique, GWR perhaps provides the most (local) insights
into the geographic nature of relationships in aggregate data sets. In addition to
the methodological complexity and technical requirements of GWR, one of the
greatest trade-offs is the simplicity in the interpretation of results. In the case
of evaluating and assessing relationships with OLS and spatial econometric
techniques there is only a single coefficient to interpret. GWR returns a coefficient
for each and every observation. Mapping such estimates certainly facilitates
interpretation, but there is also the presumption that such maps as well as the GWR
results themselves can be interpreted and understood. Furthermore, though GWR
can provide interesting if not valuable insights into the geographic instability of
regression coefficients, assessing statistical significance remains problematic,
and perhaps more importantly, GWR has the potential to introduce a prejudice
for the local. Just as OLS and spatial econometric methods can be criticized for
returning single or global parameter estimates for a data set, are we missing the
forest for the trees with GWR? Such questions and trade-offs surrounding the use
and application of OLS, spatial econometrics and GWR are more about how and
why geography matters to electoral studies, and are less about methodology per se.
Conclusion
The spatial regression techniques discussed and implemented above hold

considerable potential for electoral studies and analysis. In the absence of survey
data, or used in conjunction with such individual-level data, spatial regression
can complement and extend our understanding of voter attitudes and behavior
in several ways. For instance, such methods can be used to refine and articulate
research questions, and in particular, questions concerned with the influence of
geographic context upon political behavior and electoral change. Analyses such as
that presented above can also be used to guide subsequent studies, direct fieldwork
or even inform sampling strategies for voter surveys.
What is clear is that the spatial regression techniques provide a useful and
accessible means to evaluate and assess how spatial autocorrelation and spatial
heterogeneity are possibly implicated in models of voter behavior, and more
generally, the voting choice itself. In the worked example that modeled 2008 support
for the Lega Nord, it is clear that when the spatial structure underlying the data are
disregarded and ignored, the parameter estimates obtained via OLS are biased. The
introduction of a spatial lag term to control for spatial autocorrelation (i.e., Wy) in the
spatial lag model provides insights into the nature and degree of parameter estimate
bias, as well as a potential control for the underlying spatial structure of the data.
Geographically weighted regression (GWR) was then introduced to extend
further the geostatistical modeling of recent support for the Lega Nord. The
added value of the GWR approach, over both the standard OLS and spatial lag
models estimated, is its ability to evaluate and show how parameter estimates vary
geographically and locally. In other words, GWR recognizes that the relationships
between variables may not be the same, or homogeneous, across a given data
set. By obtaining local parameter estimates, such geographic variation provides
more detailed and nuanced insights into the relationships between, for example,
Lega Nord support, socio-economic indicators and geography. At the very least,
the spatial regression methods outlined highlight the need to consider geography
when estimating models with georeferenced data, and may in fact promote a more
explicit treatment of geography with regard to theories of political behavior.
Finally, it should be noted that the choice of how such geographic relationships
between observations are captured by GWR or by the spatial weights matrix, W, in
the case of the spatial econometrics, is non-trivial and can have implications for the
interpretation of geographic effects. In this regard, it is important to determine how and
why geography matters. This last point is a theoretical issue, and not a methodological
or technical one. For example, what processes and mechanisms account for Lega Nord
support in one comune to be related to those in neighboring comuni? How consistent
is this relationship within a given province or across Lombardy? What accounts for
the variation in the nature of the relationships between support for the Lega Nord
and unemployment, self-employment and immigrant presence across Lombardy?
As this example illustrates, spatial analysis, spatial econometrics and GWR hold a
considerable amount of promise with respect to answering and extending research
questions in electoral studies, as well as articulating and informing future directions
for research in political science and allied disciplines.
References
Agnew, J. 1995. The rhetoric of regionalism: The Northern League in Italian

politics. Transactions of the Institute of British Geographers 20: 156–72.
Anselin, L. 1995a. Local indicators of spatial association – LISA. Geographical
Analysis 27: 93–115.
Anselin, L. 1995b. SpaceStat, version 1.80 user’s guide. Morgantown, WV:
Regional Research Institute.
Anselin, L. 1988. Spatial Econometrics: Methods and Models. Dordrecht: Kluwer
Academic.
Biorcio, R. 1999. La Lega nord e la transizione italiana. Rivista italiana di scienza
politica 22: 55–88.
Biorcio, 1997. La Padania promessa. La storia, le idee e la logica d’azione della
Lega Nord. Milano: Il Saggiatore.
Brunsdon, C., A. Fotheringham and M. Charlton. 1996. Geographically weighted
regression: A method for exploring spatial non-stationarity. Geographical
Analysis 28: 281–98.
Cartocci, R. 1990. Elettori in Italia. Bologna: Il Mulino.
Charlton, M. and A. Fotheringham. 2009. Geographically Weighted Regression.
National Centre for Geo-computation White Paper. Maynooth, Ireland:
National University of Ireland.
Cliff, A. and J. Ord. 1981. Spatial Processes: Models and Applications. London: Pion.
Diamanti, I. 1996. Il male del Nord. Roma: Donzelli.
Diamanti, I. 1993. La Lega. Roma: Donzelli.
Fotheringham, A., C. Brunsdon and M. Charlton. 2002. Geographically Weighted
Regression: The Analysis of Spatially Varying Relationships. Chichester: John
Wiley & Sons.
Galli, G. and A. Prandi. 1970. Patterns of Political Participation in Italy. New
Haven, CT: Yale University Press.
Mainardi, R. 1998. Il Nord e la Padania. L’Italia delle regioni. Milano: Bruno
Mondadori.
O’Sullivan, D. and D. Unwin. 2003. Geographic Information Analysis. Hoboken,
NJ: John Wiley & Sons.
Politics. Philadelphia, PA: Temple University Press.
Chapter 5
Elections and Cultural Political Economy:
The Political Geography of the Bloque
Nacionalista Galego in the Galicia
Autonomous Community
Erinn P. Nicley
In the autumn of 1993, Spanish media reporting cast a spotlight on the rise of the
Galician Nationalist Bloc (BNG) in upcoming regional parliamentary elections
in the northwest province of Galicia. The BNG coalition and its charismatic
presidential candidate, Xosé Manuel Beiras, appeared to be gaining steadily
against other left-wing political forces in the traditionally conservative region.
After an intense electoral campaign, the ruling conservative People’s Party
(PP) regained an absolute majority control over parliament and the presidency.
The Socialist Party of Galicia (PSdeG) – the regional affiliate of the Spanish
Workers’ Socialist Party (PSOE) – retained its second-place opposition status
with nearly 24 percent of the vote and 19 parliamentary seats. And the left-wing,
regionalist BNG posted a third-place showing with 18 percent and 13 seats in the
75-member regional parliament (Parlamento de Galicia 2010).
The BNG finish as a third-place, regionalist political force in 1993, at first
glance, is an unusual foundation for a book chapter on new perspectives on
electoral geography. The 1993 Galician elections had not fundamentally changed
the electoral landscape. True, the BNG had more than doubled both voter support
and parliamentary seats won during the 1989 parliamentary election and had greatly
exceeded its 1982 electoral maiden voyage, in which the party received only 4
percent and one parliamentary seat. However, the overwhelmingly conservative
Galician political climate continued to prevail despite the internecine political
turmoil transforming progressive political forces in the race for second place.
The importance of the 1993 Galician regional elections is found not in terms of
control over the Galician political administration, but rather for what the patterns
of BNG support can tell about the mutual relationship between the political
economy of European structural integration, regionalism, and the place-specific
struggles that constitute the broader political and economic milieu. The rise and
decline of the BNG during a series of regional parliamentary elections from 1985–
2009 reflects a regionalist party and a broader nationalist movement caught in the
maelstrom of political and economic forces struggling to redefine and reorganize

the geography of political power and cultural identity across multiple scales.
This chapter approaches the political geography of elections through a
cultural political economy perspective (CPE). This CPE approach combines
political economy, cultural studies, and semiotics to reposition electoral studies
as a window onto the broader connection between macro-scale forces and the
contingent, culturally specific organization of political and economic relations.
A CPE perspective allows us to approach elections and electoral processes as
a political moment in which the grounded electoral struggles that occur within
particular places simultaneously are influenced by and play a part in reconstructing
from within the socio-spatial structure that conditions political, economic, and
social life.
In this vein, I analyze the political geography of the rise and decline of the
BNG coalition during the 1985–2005 Galician regional parliamentary elections.
For brevity, I focus on BNG electoral success during the 1993 and 1997 elections
to provide a unique, if imperfect, window into the place-specific political struggles
that have surrounded Galician and Spanish integration into the European political-
economic structure. The geographic and historical patterns of BNG electoral
support spotlight the political-economic conditions that created the political
space for the BNG political platform to attract renewed support among Galician
electorate. The political geography of BNG support demonstrates that BNG
electoral policy has been very responsive to the different political-economic
position of particular Galician localities. At the same time, the BNG has been able
to discursively incorporate existing social and cultural conditions within distinct
localities to construct regionalist narratives that attempt to reposition the BNG and
the broader Galician territory as a political force within the broader European and
Spanish political economy.
A Political Geography of Elections
This chapter begins with a question: What would or could a political geography
of elections look like? I purposefully adopt the language of ‘political geography’
to emphasize the embeddedness of elections within a more expansive view of the
political. A political geography of elections begins with the simple recognition
that the object of study extends both before and after actual elections. As the
chapters in this volume attest, such a political geography of elections affords
an opportunity to apply critical and post-structural perspectives towards many
existing electoral geography themes. For example, geographers are well-placed
to discuss the theoretical foundations and development of formal and informal
electoral institutions, political party organization and operations, campaign
finance, electoral participation (and non-participation), media election coverage,
the gendered, racialized, and sexualized geographies of elections, and post-
structural perspectives that see elections as embodied political performances
Elections and Cultural Political Economy 77
that produce particular societal discourses and effects. The shifted focus onto
critical and post-structural election studies likewise suggests that greater use of
qualitative methods may help uncover new understandings of electoral processes
as grounded, contingent, and embodied processes deeply intertwined with the
broader societal milieu.
In this chapter, I apply a cultural political economy perspective towards the
political geography of elections. Cultural political economy has emerged in recent
years as a promising theoretical approach that can bridge the divide between
critical political economy and the post-structural concern with identity politics,
discourse, and political performance (Jessop and Oosterlynck 2008; Jessop and
Sum 2001). To begin, I want to delimit my use of the term ‘political economy’.
I specifically refer to the processes of capitalist regulation that develop and
reproduce a particular political and economic regime. Regulationist thought in
recent years has developed a concept of societal regulation to capture the diverse
array of economic and extra-economic institutions and relations through which
capitalist political economy is reproduced (Jessop and Sum 2007). This has
opened new avenues to seeing capitalist regulation of the political economy in
unexpected places seemingly far removed from the macro-economic worlds of
financial regulation, market institutions, and economic class relations. For this
study, we can define capitalist regulation as a disciplinary regime that blends
together both economic and extra-economic social institutions and relations to
normalize the capitalist political and economic structure and thus secure continued
capital accumulation (Jessop and Sum 2007).
A cultural political economy perspective examines the culturally mediated
institutions and relations through which the political economy of capitalist
regulation is produced on the ground (Jessop and Sum 2007; Sayer 2001). Cultural
political economy offers a compromise between the monolithic class-centered
vision of structural political economy and the post-structural vision of multiple,
situated experiences of capitalism fractured by race, gender, and other social
identities (Gibson-Graham 1996). That is, cultural political economy stands apart
by seeing capitalist regulation as taking shape through the contingent intersections
of economic and identity politics that constitute everyday experience in particular
places. The political and economic structure in which localities are embedded is
itself constructed through the culturally distinct political relations that mediate
broader social forces within particular places. The contingent forms of capitalist
regulation that emerge through such grounded political relations establish distinct
geographic patterns of political and economic power across different scales and
among particular places in relatively stable but impermanent configurations
(Jessop and Sum 2007).
From this perspective, the political-economic structure is socially produced
through the culturally distinct political institutions and relationships that build from
within hegemonic patterns of capitalist regulation in particular places. However,
cultural political economy places limits on the range of culturally mediated
political and economic discourses that may form within the broader capitalist
context. The cultural political economy takes root through an evolutionary process
that privileges certain culturally distinct political and economic possibilities –
what Jessop and Oosterlynck (2008) call economic imaginaries – that also tend
to best support capitalist accumulation. Thus, cultural political economy involves
a reflexive process of variation, selection, and retention in which competing
economic discourses circulate within capitalist society (Jessop and Sum 2007).
The selection of certain economic imaginaries privileges a culturally distinct
set of narratives, semiotic meanings, and material practices that gain traction in
particular places to contingently reproduce the broader conditions of capitalist
reproduction. The socio-spatial structure of the political economy arises through
repeated instances of definition, negotiation, reproduction, and transformation
as it both responds to but also reproduces the culturally mediated conditions of
capitalist regulation and political economy (Jessop and Oosterlynck 2008).
This cursory overview of cultural political economy reflects the abstract state
of current theory on the actual operation of cultural political economy at the
intersection of semiosis, discourse, and critical political economy. Geographer
Martin Jones has called recently for greater attention to the embodiment of cultural
political economy as a political performance (Jones 2008). The cultural political
economy is reproduced on a daily basis through particular institutions, processes,
and political actors. In this sense, we can grasp the continued reproduction of
cultural political-economic conditions best through research focused on specific
political institutions, events, narratives, and relations that play a role in maintaining
(or transforming) the dominant conditions of capitalism in particular places. As
Jones (2008, 396) notes, the ‘centrality of the meaning and spatialities of politics
itself [becomes] the primary focus of conflict and debate’. In effect, the focus on
real-world political moments – like democratic elections – becomes a window into
the economic and extra-economic reproduction of the cultural political economy
of capitalist regulation in everyday life.
Elections are a distinct political moment within a society. The electoral process
reflects the embodied performances and struggles among political and economic
forces to define and put into practice a political and economic program. From this
perspective, elections extend beyond the limited experience of political parties
on election-day to include multiple social forces that take part in and influence
the electoral process, e.g., political parties, social movements, constituent
communities, media producers, financial interests, legal institutions. This extended
view suggests that elections in many ways reflect the prevailing political-economic
conditions within a society. Of course, political movements frequently reflect the
political economy through their internal composition, their programs, and their
constituent bases of support.
Political geographers have highlighted from different perspectives the
geographic connections between political economy, place, and electoral politics
(Agnew 1997, 2000; Clayton 2002; Flint 1996, 2001; Keating 2001; Marden
1997; Shin 2001). For such writers, a regional and place-based political economy
perspective has helped explain the geographic variation among electoral processes
and electoral outcomes across a range of political, regionalist, and ethnic nationalist
movements worldwide. The role of political parties and institutions in the social
construction of political spaces has been central to many of these arguments. For
example, Flint (2001) draws upon regional and place-based political-economic
conditions as the context upon which the Nazi party strategically tailored electoral
narratives and discourses to generate regional appeal in pre-war Germany. For
such studies, the political economy conditions the role of political discourse in
creating the distinct political geographies of elections. At the same time, there is
a recognition that political economy is not only a shaping influence, but also is
actively being produced through place-specific political struggles to define, put
into practice, and institutionalize particular institutions and relations (i.e., political
economy). Elections afford political parties an important, albeit de-centered,
opportunity to reconstruct the geographic ‘horizons’ of political and economic
discourse within the bounded concept of particular regions, places, scales, and
networks (Agnew 1997).
A cultural political economy perspective advances the political geography
focus on the construction of electoral political space with exploration of how
elections can serve to reproduce culturally mediated modes of capitalist regulation
in particular places. The electoral process in part is a political consequence of
structural political and economic forces that shape the field of political issues,
political movements, and voter mobilizations in a particular time and place.
However, we can use a cultural political economy perspective to unearth
the construction of political-economic conditions through the distinct social
institutions, relations, and narratives that constitute political and economic life on
the ground (MacLeod 2001). It is these everyday geographies that both mediate
and actively produce the political and economic conditions that both shape and
emerge through electoral outcomes.
Elections can be seen as an opening for renegotiation of the dominant cultural
political economy. Elections are a moment in which the dominant political-
economic ‘imaginaries’ are actively open to debate among political forces with
alternative visions for the organization of political-economic space. In other words,
elections are a snapshot of the culturally mediated political-economic fault lines
within a society. Yet, elections also permit the reconstruction of a new cultural
political economy through the selection of new political and economic narratives
and practices that ground capitalism within particular places and regions. Political
parties and other electoral forces selectively draw upon existing identities and
cultural themes as the foundation for rebuilding the political-economic position
of the society within the broader capitalist structure. In this way, elections are a
performance of the cultural political economy in which competing forces struggle
to arrange, reproduce, and transform the social and spatial patterns of political
and economic life through the culturally distinct field of relations that comprises
everyday life.
These tentative thoughts on a cultural political economy perspective on the
political geography of elections return us to the initial question: What can a
political geography of elections look like? The possibilities that I am beginning to

articulate in this chapter highlight that elections are at once micro-scale political
events and bound recursively into broader political and economic forces. The call
to unite the macro- and micro-worlds of electoral geography is well-worn territory
(e.g., Flint 1996). The cultural political economy perspective takes us further
with a focus on the place-specific identity politics that mediate, give meaning and
material form to capitalist political economy in time and space. Elections are a
key moment in the contingent reproduction of these culturally mediated political-
economic visions that define the position of particular places within the broader
capitalist structure.
We can consider a cultural political economy perspective to develop a roadmap
for studying the connections between capitalist regulation and the political
geography of elections. This roadmap consists of three parts:
1. examination of the political-economic context shaping the geography of

electoral outcomes;
2. examination of political campaign narratives and how political forces
select and discursively draw upon particular cultural identities and social
relations to promote particular political-economic imaginaries through the
electoral process; and
3. examination of how elections and electoral outcomes renegotiate the
cultural political economy of capitalist regulation through place and scale-
specific political processes that feed into the reconstruction of the capitalist
structure ‘from within’.
In the remaining sections of the chapter, I explore the connections between the
cultural political economy of capitalist regulation and the political geography of
regional elections in the Galician autonomous community in northwest Spain. The
roadmap described above structures the conversation to analyze the geography
of BNG political support in terms of political-economic changes that came with
Spanish adhesion to the European Union, the political response that the BNG
offered to the social dislocations experienced in particular Galician localities, and
the transformation of the Galician regional political-economic position within the
European regional structure through the political presence of the BNG and its
regionalist political discourse.
The Political Geography of Regional Elections in Galicia
The rise of the Galician Nationalist Bloc (BNG) as a central political force in
Galician political life in the past 30 years emphasizes the powerful influence
of place-based political economy upon the political geography of elections.
The historical regional context of Galician political, economic and cultural
marginalization has influenced Galician politics for over a century. The varied
Galician regionalist and nationalist movements from the 1840s to the present
have conflated Galician cultural identity with traditional genres de vie rooted
in the maritime fishing, animal husbandry, and agricultural sectors (Sharif
2006). In this sense, the BNG story is also the story of Spanish adhesion to the
European Union and its predecessor organizations during the 1980s and 1990s.
The renegotiation of the Galician political economy through the turbulent social
processes of European–Spanish integration became a focal point for Galician
regionalist politics discursively committed to defending Galician cultural and
political-economic heritage. From EU–Spanish adhesion in 1986 to the political
climax and dénouement in the early 1990s, the political geography of successive
regional parliamentary elections reflected the blended economic conditions and
identity politics that occupied the hearts and minds of the Galician population.
The Bloque Nacionalista Galego was a marginal regionalist political force in
the early years of post-Franco Spain (1978–present). Galician regionalist political
forces proliferated among both the political right and a deeply fragmented political
left eager to reclaim political space lost during the Franco dictatorship. The BNG
was no exception when it formed and first participated in the 1982 regional
election. The diverse positions on the question of independence vs. varied levels
of self-determination and autonomy as provided in the 1978 Spanish constitution
limited the success of left-leaning regionalist political forces at the polls (Botella
1989). Indeed, Galician regionalist electoral politics in the early 1980s suggested
that the center-right Galician Coalition (Coalición Galega, CG) was going to set
the Galician regionalist agenda from its electoral stronghold in the rural, interior
Ourense and Lugo provinces of eastern Galicia. The meritable 13 percent support
for the Galician Coalition in the 1985 regional parliamentary elections stands in
contrast to the BNG, who gained only 4 percent support from radicalized centers
on the Galician coast (Vilas Nogueira 1992).
A brief four years later, the BNG rose in the 1989 regional elections on a
political trajectory that ultimately positioned the BNG as the hegemonic regionalist
force on the Galician political landscape for the next 20 years. In subsequent
regional elections, the BNG continued to demonstrate its capability to attract votes
as a major opposition party: 8 percent (5 seats) in 1989, 18.4 percent (13 seats)
in 1993, and 24.8 percent (18 seats) in 1997. By the 2001 regional election, the
BNG surpassed the socialist PSdeG-PSOE as the main opposition party with 22.6
percent (17 seats) voter support. Mixed BNG success in 2005 greatly reduced voter
support to 18.7 percent (13 seats), but gave the BNG the Galician vice presidency
in a coalition government with the socialists. The BNG suffered a stark decline in
2009 regional elections, receiving only 16 percent support (12 seats), compared
to 25 seats won a resurgent socialist PSdeG-PSOE (all data from Parlamento de
Galicia 2010).
The BNG success through the 1990s and early 2000s and its uncertain future
today spotlight the question of why and how a far-left regionalist political
force was able to gain so much political ground in a territory renowned for its
traditionalist, conservative politics (Vilas Nogueira 1992). Scholars frequently
point to the proximate causes of BNG growth during these years. The incorporation
of moderate regionalist political parties into the BNG coalition, the decline of the
state-range PSdeG-PSOE, and the mobilization of new and idle voters have all
played their part in BNG success (Barreiro Rivas 2003; Márquez Cruz 2004;
Portas 2010; Vilas Nogueira 1992). However, closer examination of the political
geography of BNG electoral support during the critical 1993 and 1997 regional
parliamentary elections suggests that the Galician electorate increasingly turned
to the BNG as a place-specific political response to the European political-
economic integration and the appeal of the BNG as self-proclaimed protector
of Galician culture and position with the emerging European system (Casal and
Ventoso 1994).
In this section, I analyze the political geography of BNG electoral support in
a series of regional parliamentary elections from 1985–2005. Exploratory data
analysis of BNG percent vote received in each election using ArcMap 9.2 Gi*
hotspot analysis highlights a complex political geography punctuated by multiple
spatial clusters of significantly high and low BNG support. The BNG electoral
map series reveals spatial and longitudinal patterns of changing BNG support
among 315 Galician municipal townships that constitute the principal electoral
candidate districts throughout the region. Following an initial review of the entire
1985–2005 BNG electoral series, I focus on the 1993 and 1997 regional elections
to examine the relationship between BNG electoral clusters and the reconstruction
of the Galician cultural political economy through the electoral process.
The generalized BNG growth from the 1985–2005 regional elections
demonstrates quite graphically the consolidation of regionalist political forces
and the ideological shift towards a more moderate, social-democratic discourse
from 1987 onwards (Barreiro Rivas 2003). From 1987, the BNG formalized a
new strategy of moderation of its radical discourse, a process that began with the
acceptance of the Spanish constitution in 1985. In place of the historic rejection
of the Spanish state, a new BNG discourse emphasized maximization of Galician
autonomy with social-progressive themes and instrumental concerns involving
Galician political, economic, and cultural affairs (BNG 2010). By 1989, the
BNG reoriented its program towards protection of Galician political, economic,
and cultural interests within the complex political economy of the Spanish state
and European community. Throughout the early 1990s, the BNG increasingly
attempted to attract moderate voters in the traditional socialist-leaning interior
townships of A Coruna province, Pontevedra province and, with notable success,
the former Galician Coalition strongholds in the Ourense province of southeastern
Galicia. The early 1990s Galician political focus on economic recession, debate
regarding approval of the Maastricht Treaty on European Union, and the growing
debilitation of the Spanish socialist (PSOE) government both nationally and within
Galicia were central to BNG success (Casal and Ventoso 1994).
The 1993 and 1997 regional elections maps highlight three visible and
statistical trends in the data (Figure 5.1 and 5.2). First, the BNG from 1985–2005
has demonstrated a general expansion of political appeal throughout Galician
Figure 5.1 BNG Election Support Map, 1993 (%)
Figure 5.2 BNG Election Support Map, 1997 (%)
territory. The growth of BNG support clearly remains strongest among the
western A Coruna and Pontevedra provinces, but BNG expansion into the rural,
conservative Lugo and Ourense provinces to the east is notable. Second, the BNG
simultaneously developed deeper political support among its core constituent
townships and cities along the heavily populated Galician west coast. Third, the
1993 and 1997 regional elections most distinctly highlight the consolidation of
three distinct BNG support clusters centered on:
a. northwest A Coruna province;

b. southwest Pontevedra province; and
c. southwest Ourense province.
The BNG capitalized upon the opportunity to reposition its discourse as the sole
opposition to the ruling Partido Popular and positioning itself as the defender of
Galician economic development within the European Union. The BNG political
program during the 1990s reflected a moderate, inclusive political project that
reflected the dynamic transformations of Galician political economy during
the period (see next section). Campaign slogans motivated Galician voters
with narratives of ‘Galicia without limits’ (Galiza sen limites), ‘Open Yourself
to Galicia. Participate in a Common Project’ (Abre-te a Galiza. Participa nun
proxecto común) and ‘The Galicia that you want. Together we can’ (A Galiza que
queres. Entre todos podemos) (BNG 2010).
The BNG political program yielded results as BNG support advanced from
its traditional western enclaves into central and eastern Galicia in the 1993 and
1997 regional elections. This growth can be seen as a response to the concurrent
development and intensification of European-led political and economic
restructuring and the failure of Spanish-range political forces (and their regional
allies) to adequately address growing Galician concerns over the effect of these
processes on the daily lives of key constituents in the Galician community. BNG
support quickly extended from its base among industrial workers, agricultural and
maritime producers in western Galicia’s mid-sized townships (10,000–30,000
population) and industrial enclaves like O Ferrol towards the larger urban districts
of western Galicia with their base of culturally conscious liberal professional
classes and newly mobilized youth voters concerned with the deterioration of
Galician identity and economic opportunity both along the urban coast and the
interior rural districts (Aizpeolea 1997; Bauer 1992; Rivas 1997).
Localized clusters of strong BNG support further punctuate the regional BNG
growth across Galicia. As Figures 5.1 and 5.2 indicate, geographic patterns of
BNG support among what I call the A Coruna cluster, the Pontevedra cluster,
and the Ourense cluster suggest that BNG political strength in the 1990s
reflects particular place-specific contextual factors beyond mere demographic
variability among the Galician electorate. In a previous, unpublished study, I
have demonstrated that BNG support among these three geographic clusters are
in part a response to place-specific economic structures and underlying political-
economic transformations within each cluster (Nicley 2002). This structural-
ecological perspective suggested that BNG support had concentrated among
townships where the European integration process had introduced a series of
structural and sociological changes effecting traditional Galician economic
sectors and production complexes. Structural changes to primary sector maritime
production centered in the Pontevedra cluster, traditional shipbuilding and steel
production in the A Coruna cluster, and a mix of dairy production, timber and
lumber production, and environmental pollution concerns in the Ourense cluster
correlated statistically with the place-specific geographies of higher-than-normal

BNG support in each region (Nicley 2002).
These BNG spatial patterns corresponding to the BNG peak success in the
1993/1997 regional elections compel a closer examination of European integration
and Spanish state policy as factors driving BNG growth and stabilization during
the past 20 years. In the following two sections, I explore the Galician economic
structure and the grounded experiences of European political-economic reform
that accompanied the Spanish and Galician adhesion to European Community/
European Union policies, regulations, and institutions. The chapter concludes
with consideration of how the transformation of the Galician economic structure
during the 1990s established a political opportunity for the BNG to reconstruct
the cultural political economy of EU–Spanish–Galician relations through the
elections process.
European Integration of the Galician Political Economy (1985–2005)
To make sense of the Galician Nationalist Bloc growth during the 1990s, we
first must understand the transformations to the Galician economic structure
experienced during the late 1980s–1990s. Territorially, Galicia is a minor part of
the Spanish state – slightly smaller in area than the US state of Maryland and
with a stable, modest population of 2.8 million inhabitants. The Galician position
within the Spanish state since the mid-nineteenth century reflects the longstanding
tension between structural underdevelopment and limited Spanish state programs
to promote economic diversification (Beiras 1994). Despite considerable
development of both secondary industry and tertiary services in the past three
decades, traditional primary sector economic activities (i.e., fishing, dairy,
agriculture, forestry) remain stalwarts of the Galician economy and social structure
(Figure 5.3). Galician political economy characteristically involves a simultaneous
mixture of subsistence and market-based production. Occupational identities are
blurred by the carrying-out of multiple primary, secondary, and tertiary sector
economic activities as part of daily life among Galician households (Baylina
and García-Ramón 1998; Roseman 1999). Traditional male economic activities,
such as industrial labor, construction, forestry and fishing, are frequently coupled
with traditional female economic activities, including subsistence agriculture and
dairy production, but also with clam-digging, seafood processing, domestic labor,
hospitality and hostelry, and the putting-out system of in-house textile production
(Meltzoff 1995).
Each occupational activity contributes to the overall production and re-
production of the Galician political economy in a closely interwoven pattern.
As a result, structural economic changes have a powerful effect on the political
economy of daily life and foster support for regionalist responses to these
external forces shaping the local social condition. The secondary basis of much
Galician agriculture as a supplemental activity and the traditionally small-scale,
Figure 5.3 Traditional Primary Sector Economic Activities in Galicia
personalized social relations characteristic of the fishing industry are reflected by

the low level of primary-sector unemployment, which only equals five percent of
Galician unemployment (Meltzoff and LiPuma 1986).
The intensification of European Union economic regulation in the 1990s began
to generate structural effects among the longstanding traditional Galician modes
of primary and secondary sector economic production. These traditional bases of
the Galician economy were subjected to rapid transformations that undermined
culturally distinct production practices and relations that had been in force in
some instances for centuries (García-Ramón et al. 1993). Throughout the 1990s,
the rapid transformation of the predominant structure of primary and secondary
sector production radically disrupted occupational and household relations among
Galician maritime, agro-industrial, and peasant communities as the Galician
economy transitioned into a modestly diversified, services-dominated economy.
For perspective, from 1970 to 1996 primary sector employment declined from
56.5 percent of the active labor force to only 15.6 percent. The corresponding
primary sector contribution to regional gross domestic product declined from 28
percent to only 8.5 percent during the same period (Beiras 1994; IGE 1996). In
contrast, tertiary services rapidly doubled throughout the 1980s and 1990s into
approximately 50 percent of all employment and almost one-third of the regional
GDP (Beiras 1972; IGE 1996).
Perhaps because of these transformations, defense of the traditional Galician

primary sector and the spin-off industrial sector became a salient political issue
during the 1990s regional elections in disproportion to their size. The primary
fishing sector is at the forefront of this Galician political economy. Even by the mid-
1990s, the fishing sector generated roughly 57 percent of all Spanish production
and directly employed 37,000 workers (IGAPE 1994). Similarly, ancillary and
secondary sector maritime industry – shellfishing, aquaculture, seafood canning
and packaging, shipbuilding, and such – contributed closer to 80,000 jobs to the
Galician and particularly the Pontevedra provincial economy (Ballesteros 2000;
Fernández-Macho, Gallastegui, and González 2008). Milk and dairy production,
forestry and wood transformation, shipbuilding and steelworks, and a range of
mining, textile, and energy production industries constituted the principal primary
and secondary mainstays of the Galician economy (IGAPE 1994). Export growth
among leading industrial sectors such as transportation manufacturing, granite and
slate mining, and steel and metals manufacturing did not counter the negative
trade balance of imports over exports in the core agro-industrial sectors such as
fishing, energy production, forestry and wood products, and commercial dairy
production (IGAPE 1994).
Reconstructing a Galician Space Within the European Union
The EU Maastricht Treaty implementation in 1993 provided Galician political

society – with some justification – a focal point for the social unease that permeated
everyday life.
The deepening structural and institutional connection between Europe, Spain,
and Galicia required the BNG to articulate a political program that would define
the territorial organization of political and economic power for the benefit of the
Galician economy and culture. The resulting BNG program selectively utilized
Galician cultural identity politics to both reflect and feed into growing perceptions
of political and economic disempowerment with a regional geographic vision
for defending nominally Galician interests. Thus, the BNG political movement
came to symbolize and cultivate a core trope of Galician cultural identity bound
to the primary and secondary sector employment and a corresponding traditional
political economy centered on the peasant-based cultural complex of primary
and secondary sector production. The BNG defense of Galician agriculture,
dairy, maritime production, and heavy industry prepared the way for the BNG
to dominate Galician regionalist representation among regional elections politics
throughout the 1990s and into the 2000s.
Accordingly, BNG elections success must take into account the changes to
traditional Galician production centered on complex community- and family-
based networks of agricultural and dairy production, coastal fishing and shellfish
extraction practices, and land tenure regimes little changed from earlier generations
(Bauer 1992; García-Ramón et al. 1993; Naylon 1995; Roseman 1996). The late
1980s–early 1990s Galician experience of global political and economic forces

reinforced a political-economic ‘siege mentality’ that opened the door for the
consolidation of the BNG as the lone viable heir to Galician regionalist politics.
The deindustrialization that eroded the metallurgical and shipbuilding sectors of
western Galicia in the late 1970s and 1980s had set the stage for the socio-political
transformations to the primary and secondary sectors that accompanied the
deepening Galician incorporation into the European-centered political economy
during the 1990s period.
Regional politics surrounding the 1993/1997 regional elections solidified
around the structural and social importance of Galician primary sector and
secondary sector production under threat from European regulation and structural
reforms. Major EC/EU policies like the Common Fisheries Policy (CFP) and
Common Agricultural Policy (CAP) became discursive rallying points for the
BNG political critique and calls to protect the traditional Galician primary and
secondary sector production complexes that constituted daily life for many
Galician citizens (Meltzoff and LiPuma 1986). European and Spanish political
forces represented an externalized threat ‘from beyond’ to traditional Galician
identities rooted in economic class, gender, and place-based identity.
The effects of European Union economic regulation on particularly the
1993/1997 Galician regional elections are visible through review of the
European Common Fisheries Policy (CFP) and the territorial reorganization
of fisheries governance institutions and practices. From its formal European
Community adoption in 1983, the CFP was designed to coordinate biologically
and economically sustainable maritime stock management conservation practices
through policy instruments like the Total Allowable Catch (TAC) quotas and the
5-year Multi-Annual Guidance Plans (MAGP) focused on maritime fleet reductions
and reorganizations (Caballero Miguez et al. 2007). The governance institutions
put in place by the CFP generated a complex territorial reorganization of political
authority among the EU, Spain, and the Galician government. EU regulatory
control over so-called Community waters reinforced the increasingly hierarchical
European political-economic framework that removed policy decision-making
further away from Galician maritime producers onto higher levels of authority.
Notably, the CFP-generated territorial reorganization of fisheries governance ceded
Galician coastal waters maritime regulation to the Galician regional government,
the Xunta, controlled throughout the 1990s by the conservative People’s Party
(Caballero Miguez et al. 2007).
The complex regulatory geographies that the CFP produced among Galician
maritime producers extended throughout the Galician productive complex to
transform the social foundations of Galician society centered on small-scale
maritime, agricultural and dairy producers and secondary sector maritime
processing and shipbuilding industries. Stark reductions in Total Allowable
Catch quotas (administered by Spanish officials) and from 1992 mandatory
fleet modernization and overcapacity reduction measures had notable direct and
indirect effects on the daily lives of Galician peasant producers. As Fernandez-
Macho, Gallastegui, and González (2008) recognize, CFP fisheries policies

extended beyond both conservation and economic rationales to have direct effects
on the political and social conditions of entire communities and regions effected
by European policy decisions.
The political ramifications of EU integration and regulation on the Galician
traditional primary and secondary sectors gave sustenance to the BNG political
program and drove public support for the BNG in communities and townships
hardest hit by EU regulations and perceived governance failures among both
Spanish and Galician governments (Nicley 2002). While it is notable that
maritime sector employment dipped during the early and mid-1990s, the sector
rebounded to a strong economic position by the late 1990s along with substantial
European and Spanish financial support for Galician fishing fleet modernization
and aquaculture production (Caballero Miguez, Garza Gil, and Varela Lafuente
2007). Nevertheless, the European integration process transformed traditional
genres de vie that were at the heart of Galician cultural identity. The BNG
rise in the 1990s can be seen as a reflection of these societal dislocations that
reconstructed the Galician political economy towards closer integration into the
European-dominated capitalist political economy. As anthropologist Heidi Kelley
(1994, 77) noted:
Issues of regional identity will continue to be highlighted in coming years, as

Galicia struggles to define its role within both the Spanish state and the European
Community. As the economy and prestige structure of the community continue to
change, this tension [between pride in rural identity and urban cosmopolitanism]
will continue to be exacerbated, particularly if regional identity continues to be a
locus of resistance to what is perceived as nationally-imposed change.
The complex restructuring of the Galician maritime sector was paralleled in

European-led regulation like the Common Agricultural Policy that equally
disrupted traditional modes of production and political-economic relations
among Galician dairy cattle producers, wood products producers, avicultural
and agricultural producers (García-Ramón et al. 1993). For example, increased
European market competition during the early 1990s further burdened Galician
small-scale dairy producers attempting to comply with Common Agricultural
Policy dairy production quotas and overproduction fines levied by the EU against
member states and passed down to regional governments like the Galician Xunta
(El Correo Gallego 1999, 2001).
Taken together, European – Spanish adhesion from 1986 (and particularly
the period surrounding the 1993 formation of the European Union) introduced
institutional and regulatory reforms that reorganized the territorial governance
relations and underlying political economy of Galician society. The 1993 and 1997
regional elections thus reflect the transformed conditions of everyday practice and
cultural identity among particular Galician producers and communities.
The political geography of BNG regional elections support can be viewed as a

response to the place-specific political and economic dislocations generated by the
reterritorialization of political power onto the European Union and concomitant
disruption of traditional practices, cultural identities, and social relations in
particular Galician communities. Notably, the minifundia farm communities
hardest hit by European maritime, agricultural, and dairy sector reforms in
the early 1990s – ‘particularly … on the coast, where the average farm size is
significantly smaller than inland’ – also show some of the highest levels of BNG
support not only in the 1993 and 1997 regional elections, but throughout the 1985–
2005 regional election series (García-Ramón et al. 1993, 8). As anthropologists
Edward LiPuma and Sarah Meltzoff (1994, 32) note:
The problem, and cause for some political discontent, is that the unification
(through formalization and standardization) of the institutions of a particular
[economic] sector is seen as leading to an internationalization of culture and
language and thus a dilution of national and occupational identities … There is
a fundamental link between occupation and Galician identity and … these are
being compromised by EC proposals for restructuring primary industries.
Perhaps foretelling the BNG success later that year, the Spanish daily newspaper
El País in 1993 published public opinion poll results showing that 49 percent
of fishermen believed that the state of the fishing industry had worsened and 54
percent of farmers believed that the state of the dairy industry had worsened,
compared to only 7 percent of fishermen and 15 percent of dairy producers who
believed their respective industries had improved (Ibañez and Hermida 1993).
Reconstructing the Galician ‘Cultural Political Economy’: Variation, Selection,

Retention
Detailed analysis of the connections between European political economy, Galician

regionalist politics, and regional elections returns us to the central question of
this chapter: How does the political geography of elections play a role in the
reconstruction of cultural political economy? The European project has involved
a renegotiation of the political economy among Europe, Spain, Galicia, and
particular localities within the Galicia region. European integration policies came
together in the historical political moment of the 1990s and in particular places
to profoundly change the Galician political economy. The geographic patterns
of BNG support reflect the changing Galician landscape of identity politics in
the 1990s as communities responded to their everyday experience of capitalist
restructuring towards a globalized, European-centered political economy.
However, a cultural political economy perspective on elections takes us
beyond the top-down model of a European macro political economy generating
Galician regional and place-specific electoral effects. The recursive connection
between European integration and BNG support requires that we look more
closely at the role of elections in actually reconstructing the European (cultural)
political economy from within. The European project established an opportunity
for the rise of new political parties responding to transformed political and
economic conditions. Yet, the BNG rise was never a priori guaranteed in response
to the Galician political-economic reforms in the 1990s. The regional elections
provided an historical moment for renegotiation of the Galician political economy
through competition to define, select, and institutionalize a particular, alternative
political-economic vision for Galicia. That is, elections are constitutive as an
embodied set of practices that reconstruct the cultural political economy within
the geographically specific field of identity politics found in particular places.
BNG success in the Galician regional elections over the past 20 years (as
well as European, Spanish, and municipal elections) can be viewed as a cultural
political economy selection process from among the broader range of Galician
political forces and parties. The BNG emerged as the hegemonic voice of Galician
regionalist politics throughout the 1990s and our present century through a
political ‘selection’ process in which the alternate political-economic visions (that
is, ‘imaginaries’) offered by regionalist and regionalist forces like the Coalición
Galega fell aside or were incorporated into existing political discourses. The
BNG participation in the Galician electoral process has thus been a platform
through which its distinct visions, rhetoric and discursive practices have helped to
reconstruct the Galician political economy.
This BNG vision for cultural political-economic reconstruction has centered on
the renegotiation of the geographic relationship between Europe, Spain, Galicia,
and particular localities within the region. The BNG participation in regional
elections – from campaigns, to political demonstrations, to electoral outcomes
– has generated greater political space for a Galician regional presence within
the overarching framework of the European Union and the Spanish state. The
BNG political move from the late 1980s towards a more conciliatory discourse of
Galician ‘self-determination’ within the Spanish state structure and accompanying
visions of a strong Galicia within the European Union supports the regionalists’
effort to advance the Galician structural position within a European-dominated
political economy. The 1994 BNG political rhetoric of ‘a strong Galicia in a new
Europe’ (‘Unha Galiza forte nunha nova Europa’) during EU parliamentary
elections reflects the BNG effort to redefine the hierarchical European-Spanish-
Galician political economy towards a greater regional voice within the institutional
fabric of the European Union (BNG 2010).
Moreover, the Galician regional elections constitute a distinct political
moment in the contingent reproduction of the Galician cultural political
economy. The regional elections (and particularly the 1993 and 1997 elections)
produced complex political-economic visions for Galicia through the contingent
articulations of political power among the BNG, mainstream socialist and
conservative political forces, and other minor political groups. These contested
visions for a new Galician future involved distinct articulations of exactly what
elements of society fell within the realm of the political, the economic, and
the non-economic. The BNG defense of Galician primary sector and secondary
sector producers as cultural standard bearers threatened by the course of
European integration and Spanish economic policy found a ready audience
among constituents suffering from the economic and social dislocations of the
1990s. The BNG vision for Galicia located traditional production activities
and the everyday lives of Galician industrial labor and small-scale agricultural,
dairy cattle, and maritime professionals as simultaneously within the realm
of the economic and the social – and therefore a legitimate focus for political
renegotiation within the European Union framework. The transformed design
for the Galician position within Europe and Spain, as articulated notably
during regional elections, helped to define and establish the contingent lines
of political struggle over cultural and social identity that have become, if
fleetingly, the foundations for the production and embeddedness of the Galician
political economy within the broader capitalist structure of the European Union.
Finally, the BNG role in reconstructing the Galician cultural political economy
demonstrates the flexible modalities of capitalist regulation as mediated through
culturally contingent political relations. In their call for cultural political economy
perspectives, Jessop and Oosterlynck (2008) argue that political economy cannot
be reduced to the ‘soft sociology’ of socially constructed economic discourse
unhinged from a broader capitalist structural foundation. The Galician regional
elections in the 1990s demonstrate the hybrid conditions of cultural political
economy as socially constructed within the bounded possibilities afforded by a
culturally mediated capitalist structure. The Galician cultural political economy
envisioned by the BNG and BNG supporters (always in relation) centered on
reconstructing new economic imaginaries of the Galician position within the
European Union. The moderated BNG stance of critical rapprochement towards
the European project during the 1990s facilitated BNG electoral success and
political influence over European integration policy in ways that have continued
to yield political results in later years. For example, the European Union
inaugurated in 2007 the EU Community Fisheries Control Board headquarters
in the Galician industrial port city of Vigo. Such institutional webs that bind
Galicia within a common European regulatory framework reflect the contingent
political struggles of recent decades to raise the Galician profile politically,
economically, and socially within the overarching capitalist structure of the
European Union. In this manner, we can see regional elections as a ‘sorting out’
process in which competing electoral-political forces like the BNG, PSOE, and
PP reproduce through electoral processes the broad contours of the European
capitalist structure even as the institutional and governance dynamics among
Galician intra-community relations, Galician-Spanish relations, and Galician-
European Union relations remain in a constant state of evolution towards new
forms of culturally grounded political economy.
Conclusions
This chapter explored possible connections between the political geography of

elections and cultural political economy. The benefit of cultural political economy
is a basic reminder that the capitalist socio-economic structure is a social product
that takes shape through a geographically contingent mix of economic and identity
politics. In other words, capitalist political economy is always in a state of being
reproduced and that reproduction process opens the door for the transformation of
political-economic structure through political struggle. Following Jones (2008), I
have suggested that we can approach elections as a particular political moment in
the contingent reproduction of a culturally mediated capitalist political economy.
The Galician regional elections and the rise of the Galician Nationalist
Bloc (BNG) provided a strong case of capitalist political economy as a shaping
factor in electoral outcomes. At the same time, European political and economic
integration afforded an opportunity window for the negotiated reconstruction of
the European-centered political economy through the political struggles among
European, state-range, and regionalist political forces to redefine and consolidate
a distinct political-economic vision for the future. The Galician regional elections
highlighted the complex geography of these reconstruction processes. Geographic
concepts of power, place, and scale are central to the ongoing reconstruction of
European-centered cultural political economy, from the struggles over territorial
governance between the EU, Spain, and Galicia to the place-specific political
struggles over political, economic, and cultural identity that generated bases of
support for the BNG program to rebuild the Galician political-economic position
within a dominant European capitalist system.
Approaching the political geography of elections as a point of entry for
research on the production and changing characteristics of capitalist political
economy gives political geography an opportunity to bring together research on
macro-political and macro-economic forces and the contingent, grounded political
practices that go into reconstructing capitalist political economy on a daily basis.
The connections between elections and cultural political economy remain open
to significant refinement, clarification, and modification – this chapter offers
only an initial consideration of the possible avenues that political geographers
might explore. However, the potential benefits for political geography rest in the
opportunity to reinvigorate electoral studies on the leading edge of contemporary
social theory on the intersections between capitalist regulation and post-structural
thought.
References
Agnew, J. 1997. The dramaturgy of horizons: Geographical scale in the

‘Reconstruction of Italy’ by the new Italian political parties, 1992–95. Political
Geography 16 (2): 99–121.
Agnew, J. 2000. From the political economy of regions to a regional political

economy. Progress in Human Geography 24: 101–10.
Aizpeolea, L. 1997. El gobierno atribuye el alza del nacionalismo gallego a su
componente de izquierdas. In El País print edition. Published October 21, 1997.
Available at: http://www.elpais.com/articulo/espana/espana/galicia/espana/
bloque_nacionalista_galego_BNG/partido_popular_/PP/partido_socialista_
obrero_espanol_/PSOE/elecciones_autonomicas_gallegas_2001_/21-10-2001/
elpepiesp/19971021elpepinac_4/Tes [last accessed: October 27, 2010].
Ballesteros, M.A. 2000. Technology and employment, new ways of organising
production and labor. Modernisation of traditional economic industries:
The fishing in Galicia (Spain). The public administration as advocacy actor.
Paper presented at The Ionian Conference: Facing the Challenges of the New
Millennium, at Corfu, Greece.
Barreiro Rivas, X.L. 2003. Da UPG ao BNG: O proceso organizativo do
nacionalismo galego. In X.M. Rivera Otero (ed.), Os Partidos Politiicos en
Galicia. Vigo: Xerais de Galicia, 99–262.
Bauer, R.L. 1992. Changing representations of place, community, and character in
the Spanish Sierra del Caurel. American Ethnologist 19 (3): 571–88.
Baylina, M., and M.D. García-Ramón. 1998. Homeworking in rural Spain: A
gender approach. European Urban and Regional Studies 5 (1): 55–64.
Beiras, X. 1994. O Atraso Económico de Galicia. Second edition. Santiago de
Compostela: Galaxia.
BNG. 2010. A Historia do BNG [BNG Official Website]. Available at: http://www.
bng-aliza.org/opencms/opencms/BNG/global/contidos/textos/historia/textos/
texto_0001.html [last accessed: October 20, 2010].
Botella, J. 1989. The Spanish ‘new’ regions: Territorial and political pluralism.
International Political Science Review 10 (3): 263–71.
Caballero Miguez, G., M.D. Garza Gil, and M.M. Varela Lafuente. 2007.
Institutions, governance and fishing policy: Theoretical foundations and
the analysis of the Galician case. Paper read at European Association for
Evolutionary Political Economy (EAEPE), at Porto, Portugal.
Casal, C., and L. Ventoso. 1994. Un Solo de Gaita. Santiago de Compostela:
Biblioteca Gallega.
Clayton, T. 2002. Politics and nationalism in Scotland: A Clydeside case study of
identity construction. Political Geography 21: 813–43.
El Correo Gallego. 1999. Os criadores avícolas, contra as propostas de UE. El
Correo Gallego (Digital edition), 15 June 1999.
El Correo Gallego. 2001. Milleiros de gandeiros reclaman en Santiago axudas
directas para o sector. El Correo Gallego (Digital edition), April 8, 2001.
Fernández-Macho, J., C. Gallastegui, and P. González. 2008. Economic impacts
of TAC regulation: A supply-driven SAM approach. Fisheries Research 90
(1–3): 225–34.
Flint, C. 1996. Whither the individual, whither the context? Political Geography
15 (2): 147–51.
Flint, C. 2001. A timespace for electoral geography: Economic restructuring,

political agency and the rise of the Nazi Party. Political Geography 20: 301–29.
García-Ramón, M.D., M. Vilarino, M. Baylina, and G. Canoves. 1993. Farm
women, gender relations and household strategies on the coast of Galicia.
Geoforum 24 (1): 5–17.
Gibson-Graham, J. 1996. The End of Capitalism (As We Knew It): A Feminist
Critique of Political Economy. Cambridge, MA: Blackwell.
Ibañez, J.G., and X. Hermida. 1993. Una ligera mayoría cree que en Galicia se
vive mejor. El País, October 11, 1993, 17.
IGAPE. 1994. Galicia: Good Business Opportunities. Edited by I.G.d.P.
Económica. Santiago de Compostela: Xunta de Galicia.
IGE. 1996. Padrón de Habitantes de 1996: Instituto Galego de Estatística.
Jessop, B., and S. Oosterlynck. 2008. Cultural political economy: On making
the cultural turn without falling into soft economic sociology. Geoforum 39:
1155–69.
Jessop, B., and N.-L. Sum. 2001. Pre-disciplinary and post-disciplinary
perspectives in political economy. New Political Economy 6 (1): 89–101.
Jessop, B., and N.-L. Sum. 2007. Beyond the Regulation Approach: Putting
Capitalist Economies in Their Place. Cheltenham, UK and Northhampton,
MA: Edward Elgar.
Jones, M. 2008. Recovering a sense of political economy. Political Geography
27: 377–99.
Keating, M. 2001. Rethinking the region: Culture, institutions and economic
development in Catalonia and Galicia. European Urban and Regional Studies
8 (3): 217–34.
Kelley, H. 1994. The myth of matriarchy: Symbols of womanhood in Galician
regional identity. Anthropological Quarterly 67 (2): 71–80.
LiPuma, E., and S.K. Meltzoff. 1994. Economic mediation and the power of
associations: Toward a concept of encompassment. American Anthropologist
96 (1): 31–51.
MacLeod, G. 2001. New regionalism reconsidered: Globalization and the
remaking of political economic space. International Journal of Urban and
Regional Research 25 (4): 804–29.
Marden, P. 1997. Geographies of dissent: Globalization, identity and the nation.
Political Geography 16 (1): 37–64.
Márquez Cruz, G. 2004. Las elecciones locales y provinciales en Galicia
(1979–2004). 1. Comportamiento electoral y continuidad/renovación de las
autoridades locales. Revista de Investigaciones Políticas y Sociológicas 3 (2):
111–33.
Meltzoff, S.K. 1995. Marisquadoras of the shellfish revolution: The rise of
women in co-management on Illa de Arousa, Galicia. Journal of Political
Ecology 2: 20–38.
Meltzoff, S.K., and E. LiPuma. 1986. The troubled seas of Spanish fishermen:
Marine policy and the economy of change. American Ethnologist 13 (4): 681–
99.
Naylon, J. 1995. On the periphery: A view from rural Galicia. Leeds Papers on
Spain in Europe n.d.: 39–79.
Nicley, E. 2002. Somos Galegos!: Regionalism, political economy, and place
context in Galicia, Spain. Master’s Thesis, Department of Geography, Florida
State University, Tallahassee, FL, USA.
Parlamento de Galicia. 2010. Resultados electorales. Published by the
Parliament of Galicia. Available at: http://www.es.parlamentodegalicia.
es/sites/ParlamentoGalicia/ContenidoGal/pargal_II_ResultadosElectorais.
aspx [last accessed: October 10, 2010].
Portas, P. 2010. The myth of unity and the BNG: Storytelling, nationhood and
power. In Political Studies Association. Edinburgh: Unpublished manuscript.
Rivas, M. 1997. ¿Por qué sube Beiras? El País print edition. Published October
13, 1997. Available at: http://www.elpais.com/articulo/espana/espana/galicia/
espana/Bloque_Nacionalista_Galego_/BNG/elecciones_autonomicas_
Gallegas_1997_/19-10-1997/sube/Beiras/elpepiesp/19971013elpepinac_11/
Tes?print=1 [last accessed: October 29, 2010].
Roseman, S. 1996. ‘How we built the road’: The politics of memory in rural
Galicia. American Ethnologist 23 (4): 836–60.
Roseman, S. 1999. ¿Quen manda? (Who’s in charge?): Household authority
politics in rural Galicia. Anthropologica 41 (2): 117–32.
Sayer, A. 2001. For a critical cultural political economy. Antipode 33 (4): 637–708.
Sharif, G. 2006. Galicia: A Concise History. Cardiff: University of Wales Press.
Shin, M. 2001. The politicization of place in Italy. Political Geography 20 (3):
331–52.
Vilas Nogueira, J. 1992. Las elecciones en Galicia (1976–1991). Barcelona:
Institut de Ciències Polítiques i Socials.
Chapter 6
Place, Identity, and Taiwan’s Presidential
Elections, 1996–2008
Daniel McGowin
On 23 March 1996, the Taiwanese people celebrated a monumental day in

the history of the Republic of China (ROC). On that day, Kuomintang (KMT)
nominee Lee Teng-hui beat out two opponents to remain the president of the
ROC. This election was significant not because Lee was re-elected, but it marked
the first time that the Taiwanese people were able to directly elect the leader of
the country.
Previous to the 1996 presidential election, the now defunct National Assembly
held the responsibility of selecting the president of the ROC. The National
Assembly, established in 1948 in accordance to the constitution of the ROC, only
elected into office two presidents during its existence – Lee Teng-hui in 1988 and
Chiang Ching-kuo in 1978.1 When a new National Assembly was established in
1991, the decision to abandon indirect presidential elections was made and the
process towards the country’s first direct presidential elections commenced. This
culminated in the historic re-election of Lee in 1996.
Since this election, Taiwan has held direct presidential elections every four
years, with the most recent occurring in 2008. During the past 12 years, Taiwan
experienced a shift in political parties, with the opposition Democratic People’s
Party (DPP) winning both the 2000 and 2004 presidential elections. By the time
a KMT candidate returned to the presidency in 2008, a clear geographic pattern
had emerged among the Taiwanese electorate. While the 1996 elections in Taiwan
were a landslide victory for the KMT in terms of counties and municipalities won,
the 2000 elections experienced a dramatic change in the electoral geographic
landscape. Due to a split in the KMT, the opposition DPP was able to garner a
plurality of votes and its nominee Chen Shui-bian won the presidency. With this
election, a political cleavage emerged in the southwestern part of the island. Chen
succeeded in winning re-election in 2004, where the distinct voting pattern from
2000 held true. With the 2008 victory by KMT nominee Ma Ying-jeou, there was
a shift back to the old guard. However, the pattern that emerged in 2000 was once
again reinforced in the 2008 presidential elections.
1 Yen Chia-ken, who succeeded Chiang Kai-shek, was vice president at the time of
Chiang’s death and therefore ascended to president without the National Assembly electing
him to the position.
Because political parties in Taiwan tend to carry a particular position on

Taiwan’s international status, the emergence of such patterns are likely linked
to locally-held opinions on independence, as well as questions of identity and
regional rivalries. The purpose of this chapter is to explore the voting patterns
of the four presidential elections in Taiwan and how these patterns represent the
ongoing debate over Taiwan’s international status. The chapter seeks to further
the work done by electoral geographers linking votes and elections with debates
over identity, and add to the limited number of studies of Asian elections with the
English-language electoral geography literature (see Lee and Brunn [1996] for one
example from South Korea).
The chapter begins with a history of political parties in Taiwan, followed by
discussion of the intersections of identity politics and political parties. The chapter
concludes with an analysis of the 1996, 2000, 2004 and 2008 presidential elections,
and suggests how place and identity are crucial concepts in understanding the
geographic patterns of these votes.
Political Parties in Taiwan
The Origins of the Kuomintang
Since its founding, the Republic of China has been dominated by the Kuomintang
(KMT), be it in the capacity of the presidency, in the legislature, or in the local
offices. This dominance was so prevalent that the KMT was synonymous with
‘Nationalist China,’ a term used to differentiate the ROC from the People’s
Republic of China (Wright 1955).
The roots of the KMT date back to the Qing Dynasty (Booth 1999). The
original purpose of the KMT was revolutionary in nature; ‘a revolution to
establish the Republic’ (KMT website 2009). Once the Republic of China was
established and the KMT gained power over the mainland in 1912, the ideology
of the party shifted from one of revolution to one of state-building. However,
the wars with the Communist and with Japan made it difficult for the KMT to
establish itself internationally. Despite Western support, the Nationalist suffered
numerous losses to the Communist and eventually fled to the island of Taiwan.
It was here that a new KMT ideology developed, one that continues to influence
the party even today.
However, it is important to note the transformation of the KMT’s ideology
did begin while the party and the ROC controlled the mainland. The switch
from a revolutionary ideology to a restoration ideology was based on the need to
legitimate its claim as the leader of China. In this sense, the KMT felt it best to
project itself as a continuation of Chinese history rather than a completely new
beginning. Party officials, including Chiang Kai-shek, also sought to promote the
Confucian traditions that were prevalent, albeit intermittently, throughout China’s
long history (Wright 1955). Yet, the KMT’s shift was also influenced by interaction
Place, Identity, and Taiwan’s Presidential Elections, 1996–2008 99
with the Soviet Union, as in 1924 Sun Yat-sen adopted Leninist elements from the
Soviets, even going as far as to using the Soviet Communist Party as a model for
reorganization (Dickson 1996). This Leninist influence played a significant role in
the reorganization of the KMT once it fled to Taiwan.
The Reorganization of the KMT and Its Current Philosophies.
The goal of the KMT/ROC was never to remain on the island of Taiwan, but to
reorganize and retake the mainland from the Communist. But coexisting on the
island with the native Taiwanese would prove to be a difficult task.
The term ‘native Taiwanese’ refers to those people who arrived on the island
previous to the ROC gaining possession of the island in 1945. Bedford and Hwang
(2006, 5) note that:
85 percent of Taiwan’s current residents have ancestors who emigrated from

China three- to four-hundred years ago … The remaining 13 percent of
nonindigenous residents arrived in Taiwan or are descended from those who
arrived in Taiwan between 1945, when Japan ended its colonization, and 1949
when Chiang Kai-shek and his armies and followers arrived.
This latter group is referred to as waisheng, which references their ‘foreign’ nature
to Taiwan. This distinction, while not necessarily visible, is an important ethnic
difference that plays out in present-day Taiwanese politics. Because the KMT
were viewed as outsiders and were the minority on the island, they needed a way
to consolidate power.
The KMT did little to enamor themselves with the Taiwanese in the immediate
years following the end of the Pacific War. For all the negative actions that
the Japanese took during the Pacific War, it is also clear that they modernized
and invested in Taiwan (Cumings 1984). In fact, by the time the Japanese were
forced out, Taiwan had actually advanced past China. This notion was reinforced
by the actions of the Chinese army that arrived on the island. Due in part to
receiving low pay, the soldiers plundered the local community. Furthermore,
the KMT ‘dismantled Taiwanese infrastructure including telephone wires, pipes,
metal roofing, fire hydrants, and railroad switches and shipped everything back
to the mainland to support the effort there, asserting that Taiwan’s citizens owed
a debt to the KMT for liberating them from the Japanese’ (Bedford and Hwang
2006, 6). As crime rates rose and the KMT took control of major industries, a
feeling of resentment began to fester. The local feeling at the time is summed
up best by a local saying quoted by Bedford and Hwang (ibid.) ‘that the dogs
(who could at least protect your property) had been chased away, but the pigs (who
only make a mess) had come.’
Taiwanese resentment culminated in what is known as the 228 incident.
Beginning on 27 February 1947, protests began in Taipei and spread throughout
the country. The response was brutal as the Republic of China’s army, backed by
the KMT, violently suppressed the uprising (Wu 1995; Cooper 1998; Huat 2001).
This anti-government revolt not only galvanized Taiwanese resentment towards
the KMT, but it also solidified a mainland identity to the KMT. The 228 incident
‘resulted in considerable bloodshed after soldiers were called from the mainland
to put it down, and in its aftermath there was deep antipathy on the part of the
Taiwanese toward Mainland China, the KMT, and the government’ (Cooper 1998,
15). The subsequent suppression of the majority Taiwanese by the KMT left an
impact that continues to define the political landscape in Taiwan. As Huat (2001,
135) notes:
local Taiwanese suffered a regime of ‘white terror’ under the KMT’s security
apparatus, identified with the colonizing mainlanders, who constituted no more
than 15 percent of the population. The terms ‘mainlanders’ and ‘locals’ thus
became an ideological/identity/political divide providing one of the central
discursive resources for the organization of opposition to the KMT and preventing
the formation of a Taiwan ‘nation’ so long as the ‘mainlanders’ are in power.
Thus, the KMT was ascribed with the mainlander identity, regardless of the
ethnicity of its members. Additionally, any party that came to oppose the KMT
could potentially be ascribed with a Taiwanese identity.
After the ROC leadership fled the mainland for Taiwan, the KMT struggled to
find its place on the island. While the KMT initially attempted to undercut identity
politics by sinicizing the Taiwanese (Huat 2001), the party soon realized that it
needed to adapt to its new political environment. As Dickson (1995, 49) notes
that ‘the main goals and tasks of the KMT changed: Rather than concentrating
on plans to retake the mainland, the party devoted more of its energy to issues
of immediate concern to Taiwan and its own reputation.’ One way to accomplish
this was to Taiwanize the government, as well as the party itself, by allowing
Taiwanese to join its ranks. In order to do this, the KMT utilized local elections to
integrate Taiwanese into the political process. However, the central government
remained in the hands of the KMT. This integration process created the façade of
an all-inclusive government while maintaining the legitimacy of the ROC state
(Rigger 2001). A similar tactic was repeated in the 1970s as the KMT government
attempted ‘to maintain the legitimacy of its power’ (Bedford and Hwang 2006, 7).
In this case, Taiwanese elites were allowed to join the party.
Not surprisingly, this shift in attitude coincided with an international shift
in recognition. KMT leadership recognized that reclaiming their stake to the
mainland would have to be put off and a new focus would need to be shift to
simply maintaining some form of legitimacy. There was some relaxing of the
KMT’s political control, but martial law, for example, remained in effect until
the mid-1980s. Taiwanization of the party placed Taiwanese members into local
offices and helped integrate Taiwanese, including future president Lee Teng-hui,
into the central government (Dickson 1995). While change was slow, by 1985 the
influx of Taiwanese members of the KMT had made a profound impact. Using data
gathered by Yang (1985), Dickson (54–5) notes that ‘the proportion of Taiwanese
among district cadres rose to 56.6 percent in 1975 and to 73.3 percent in 1985. At
the county level, the proportion of Taiwanese increased from 34.5 percent to 53.9
percent between 1975 and 1985.’ While some of this could be explained by the lack
of immigration from the mainland – and therefore fewer mainlanders from which
to choose – this pattern marks a new trend within the KMT; one implemented by
Chiang Ching-kuo that differed from his father Chiang Kai-shek, who preferred
mainlanders to Taiwanese. And even though the numbers of Taiwanese in the
provincial cadres were not as high, the impact of locally visible Taiwanese KMT
members was important. ‘In local offices where society came into direct contact
with the party, the increased visibility of Taiwanese in party posts created at least
the appearance of greater responsiveness to local wants and needs, a basic goal of
the Taiwanization strategy’ (Dickson 1996, 55).
The reign of Chiang Ching-kuo also marked a departure from the tight-fisted
control demonstrated by his late father. The younger Chiang began to loosen up
restrictions on opposition parties, something that was forbidden during the era
of martial law. In 1986, the opposition Democratic Progressive Party (DPP) was
established and Chiang took ‘a tolerant position’ (KMT website 2010). One year
later, martial law was officially lifted on the island. These steps not only helped
establish an opposition to the KMT, but it also provided a much louder voice for
the Taiwanese people; a voice that signaled the beginning of ethnic politics.
From Dangwai to Opposition: The Rise of the DPP
One way the KMT kept Taiwanese from directly electing the president, or even
establishing political parties, was the notion that the Republic of China was not
confined solely to the island of Taiwan. Indeed the KMT-controlled ROC stressed
the fact that they represented the mainland. Therefore, to allow the people of Taiwan
to be the only ones selecting the leader for all of China would disenfranchise
their fellow countrymen and women on the mainland. While this line of reasoning
worked in the 1950s and 1960s, the changing political climate of the 1970s eroded
this façade and led the KMT to reconsider its position.
Although the KMT dominated politics on the island, there were a few
independent candidates who were able to break through and reach the upper
echelon of the Taiwanese political hierarchy. One of the first was Kao Yu-shu
(Henry Kao), who was able to defeat KMT opponents to become mayor of Taipei
in 1954 and then again in 1964. Kang Ning-hsiang was able to work his way into
the Legislative Yuan by 1972. Since he was not a member of the Kuomintang, he
took up the moniker of a Dangwai candidate, or an ‘outside the party’ candidate.
This denotation, once held in a negative light and seen as a threat to the KMT, is
an important step towards creating an organized opposition to the KMT. Although
the Dangwai had its share of factionalism (Rigger 2001), the rise of Dangwai
candidates in the 1970s helped stoke the fire of opposition in Taiwan. It is from
this movement that the DPP gained its initial momentum.
The wave of democratization and the loosening of the authoritarian system

on the island began in the early 1970s. Triggered by the supplementary elections
to replace members of the Legislative Yuan who either retired or passed away,
non-KMT politicians began to break into the political system. However, it would
be a mistake to assume that the opposition held any power or that it was a solid
cohesive unit. There were some attempts to organize the opposition movement,
most notably the Dangwai and its accompanying political publication Formosa.
The magazine set up offices throughout the island and helped create a network
in which to build opposition. The threat posed by the Dangwai via Formosa
was very real in the eyes of the KMT leadership. The rising opposition and the
KMT’s fear culminated in a clash in 1979 in the southern city of Kaohsiung. This
incident, known as the Kaohsiung Incident or the Meilidao Incident, occurred
on 10 December 1979 as opposition leaders held protests against the KMT-led
government. Police, both civilian and military, responded with violence and
several opposition leaders, including those that did not participate in the protest
were jailed (Rigger 2001; Bedford and Hwang 2006; Wu 1995). While some of
the leaders were removed from the political process, the fallout of the Kaohsiung
Incident led to a rise of new opposition leaders, including Chen Shui-bian, who
was an attorney for the jailed dissidents (Rigger 2001).
The Kaohsiung Incident marked a transition into a comparatively more relaxed
political climate. Ching-kuo allowed for some publications to continue printing
dissenting voices. One such publication was The Eighties, headed by moderate
opposition leader Kang Ning-hsiang. Kang represented a more tolerable opposition
in comparison to the Formosa group. However, despite the arrests and depletion of
activist opposition leaders, there were still factions in the Dangwai movement that
challenged the moderate position, namely the defense lawyers and a new group
of radicals (Rigger 2001; Wu 1995). While some of the defense attorneys and
even family members of jailed dissidents won seats in the 1980 elections, the
Dangwai movement factionalized even more. Radical magazines such as Deep
Plow accused the moderate Kang of having close ties with the KMT. When the
two sides attempted to work together in 1983, disagreement arose about how the
selection process should be conducted; moderates preferred a top-down approach
while the activists favored a bottom-up method. By 1984, there were at least four
factions: ‘the Writers and Editors Alliance, the Dangwai Public Policy Research
Association (DPPRA), the Progress Faction … and the imprisoned Formosa
Faction’ (Rigger 2001, 23). The radical Writers and Editors Alliance would soon
establish the New Tide Magazine, which continued to be the voice of the radical
faction until 2006 when DPP factions were abolished (DPP website 2010; Rigger
2001).
Despite factionalism, the Dangwai collectively pursued the establishment of a
political party. Under martial law, political parties other than the KMT were illegal.
Nevertheless, the opposition recognized the importance of having a political
party. In 1986, Kang established a DPPRA branch in Taipei. After that time, the
opposition spent the summer carefully planning their move. On 28 September,
the opposition arrived at a consensus and established the Democratic Progressive

Party. ‘The party represented a coalition of all camps within the Dangwai: the
Mainstream Faction (or Kang Group), New Tide (formerly the Writers and Editors
Alliance), and Lin Cheng-chieh’s Progress Faction. The Formosa Faction was
absent at the founding only because the leading Formosa activists were still behind
bars’ (Rigger 2001, 25). It is important to note that the party was founded despite
the fact that such parties were illegal. However, Chiang Ching-kuo, who had
advocated for reform, turned a blind eye to the DPP and held back conservative
calls for prosecution (Wu 1995).
The DPP wasted little time breaking into the upper tier of the Taiwanese
political system, making headway in 1986, 1989 and 1992. By March 2000, just
shy of 14 years after its founding, the DPP claimed the presidency in Taiwan’s
second direct presidential election.
What does the 2000 victory by Chen Shui-bian represent for Taiwan? Or, more
importantly, how does this victory reflect the identity debate in Taiwan? To answer
this, it is imperative to understand the identities of the political parties of Taiwan,
namely the KMT and DPP.
Identity Politics and Political Parties in Taiwan
As noted above, the KMT carries with it the mainlander identity. This denotation
is due in large part because of the migration of the party from the mainland to
Taiwan. However, the connection runs deeper than a simple transfer of location
from Nanjing to Taipei. Under Chiang Ching-kuo, the KMT went to great lengths
to Taiwanize the party. However, events such as the 228 Incident and the Meilidao
Incident injured the party’s reputation. With the KMT resembling more of a
colonizer rather than fellow citizens during the early years, the KMT was stained
with the outsider stigma. Regardless of the steps taken by Ching-kuo, Lee Teng-
hui and other party leaders, the KMT continues to wear the mainland label.
The DPP, by way of comparison, carries with it the party of the Taiwanese.
The 228 incident, which created a great rift between the Taiwanese and the
mainlanders, remains an important symbol of the opposition movement. That
incident came to represent the resistance against the mainlanders and, by extension,
the KMT (Hsieh 2008). Through this connection to the 228 Incident, as well as the
Meilidao Incident, the birth of the DPP was in direct opposition to mainlanders as
a whole, and the KMT in particular. Through these means, the DPP is identified
as ‘Taiwanese.’
However, to breakdown the KMT and DPP solely on the basis of ethnic identity
is erroneous. While the parties may have carried a strict ethnic identity early on,
especially the KMT, that is no longer the case. As noted above, through the efforts
of Chiang Ching-kuo, the KMT has become more diversified in its ethnic make-
up. According to Huang (1996, 115), the ethnic composition of the KMT shifted
from 60.6 percent mainlanders in 1969 to only 30.8 percent in 1992. Additionally,
the composition of the KMT Central Committee grew from 3.1 percent Taiwanese
in 1952 to 53.3 percent Taiwanese in 1993 (118). This is also reflected in the
electorate, as Taiwanese and Mainlanders were equally likely to support the KMT
(Rigger 2001, 164). This trend indicates that the KMT’s composition has lost
its mainlander identity, even if it continues to be painted as a mainlander party.
The reason for this has more to do with support for the DPP. As Rigger notes,
‘Taiwanese were far more likely to support the DPP than were Mainlanders.’ Thus,
Rigger concludes, the characterization of the DPP as the party of the Taiwanese
remains an accurate one. If the KMT is to continue to carry with it the identity of
the mainland, it is only because of its historical antecedents and by way of contrast
to the DPP rather than anything based on reality.
However, the primary difference between the DPP and KMT rests on the
issue of national identity rather than ethnic identity. Hsieh (2008, 12–3) notes
that ‘as Taiwan became more and more democratic, Taiwanese, as the majority
population, gradually got the upper hand, and the salience of ethnicity in Taiwanese
politics declined. In its stead, the issue of national identity became more and
more prominent in Taiwan’s politics.’ It is important to note what is meant by
‘national identity.’ In this case, the sides fall either for independence (from China),
the status quo, or reunification (with China). Political parties that are considered
to be part of the pan-Blue coalition, namely the KMT, People First Party (PFP)
and the New Party (NP) tend to support the status quo with a slight lean towards
reunification. The pan-Green coalition tends to support independence and is led by
the DPP, the Taiwan Solidarity Union (TSU) and the appropriately named Taiwan
Independence Party (TAIP). It is worth noting here that support for independence
or for status quo/unification does not necessarily fall along ethnic lines. The issue
of national identity has existed since the KMT arrived on the island. While this
issue coexisted along with the ethnic divide, the Taiwanization of the KMT and
the rise to power of Taiwanese has lessened the role of ethnicity and given way to
the effect of national identity. Nevertheless, there are some mainlanders who favor
independence and some native Taiwanese who support the status quo. The issue of
national identity could not easily fall along ethnic lines because, as Bedford and
Hwang (2008, 11) note, ‘with native Taiwanese constituting the vast majority of
the population, if the independence-unification issue were decided along purely
ethnic lines, the pan-blues would have no hope of getting elected.’
That the national identity issue is the main division between the two major
parties, and more generally the two coalitions, is portrayed in a table provided by
Rigger (2001, 165). In a post-election survey after the 1995 legislative election,
41 percent of Taiwanese and 44.7 percent of mainlanders were likely to support
the KMT (21.9 percent and 5.4 percent, respectively, for DPP support). In terms
of self-identification, 34.5 percent of those who consider themselves Taiwanese
supported the KMT while 51 percent of self-identified Chinese sided with the
KMT. In general, Taiwanese and mainlanders were equally likely to support the
Kuomintang. However, when it came to the question of national identity, 49.5
percent of unification supporters and 45.7 percent of those that wanted to maintain
the status quo preferred the KMT. In contrast, only 20.2 percent of supporters for
independence were likely to back the KMT. Conversely, 52.8 percent of voters
who backed independence preferred the DPP, with supporters of unification (10.7
percent) and status quo (15.7 percent) not likely to favor the DPP.
What these numbers suggests is that the division between the two main
Taiwanese political parties is along national identity issues rather than ethnic
issues. However, it would be erroneous to suggest that this is clear cut, that the
national division is the only difference between the parties, or that ethnic identity
plays no role. The characteristics of the KMT are more muddied than that of the
DPP. In fact, those that support independence who still backed the pro-unification
party was double the percentage of those unification supporters who still preferred
the pro-independence DPP. As Rigger concludes, ‘party identification is not a
simple matter of ethnic identity or policy preference.’ However, it is clear that the
DPP elicits a certain type of voter – ethnic Taiwanese; pro-independence – than
does the KMT. Thus, the supporters of the DPP are more defined by their ethnic
composition and policy preference than the KMT. Ethnic identity is still important
with regards to the DPP given the parties pro-independence lean – a policy that
resonates well with many Taiwanese. With the number of self-identified Taiwanese
tripling since the 1990s (Bedford and Hwang 2006), ethnic identity continues to be
important for the DPP.
What can be stated here is that because of how the supporters of the DPP
are defined, supporters of the KMT are ‘othered’ as being mainlanders and pro-
unification, even if it is not entirely accurate. Yet, the ethnic divide has become
less of an indicator of support while the national identity issue continues to
provide a clearer understanding for one’s support for a given party. Certainly
ethnic identity is an indicator of support for independence (Rigger 2001; Hwang
and Bedford 2006; Hsieh 2008), but the national identity issue still remains the
better determinate for party support in Taiwan. And, given the platforms of the
two parties, the national identity issue is still a central division to the parties itself
even if the electorate may not be as clearly defined, at least with regards to the
Kuomintang.
Thus, how does geography aid in better understanding this division? Does place
prove to be a clearer determinate for party support than ethnic identity or policy
preference? In turn, how does place reflect the debate over national identity; in
other words how does geography explain the discourse over what it means to be
‘Taiwanese’? As a way to illustrate this, the remainder of this chapter will examine
direct presidential elections since 1996. However, before doing so it is necessary to
briefly explain the process of selecting Taiwan’s president prior to 1996.
The Democratic Transition in Taiwan
As noted above, the National Assembly, the constitution-amending body of the

Taiwanese government, elected the president for the Republic of China prior to
1996. But, elections at all levels have been limited to some degree by the KMT-
controlled government. The freeze on elections for legislative members as well
as the ban on political parties created a one-party system that ensured the KMT
remained in power while it attempted to reorganize and plan its return to the
mainland. With the shift in attitudes in the international community, the KMT
began to soften its stance and slowly allowed for a more open and democratic
system to take effect.
Tien (1996) notes that the transition from an authoritarian system of government
to modern democracy took place in three phases. The first stage, from the 1950s
until 1971, witnessed elections at a local level. The first local elections took place in
1952, allowing Taiwanese to vote for town chiefs, county and city councilpersons,
and city mayors. Direct elections were not allowed above the provincial level.
During this time non-KMT candidates were able to receive as much as 35 percent
of the vote. However, because of the implications of martial law, these non-KMT
candidates could not organize as a party. In fact, the KMT banned candidates from
engaging in political activities outside of their voting districts.
The second period (1972–85) was marked by the decline in international
recognition of the Republic of China. As noted above, as a means of maintaining
some semblance of political legitimacy the KMT began to Taiwanize its party and
by extension Taiwanize the political hierarchy. The KMT also began to allow an
expansion of the election process in Taiwan. There were supplementary elections
for the National and Legislative assemblies in 1969 and 1972, then more regularly
beginning in 1980. The dangwai were allowed to participate as a quasi-party and
began making their way into the various assemblies. Tien attributes this expansion,
in part, to the rising economic condition in Taiwan. Economic growth in Taiwan
led to more opportunities for higher education attainment, greater personal wealth,
a larger middle-class, and a more diverse media industry. ‘These developments
provided not merely the requisite conditions for democratic change but also the
additional driving forces for liberalizing the political system.’10 Additionally, more
Taiwan-born political activists meant fewer connections to the mainland and a swing
in momentum in favor of opposition. This trend set the stage for the third phase.
With the birth of the DPP in 1986, the third phase opened the door for
elections under a competitive party system. However, it was not until after the
death of Chiang Ching-kuo in 1988 that political parties were legalized. Under the
leadership of Lee Teng-hui, more steps were taken to liberalize the political system
of Taiwan. The most important here is the move to direct elections for president.
Lee worked with both KMT and DPP leaders to hatch out a plan to allow the voters
of Taiwan to decide on their president. In 1992, the Legislative Assembly passed
an amendment that would abolish the previous method for electing the president.
However, the amendment did not immediately decide on a new method, choosing
instead to wait until one year before the next election, which was in 1996 (Lu
2002). The amendment to the 1947 constitution, which was enacted in 1994, also
set the length of terms to four years each with a maximum of two consecutive
terms (Lu 2002; Myers et al. 2002).
The effects of the democratization process of Taiwan can be seen in the

early open elections of the Legislative Assembly. The Dangwai/DPP saw its
representation in the Legislative Yuan jump from 13 percent in 1980 to 33.2
percent in 1995 before sliding down slightly to 29.6 percent in 1998 (Tien 1996,
16; Diamond 2001, 53). Additionally, Diamond notes that by 1995, the KMT’s
representation in the Legislative Yuan dipped below 50 percent. Similarly, the
KMT’s representation among county magistrates and city mayors fell below 50
percent in 1993 (Tien 1996, 17). This pattern culminated in the 2000 election of
Chen Shui-bian as president, giving the DPP a plurality of the votes. Four years
later, Chen secured just over 50 percent of the votes for the DPP.
Since this election, Taiwan has held direct presidential elections every four
years, with the most recent occurring in 2008. During the past 12 years, Taiwan
experienced a shift in power, with the opposition Democratic People’s Party
(DPP) winning both the 2000 and 2004 presidential elections. By the time a
KMT candidate returned to the presidency in 2008, a clear geographic pattern had
emerged among the Taiwanese electorate. And because political parties in Taiwan
tend to carry a particular position on Taiwan’s international status, the emergence
of such patterns are likely linked to locally-held opinions on independence, as well
as questions of identity and regional rivalries.
Four Presidential Elections in Taiwan: 1996–2008
The presidential election in 1996 witnessed an overwhelming victory for the KMT,
with their candidate Lee Teng-hui winning 24 of the 25 districts. Independent
candidate Lin Yang-gang, who finished third in the voting, captured his home
county of Nantou in central Taiwan. Lee garnered a majority of the votes in 18 of
Taiwan’s 25 districts and at least 40 percent of the votes in five other districts. The
only two districts where he received less than 40 percent of the votes were in the
aforementioned Nantou County (31.5 percent) and Taipei City in the north (38.9
percent). Thus, the 1996 election pattern represents a fairly solid voting bloc in
Taiwan as the KMT was able to secure victory throughout the country.
While the 2000 presidential election involved five candidates, the competition
was primarily between three individuals – Lien Chan of the KMT, Chen Shui-
bian of the DPP, and James Soong of the newly formed People First Party (PFP).
With 39.3 percent of the vote, Chen won the election and for the first time it ROC
history the country was led by someone not of the KMT party. Not only did the
KMT lose, Lien did not win a single electoral district. While Chen won the popular
vote, Soong actually won more districts – 15 to Chen’s ten. Here, a geographic
pattern begins to emerge as the DPP won districts in the south and west as well as
Yilan County in the north, and Soong won a contiguous string of districts from the
west to the north, including the three western island districts – Kinmen County,
Lienchiang County and Penghu County. It should be noted that Chen only gained a
majority of votes in one district, his home county of Tainan, while Soong managed
to receive a majority of votes in five districts.
Both the 2004 and 2008 presidential elections pitted only two candidates in
each election – one from the KMT and one from the DPP. In 2004, Chen sought
re-election under the DPP endorsement as Lien Chan attempted to restore power
to the KMT. In a very narrow and somewhat controversial election, Chen was able
to capture a second term for the DPP, gaining 50.1 percent of the popular vote.
He carried the same ten districts as he did in 2000 while also picking up Taichung
County in central Taiwan. While obvious that Chen would win the majority of
votes in those 11 districts, the margin of victory is noteworthy as he was either
close to or exceeds a three-fifths majority in seven of those 11 districts. His largest
margins of victory were in the south.
In 2008, Chen was unable to run again due to term limits. Therefore, the DPP
nominated (Frank) Hsieh Chang-ting while the KMT nominated former Taipei
mayor Ma Ying-jeou. The KMT was able to return to power with a landslide victory
(58.45 percent versus 41.55 percent). Ma wrested away six districts and secured all
seven municipalities. Hsieh’s five victories were concentrated in the south.
Place, Identity, and Taiwan’s Presidential Elections
The 12-year evolution from a seemingly ‘solid Taiwan’ to the cleavage in the south
is rooted in both the historical experience of the island and the political dynamics
of Taiwan.
It should first be noted that the appearance of a ‘solid Taiwan’ had more to
do with Lee Teng-hui’s overall popularity rather than an affinity towards the
KMT or its traditional stance. Lee’s approval rating remained above 60 percent.
An ethnic breakdown of his approval ratings shows that he ‘enjoyed rather high
popularity among Taiwanese of Fujian origin, moderately high popularity among
Hakka speakers, and rather low popularity among mainlanders’ (Lu 2002, 62). It is
important to note that his low approval rating among mainlanders, a stalwart of the
KMT, demonstrates fractures in the base of the party. As an ethnic Taiwanese, he
easily appealed to the ‘native Taiwanese’ that tend be pro-independence (Bedford
and Hwang 2006). Furthermore, Lee was instrumental in deconstructing the old
system and democratizing the island, something that was seen as a challenge to
mainland China. Although it should be noted president Lee stated that he was not
in favor of independence (Lu 2002).
The results of the 2000 election can be informed by the work of Webster
(1992), who notes that the demise of the ‘solid South’ in the United States was
due in part to the emergence of a third party. The emergence of a third party in
Taiwan is crucial in allowing the opposition DPP to gain control of the Taiwanese
presidency. James Soong’s candidacy as an independent greatly divided both the
mainlander vote, as well as the native Taiwanese vote. Cheng and Hsu (2002) note
that the KMT nominee, Lien was not as popular as Soong. Lee, as chair of the
KMT, did not support the more popular Soong due to previous disagreements and
chose the less popular Lien. As a result, Soong opted to run as an independent and
immediately became the front-runner to win the election (Rigger 2001).
Because many KMT supporters did not feel that Lien would win the
presidency, voters shifted their allegiance to the candidate best suited to defeat
Chen – Soong. ‘[T]he KMT’s Lien was the second preferred candidate for most
voters, mainlanders and [native Taiwanese] alike. … [C]onsistently lagging
behind the other two contenders in polls rendered Lien a nonviable alternative,
and eventually he became the victim of strategic voting’ (Cheng and Hsu 2002,
166–7). The division of the KMT vote is important in that if we were to combine
the votes for Lien and Soong, then the numbers would reflect previous elections
in Taiwan (a 60/40 victory for the KMT). Thus, this division allowed for the DPP
and Chen to narrowly wrest power away from the KMT.
Additionally, Webster (1992) suggests that the impetus for electoral change
in the US South is situated in urban areas. While DPP candidate Peng Ming-min
was unable to win a single district in the 1996 election, his highest shares were in
metropolitan areas and provincial cities. Peng also did well in what would become
staples of the DPP electorate – counties in the south as well as Yilan County in
the northeast. Similarly in 2000 and 2004, in districts won by the KMT, Chen
Shui-bian fared better in city districts than in county districts. Furthermore, DPP
support tends to be the lowest in the rural counties of Hualien and Taitung. This
pattern is reflected in numbers cited by Rigger (2001, 166), where she notes that
DPP support is weakest among those involved in agriculture.
Another key point in the 2000 election is that Chen and the DPP distanced
itself from the pro-independence platform. Chen quickly noted that the DPP only
wanted a referendum on independence, rather than outright independence. For
Chen’s part, he moved his position more towards the middle, aligning himself
more the with Lee’s policies. While Lee Teng-hui never publicly endorsed Chen,
many of those close to Lee backed the DPP candidate in the weeks leading up
to the election. These endorsements included Lee Yuan-tseh, Noble laureate, and
Hsu Wen-lung, president of plastic producer Chi Mei Corporation who claimed
that Chen was best suited to ‘continue Lee Teng-hui’s course’ (Rigger 2001, 193).
These late endorsements gave Chen a final push to the presidency.
It is important to note that Chen’s emphasis on self-determination was not
simply political rhetoric during his campaign for presidency; Chen publicly
espoused such views as early as 1991. While hardline leaders of the DPP wanted
a platform seeking independence, Chen sought to soften the stance by stating
that a move should be decided by the people and not by one political party.
After a disastrous showing in the 1991 National Assembly elections, the DPP
began moving further away from independence ‘plank’ and by 1996, once the
hardliners left the DPP to form the Taiwan Independence Party, the DPP’s position
was closer to Lee Teng-hui’s middle position. This stance placed the DPP in a
more comfortable position to claim the presidency. Chen reiterated this stance
in his 2000 inauguration speech in which he advocated the ‘five noes’ and the
maintenance of the status quo (Chen 2000).2
While the DPP carries the mantle of the independence movement, in fact it
has more in common with the KMT than public perception indicates. As Rigger
(2001, 130) writes, ‘the DPP still is no fan of unification or ‘One China,’ but it
has accepted the two Chinas arrangement that exists, in effect, today.’ This stance,
which supports the status quo backed by the KMT, is promoted primarily because
Taiwan is, for all intents and purposes, independent, but also because of security
concerns. ‘The DPP’s current position is that because Taiwan already enjoys de
facto independence, a declaration of de jure independence is unnecessary and
would only antagonize Beijing.’ Thus, DPP leaders ‘accepted the logic that since
Taiwan already enjoyed independence, there was no need to change the status
quo’ (128 and 130). Therefore, even though the DPP advocates for a referendum
on Taiwanese independence (DPP website 2010), in general there is acceptance
by the party of the status quo. It is through this ‘new middle way’ (Bedford and
Hwang 2006) that Chen was able to calm fears of an independence push and work
his way towards the middle.
The resulting victory by Chen and the DPP gave the opposition footing that
carried over to the 2004 election in which Chen narrowly defeated a joint ticket of
Lien and Soong. By 2008, however, the KMT was able to emphatically return to
power as Ma Ying-jeou defeated DPP candidate Frank Hsieh. At this point, a clear
geographical pattern emerged.
By taking the election results in terms of regional blocs (see Figure 6.1),3 it is
clear to see that the opposition DPP holds a firm grip on the south, while the KMT
is strongest in the east and north (see Figures 6.2 through 6.5). The central districts
in Taiwan do not clearly fall into one camp as it has voted for both KMT and DPP
candidates. Also, there is one outlier as the DPP has claimed Yilan County in the
north in two of the four elections – 2000 and 2004.4
So, what explains this pattern? There are several factors that come into play
here. First, there is an ethnic difference. Rigger (2001, 164) notes that ‘Taiwanese
2 While Chen did push for a referendum to rewrite the constitution, a move some
argued was akin to declaring independence, he stated that a new constitution would not
address independence. Chen, therefore, stood by his five noes and maintained the status quo.
3 While arbitrary, the author breaks down the regions of Taiwan as follows: Central
(Changhua, Miaoli, Nantou, Taichung and Yunlin counties, and Taichung city), East
(Hualian and Taitung counties), North (Hsinchu, Taipei, Taoyuan and Yilan counties, and
Keelung, Hsinchu and Taipei cities), and South (Chiayi, Kaohsiung, Penghu, Pingtung and
Tainan counties, and Chiayi, Kaohsiung, and Tainan cities). For this analysis, the island
counties of Lienchiang and Kinmen were not included. These regions are consistent with
other regionalizations of Taiwan.
4 Yilan County has demonstrated a long history of opposition to the KMT. It was
one of the first to locally elect a non-KMT leader and, despite the 2008 election, remains a
stronghold for the DPP (Wang 2009).
Figure 6.1 Map of the Regions Figure 6.2 Map of the 1996
of Taiwan Presidential Election
Source: Author. Source: Author.
Figure 6.3 Map of the 2000 Figure 6.4 Map of the 2004
Presidential Election Presidential Election
Source: Author. Source: Author.
Figure 6.5 Map of the 2008

Presidential Election
Source: Author.
and Mainlanders were equally likely to support the KMT. However, Taiwanese
were far more likely to support the DPP than were Mainlanders.’ Historically,
Mainlanders are concentrated in the north and east, both KMT strongholds, and
are sparser in the south, the DPP’s territory. Additionally, the eastern counties
of Hualian and Taitung have the largest percentage of indigenous peoples, a
traditional supporter of the KMT.
Another factor that Rigger (167) notes is ‘political differences between the
north and the south.’ In their analysis of voting in South Korea, Lee and Brunn
(1996) note that historical rivalries and resentment between the southwest and
other parts of the country, especially the north, explain political cleavages in
South Korea. A similar north–south rivalry is present in Taiwan. Because of this
rivalry, ‘the DPP has benefited from resentment in southern Taiwan at the wide
discrepancy in government spending between north and south’ (Rigger 2001,
167). Considering that the KMT long-controlled government spending, the south’s
affinity towards the opposition, in this case the DPP, makes sense.
But this resentment reflects the center of opposition that the south came to
represent during the democratizing process and the emergence of (new) political
parties. Rigger (167, emphasis added) notes that ‘the DPP’s emphasis on Taiwanese
identity, including language and culture as well as Taiwan independence, has been
well received in the south, where Taiwanese consciousness is strong.’ Given that
the KMT’s platform with regards to mainland China is one of cooperation and
reconciliation, the DPP is often held as the party of independence and therefore
Taiwanese nationalism. This perception coupled with Taiwanese consciousness
centers the south as a site of opposition.
The connection between the DPP and the south is galvanized by the Meilidao
Incident, also known as the Kaohsiung Incident, which is held as one of the
catalysts for the formation of the DPP. On December 10, 1979 in the southern
city of Kaohsiung, a human rights protest was violently put down and three days
later several political opposition leaders were arrested. The resulting outcry over
this incident led to an increase in opposition movements, including a group of
defense attorneys that included Chen Shui-bian. Seven years later, the Democratic
Progressive Party was established in Taipei (Rigger 2001, Bedford and Hwang
2006). Although it was created in a hotel in the north, it was the incident in the south
that fueled the opposition movement. The Meilidao Incident and its connection to
the DPP places the opposition movement in the south.
What this all represents is that above any other indicator, place is a main
factor in understanding the voting patterns in Taiwan. Agnew (1987, 34) writes
that places can ‘become collective actors.’ This is important to note with regards
to the role of political parties. ‘The ideologies and policy proposals of different
parties appeal more successfully in some places than in others as they strike
responsive and nonresponsive chords among dominant sectors of the electorates.’
Agnew continues by noting that certain actions, such as opposition movements,
are rooted in places, which act as a seat of sentiment. In other words, a sense of
place ‘can present itself through different manifestations,’ such as the DPP. ‘Place,
therefore, is not the outcome on a map of an abstract social process beyond place.
Place refers to the process of social structuration’ (36, emphasis in original). The
pattern that takes shape in Taiwan is a result of place-based factors – concentration
of mainlanders in the north; Kaohsiung as a symbol of opposition; KMT support
among agricultural workers in the east – more so than simply a matter of ethnic
difference or policy preference. Certainly the latter two variables are important,
but each is rooted in space and helps create the place-based identity that gives rise
to the pattern in Taiwan.
References
Agnew, J. 1987. Place and Politics. Boston: Allen and Unwin.

Bedford, O. and K.K. Hwang. 2006. Taiwanese Identity and Democracy: The
Social Psychology of Taiwan’s 2004 Elections. New York: Palgrave McMillian.
Booth, M. 1999. The Dragon Syndicates: The Global Phenomenon of the Triads.
New York: Carroll & Graf Publishers.
Chen, S.B. 2000. Taiwan stands up: Toward the dawn of a rising era. [speech]
Taiwan Public Policy Council. Transcript available at: http://www.tppc-us.org/
LinPDF/ChenShuiBian5-20-00.pdf.
Cheng, T.J. and Y.M. Hsu. 2002. The March 2000 election in historical and
comparative perspectives: Strategic voting, the Third Party, and the non-
Duvergerian outcome. In B. Dickson and C.M. Chao (eds) Assessing the Lee
Teng-hui Legacy in Taiwan’s Politics: Democratic Consolidation and External
Relations. Armonk, NY: M.E. Sharpe.
Cooper, J. 1998. Taiwan’s Mid-1990s Elections: Taking the Final Steps to
Democracy. Westport, CT: Praeger Publishers.
Cumings, B. 1984. The origins and development of the Northeast Asian political
economy: Industrial sectors, product cycles, and political consequences.
International Organization, 38 (1): 1–40
Democratic Progressive Party. 2010. Platform. Available at http://www.dpp.org.
tw/index_en/.
Diamond, L. 2001. Anatomy of an electoral earthquake: How the KMT lost and
the DPP won the 2000 presidential election. In M. Alagappa (ed) Taiwan’s
Presidential Politics: Democratization and Cross-Strait Relations in the
Twenty-first Century. Armonk, NY: M.E. Sharpe.
Dickson, B. 1996. The Kuomintang before democratization: Organizational
change and the role of elections. In H.M. Tien (ed.)Taiwan’s Electoral Politics
and Democratic Transition: Riding the Third Wave. Armonk, NY: M.E. Sharpe.
Hsieh, J. 2008. Democracy in a mildly divided society. In P. Paolino and J. Meernik
(eds) Democratization in Taiwan: Challenges in Transformation. Burlington,
VT: Ashgate Publishing.
Huang, T.F. 1996. Elections and the evolution of the Kuomintang. In H.M. Tien
(ed.) Taiwan’s Electoral Politics and Democratic Transition: Riding the Third
Wave. Armonk, NY: M.E. Sharpe.
Huat, C.B. 2001. Defeat of the KMT: Implications for one-party quasi-democratic
regimes in Southeast Asia. In M. Alagappa (ed.) Taiwan’s Presidential Politics:
Democratization and Cross-Strait Relations in the Twenty-first Century.
Armonk, NY: M.E. Sharpe.
Kuomintang. 2010a. ‘Introduction.’ [Online]. Available at http://www.kmt.org.tw/
english/page.aspx?type=para&mnum=105.
Kuomintang. 2010b. ‘History.’ [Online]. Available at http://www.kmt.org.tw/
english/page.aspx?type=para&mnum=108.
Lee, D.O. and S. Brunn. 1996. Politics and regions in Korea: An analysis of the
recent presidential election. Political Geography, 15(1): 99–119.
Lu, Y.L. 2002. Lee Teng-hui’s role in Taiwan’s democratization: A preliminary
Assessment. In B. Dickson and C.M. Chao (eds) Assessing the Lee Teng-
hui Legacy in Taiwan’s Politics: Democratic Consolidation and External
Relations. Armonk, NY: M.E. Sharpe.
Myers, R., L. Chao and T.C. Kuo. 2002. Consolidating democracy in the Republic
of China on Taiwan, 1996–2000. In B. Dickson and C.M. Chao (eds) Assessing
the Lee Teng-hui Legacy in Taiwan’s Politics: Democratic Consolidation and
External Relations. Armonk, NY: M.E. Sharpe.
Rigger, S. 2001. From Opposition to Power: Taiwan’s Democratic Progressive

Party. Boulder, CO: Lynne Rienner Publishers.
Tien, H.M. 1996. Elections and Taiwan’s democratic development. In H.M. Tien
(ed.) Taiwan’s Electoral Politics and Democratic Transition: Riding the Third
Wave. Armonk, NY: M.E. Sharpe.
Webster, G. 1992. Demise of the solid South. Geographical Review, 82(1): 43–55.
Wright, M. 1955. From revolution to restoration: The transformation of
Kuomintang ideology. The Far Eastern Quarterly, 14(4): 515–32.
Wu, J. 1995. Taiwan’s Democratization: Forces behind the New Momentum. Hong
Kong: Oxford University Press.
Chapter 7
The Power of Definition: The Puerto Rican
Diaspora and the Island’s Status Referenda
Luis D. Sánchez-Ayala
Almost every Puerto Rican agrees that sports raise passions and wake profound
sentiments among their compatriots. The list of those sports so highly regarded
includes basketball, baseball, boxing, and of course, at the top of the list, voting in
any electoral event. For Puerto Ricans, the electoral debate is present in everyday
popular discourse and in the spaces of everyday life. Politics is the main subject
discussed at the coffee shop, in the beauty parlor, or while waiting in line at the
supermarket or the bank. Also, often Puerto Ricans choose the colors of their
houses, cars, or even their clothes according to their electoral preferences.
This is the reality of a country whose political future is still uncertain after
centuries of foreign domination through a variety of colonial arrangements.
Consequently, for Puerto Ricans any electoral event in the island has a potential to
influence the political future and final status of the nation. However, Puerto Ricans
have been unable to agree on one particular status formula to solve the island’s
political dilemma; they continue to disagree over which direction would be best
to follow.
There have been three popular referenda on the island offering a variety of
formulas as options for a final political status of Puerto Rico. Those referenda
were celebrated in 1967, 1993, and 1998, but none has succeeded in solving the
political limbo of the island.
Since the last referendum in 1998, many more initiatives to celebrate another
referendum in the island have been discussed in both Washington and San Juan,
but the debate regarding Puerto Rico’s political status referenda is much more
complex than this would suggest. This complexity exists because the universe
of Puerto Ricans interested and involved in this matter expands well beyond
the territorial boundaries of Puerto Rico. Currently there are nearly five million
self-identified Puerto Ricans living in the United States, many of whom are still
attached to the political issues of the island. However, these members of the Puerto
Rican diaspora are not officially authorized to vote on any electoral event in the
island. Yet discussions about whether Puerto Ricans living in the United States
should be included in any consideration regarding the final decision about the
island are taking place on the island and in the United States. The controversy rests
on who would be considered “Puerto Rican” for the purpose of participating in any
potential status referenda.
This twist gives a new significance to the debate over Puerto Rico’s status
situation. Not only is the political situation of the island being deliberated, but
there also debates about what it means to be Puerto Rican among members of the
Puerto Rican diaspora. In this regard, determining who can vote or participate
in any decision-making process dealing with the island’s political future implies
issues of identity definition and construction for the nearly five million Puerto
Ricans who live in the United States. Nonetheless, in the case that Puerto Ricans
living in the United States are included in any potential status referendum, they
will represent a significant force in deciding the political future of Puerto Rico and
the Puerto Ricans living in the island.
The Status Debate
The debates over the political status of Puerto Rico unfold on two different fronts,
those that originated in Puerto Rico and those that originated in Congress or within
the United States government.
The status debate in Puerto Rico is present in the everyday popular discourse
of the island. The entire political system is arranged around the status issue. The
three existing political parties in the island were founded to respond to the status
issue, each one representing a different status option: statehood, commonwealth,
and independence.
The debates originating in Puerto Rico have led to referenda on the issue on
three different occasions. Those referenda were held in 1967, 1993, and 1998
by initiative of the government of Puerto Rico in agreement with the three local
political parties (the pro-statehood New Progressive Party, the pro-commonwealth
Popular Democratic Party, and the pro-independence Puerto Rican Independence
Party). However, in all three cases the referenda have not succeeded in their goal
of solving the political limbo of the island and have not been approved or endorsed
by the government of the United States. Without Congressional approval or
endorsement, the referendum results were reduced to popularity contests between
the three political parties in the island.
Nonetheless, lack of attention from Congress was not the only obstacle to the
effectiveness of these referenda. Out of the three referenda, two were boycotted
by at least one of the political parties on the island. In 1967 the Puerto Rican
Independence Party, known by its Spanish-language acronym, PIP, boycotted the
referendum, arguing that its only purpose was to reinforce the Commonwealth
image in front of the international community. On the other hand the Statehood
Republican Party, which at that time was the pro-statehood party, was divided, and
although its leaders supported the boycott, a group of the party’s young leadership
led a movement to participate in the referendum.
The other referendum that was boycotted was the one held in 1998. This
referendum was initiated by the governing New Progressive Party, known as
The Power of Definition 119
“PNP,” the successor of the Statehood Republican Party), and supported by the
Puerto Rican Independence Party. The referendum included four options:
1. Commonwealth status, or the application of the US Congress sovereignty

over Puerto Rico;
2. Free Association, a non-colonial association with the US;
3. statehood, the integration of Puerto Rico into the union as the 51st state; and
4. independence, the recognition of Puerto Rico as a free and sovereign
country.
The pro-commonwealth Popular Democratic Party (PPD), which never has

initiated any debate or conversation about the political status issue, opposed the
referendum. The leaders of the party did not agree with the PNP and the PIP on
the definition that was given to the term “Commonwealth.” As a result, the three
parties agreed to include a fifth column that would say “None of the Above.” The
PPD then started a campaign promoting “None of the Above,” arguing that the
referendum was not legitimate.
In the 1967 referendum, the commonwealth option prevailed with 60 percent
of the votes, against 39 percent for statehood and one percent for independence
(Figure 7.1). The Commonwealth option dominated almost every municipality
of the Island, obtaining more than 50 percent of support in most. However, the
Puerto Rican Independence Party’s boycott and the Statehood Republican Party’s
internal division set the stage for the victory of the commonwealth option and its
promoter, the PPD.
In the early 1960s the island was still dominated by a strong PPD. Just a
decade after the creation of the Commonwealth of Puerto Rico, the party (main
actor in the creation of the commonwealth status) was in the last stage of its
electoral momentum. Its involvement in the referendum had a deep effect on the
population’s voting behavior.
The 1993 referendum showed a changing pattern. Even though the
commonwealth option prevailed, its margin of support had decreased dramatically
since the 1967 referendum. On the other hand, support for statehood increased
significantly, obtaining significant support. Likewise, the difference in support
percentages between commonwealth and statehood grew closer in comparison to
the 1967 referendum.
In the case of independence, there were significant changes. Support for
independence increased significantly, from barely one percent to four percent,
although it remained a small minority. In Puerto Rico, both the local government
and US Government have often condemned those who support the independence
option. Systematic vigilance against and repression of independence supporters
were common practices, applied simultaneously with an exaggerated and
unrealistic fear campaign conducted among the population against independence.
The fact that the independence movement has resisted and survived all those
obstacles for so long is a demonstration of its strength.
Figure 7.1 Status Referendums in Puerto Rico

Source: Comisión Estatal de Elecciones de Puerto Rico.
Finally, the 1998 referendum deviated from the electoral patterns found in the
previous two referenda. Held several months after hurricane George devastated
the island, and with the strong opposition of the PPD, the population was not
enthusiastic about the election. The low level of enthusiasm was demonstrated by
the referendum’s low turnout.
In this referendum, 50 percent of the voters voted for “None of the Above,”
a fifth column championed by those who were opposed to the celebration of the
referendum. The PPD made a call for voters to select “None of the Above,” calling
it “the option of the people.” In the commonwealth-statehood-independence
referenda, the PPD, who opposed to the Commonwealth definition on the ballot,
managed to divert the debate and change the referendum into a punishment vote for
the government. Regardless of this, the statehood option maintained its support (47
percent). Likewise, independence got three percent, commonwealth (as appeared
in the ballot), 0.1 percent, and the fifth column “None of the Above” obtained 50
percent, just one percent more than the commonwealth option obtained in the 1993
referendum.
Most choices in politics are framed in terms of a new proposal being pitted
against the status quo, and in such a situation the status quo is at an advantage
(Christin, Hug, and Sciarini, 2002: 764). In Puerto Rico, the commonwealth
option has an advantage for those who are uncertain of whether the new political
status options will achieve their goals. However, there is enough evidence to state
that political status preferences are currently changing.
The results of the three referenda demonstrated the shift of the voting
behavior in Puerto Rico. Currently the popularity of the commonwealth option
is in decline, while support for statehood has been rising. Support for remaining
a commonwealth has gone from 60 percent in 1967 to 0.1 percent in 1998, while
support for statehood has gone from 39 percent in 1967 to 47 percent in 1998.
Likewise, support for independence went from one percent in 1967 to 3 percent
in 1998. But if we take out the 1998 results, since the commonwealth supporters
boycotted the commonwealth option and instead voted for the “None of the Above”
option, the same pattern can be seen. Also, if we add the percentage of support for
None of the Above (50 percent) with the percentage in favor of the commonwealth
(.1 percent), it would not represent a significant difference. Overall, support for
the commonwealth is still showing a decline, contrary to the increase in support
for statehood.
These patterns can also be seen in the results of the general elections. In Puerto
Rico, the elections are intimately related to the political status discussion. The role
of the three political parties is to promote a political status formula (statehood,
commonwealth, or independence). The general elections are dominated by status
issues, and the status referendum campaigns in turn are dominated by the political
parties. As LeDuc (2002) stated, “the positions that political parties take, either in
the evolution of the referendum issue or during the campaign itself, provide one of
the strongest available information cues to voters.”
In 1989, the heads of the three political parties in Puerto Rico (among them the
governor of Puerto Rico) wrote a letter to President Bush demanding a response to
the island’s status situation. As a result, President Bush instructed Congress to take
action. Senator Bennett Johnston (D-Louisiana) presented in the Senate the S. 712
“Puerto Rico Status Referendum Act.” The bill was referred to a Senate Committee,
where it died in 1991; it never made it to the floor to be considered and voted on.
In 1996, the Republican leaders of the House, through the chair of the
Resources Committee, Don Young (R-Alaska), presented H.R. 856, the “United
States-Puerto Rico Political Status Act.” The House approved the bill by one vote,
after an overnight session and a heated debate. However, the bill was not passed
by the Senate; it did not even make it to the floor for consideration, as was the case
for the S. 712 in 1989.
More recently, discussions over the political status of Puerto Rico have taken
place in Congress. This time Congressional action has responded to the White
House’s report released in December 2005 by the President’s Task Force on Puerto
Rico’s Status. Following the report’s recommendations, Congress is debating
possible legislation to address the political situation of Puerto Rico.
Soon after the White House’s report was released, the political parties in Puerto
Rico started to pressure Congress. Consequently, two bills were introduced in
Congress, one in the Senate and one in the House. The bill introduced in the Senate,
designed and supported by the pro-commonwealth party in Puerto Rico, discards
the White House’s recommendations of a referendum only offering statehood and
independence. This bill does not contemplate a referendum or any further debate
over Puerto Rico’s stature. Rather, it proposes that Puerto Ricans should first decide
internally which path they want to follow. On the other hand, the bill introduced in
the House has been designed by the pro-statehood party with the support of the pro-
independence party. This bill follows the recommendations of the Presidential Task
Force on Puerto Rico’s Status, so it proposes holding a referendum on the island
offering statehood or independence. The Senate bill died soon after its introduction;
however, the House bill H.R.2499, the “Puerto Rico Democracy Act of 2010”
passed on April 29, 2010 and now must be taken up by the Senate. This bill would
introduce a two-step ballot measure for Puerto Rico to decide if Puerto Ricans want
to change their current relationship with the United States. If they vote to change
their status, they can then choose to become a state, pursue independence, or seek
some other “political association between sovereign nations.”
The Puerto Rican Diaspora
Over the past century, millions of Puerto Ricans have migrated to the United States.
The 2008 American Community Survey conducted by the US Census Bureau
estimated the population of Puerto Rico at 3,954,037 and the number of Puerto
Table 7.1 Puerto Rican Population in the Island and the United States
Year Puerto Rico United States

1940 1,869,255 69,967
1950 2,210,703 301,375
1960 2,349,544 887,662
1970 2,712,033 1,492,396
1980 3,196,520 2,013,945
1990 3,522,037 2,727,754
2000 3,808,610 3,406,178
2008 3,954,037 4,216,533
Source: US Census Bureau.
Ricans living in the US at 4,216,533 (Table 7.1). According to these numbers,

currently there are more Puerto Ricans living outside of Puerto Rico than there are
living in the island. This situation has given rise to a diasporic community larger
than the total population of the island. Therefore, defining Puerto Ricanness is a
process that reaches far beyond the island’s boundaries.
Puerto Ricans have been migrating to the mainland since the island became a
possession of the United States in 1898. By the first decades of the 1900s, Puerto
Ricans were being hired to work as seasonal workers as far away as Hawaii (Rosario
1983; Maldonado-Denis 1980; Mintz 1955). The 1910 US census estimated the
Puerto Rican population on the mainland at 1,500. This figure increased even more
after the approval of the Jones Shafroth Act of 1917, which established the island
as an organized but unincorporated territory of the United States and granted US
citizenship to Puerto Ricans. This act facilitated the movement of labor to the
mainland, significantly raising the net migration from the island to the mainland
in the 1920s. However, the peak period of migration began with the end of World
War II. During this period, which became known as the Great Migration, the
Puerto Rican community grew to more than half a million.
Nonetheless, the Great Migration led to the creation of Puerto Rican barrios
in New York City’s South Bronx, Spanish Harlem, Manhattan’s Lower East Side,
and Brooklyn’s Atlantic Street. These barrios began to resemble “Little Puerto
Rico,” characterized by Puerto Rican ways and customs.
During the 1960s and 1970s, the migratory flow from the island to the mainland
took other forms. At this time a drastic decrease in the outflow of migrants from
the island to the mainland and a rapid increase of return migration to Puerto
Rico occurred. Consequently, beginning in 1965, Puerto Rican migration was
characterized by continuous back and forth movement between the island and the
mainland (Lorenzo-Hernández 1999: 989).
From the 1980s to the present, an increase in migration to the United States
has been taking place. The new migrants have demonstrated dispersion patterns
outside of the northeast of the United States, reaching to the South and West in
states like Florida, California, and Texas (Figure 7.2).
In the 1980s, over 400,000 Puerto Ricans left the island for the mainland. In
fact, 12.3 percent of the Puerto Rican population that resided on the island in
1980 was no longer living there in 1990 (Rivera-Batiz and Santiago 1990). Puerto
Ricans’ net migration to the mainland increased during this time period, reaching
126,465 in the 1980s and 111,336 in the 1990s. The combined total of these
numbers represented over a quarter of a million people, or more than 7 percent of
the island’s population in 1980 (Marquez 2004).
From 1990 to 2000 the Puerto Rican population in the United States increased
faster than the population on the Island. Between 1990 and 2000, the population
of mainland Puerto Ricans grew by 25 percent while the population in the island
increased by approximately 10 percent (Marquez 2004). The 2008 US Census
Bureau American Community Survey estimated the island population at 3,954,037
Figure 7.2 Distribution of Puerto Ricans in the United States, 1990–2009

Source: Author.
and the number of mainland Puerto Ricans at 4,216,533: for the first time there are
more Puerto Ricans living in the mainland United States than on the island.
Historically, Puerto Ricans in the United States concentrated in New York,
New Jersey, Pennsylvania, Ohio, Illinois, and other Northeastern and Midwestern
states. However, during this last emigration wave the mainland Puerto Rican
population started to disperse much more widely than in the previous emigration
movements. Consequently, the proportion of Puerto Ricans significantly increased
in other states such as Florida, which has had the greatest increase, and Texas,
and California. Yet regardless of the fact that New York still has the largest
percentage of Puerto Rican residents in the US, the growth of Florida’s Puerto
Rican population has been spectacular, from slightly more than two percent of all
stateside Puerto Ricans in 1960 to more than 17 percent by the year 2006. These
numbers placed Florida with the second largest concentration of Puerto Ricans in
the United States and as the fastest growing Puerto Rican community.
Status Referenda and Puerto Ricanness
The political status debate over Puerto Rico’s political situation, besides dealing
with the issue of defining the island’s political future, had been broadened to
include issues of Puerto Rican identity in Puerto Rico and the diaspora. This is a
particularly important issue as identities are more than just labels; identities also
involve people’s acceptance, resistance, choice, particularity, invention, perceptions,
rejections, and so forth (Cornell and Hartmann 1998: 77). Therefore, identities are
socially constructed through structural processes rather than being innate or pre-
given (Bondi 1993: 86). This process gives meaning to identities on the basis of
cultural attributes. Identity is a product, and, at the same time, the producer of social
phenomena; it is always an incomplete process. Accordingly, Keith and Pile (1993:
28) suggest that “identity formation is endless because it is based on a moment of
arbitrary closure which in the same fashion is both true and false simultaneously.”
In that sense the concept of identity is a highly contested one.
Puerto Ricans, regardless of their location inside or outside of Puerto Rico,
show a strong sense of national identity. However, when it comes to defining their
community, each Puerto Rican has his or her own interpretation and meanings
(Sánchez 2009: 66). In that sense, when it comes to identity, each individual
potentially experiences a unique understanding of his or her surroundings, and
this understanding is shaped by mental processes of information gathering and
organization that make self-consciousness (Gold 1980). Geographers such as
Peet (1998) argued that the spatiality of everyday life generates fundamental self-
consciousness.
However, the significance of all this is the fact that the debate over Puerto
Rico’s political situation is put in motion a process of identity construction for
Puerto Ricans in Puerto Rico, but equally importantly, in the United States. In this
context, the discussions taking place in the US Congress about whether Puerto
Ricans living in the United States should be included in any consideration regarding
the final decision about the island are increasingly dominating the status debate.
The controversy rests on how and who would be considered Puerto Rican in order
to participate in any potential status referendum. Therefore, this controversy could
mean profound changes in the way many Puerto Ricans living in the United States
have constructed their identities.
That congressional debate over the political future of Puerto Rico has many
angles that directly affect the Puerto Rican diaspora. Perhaps the most prominent
ones can be summarized in the views of Congressional Representatives José
Serrano from New York and Luis Gutiérrez from Chicago. Serrano (born in Puerto
Rico) has argued that all persons who were born on the island should participate
in any potential mechanism to solve Puerto Rico’s status issue. On the other hand,
Gutiérrez (born in the United States, first generation of Puerto Rican immigrants)
maintained that all persons of Puerto Rican descent should be included in any
decision regarding the political future of Puerto Rico. Along with these views
there are many others that range from allowing participation to anyone claiming
to be Puerto Rican to only including the residents of the island in such a decision.
Each possible scenario has different implications for the identity-making process
of the Puerto Rican diaspora. If Congress chooses to follow the interpretation of
Serrano, affirming that all persons who were born on the island should participate,
Congress will then determine that a Puerto Rican is only a person born in the
island. Therefore, in that sense the message to the diaspora would be that Puerto
Ricanness is restricted to place of birth and any other individual, regardless of his
or her ancestry, would not be recognized as such. This scenario would drastically
affect the way that millions of “Puerto Ricans” born in the United States produce
and reproduce their individual and collective identities.
On the other hand, if Congress chooses to favor the views of Luis Gutiérrez,
affirming instead that the participation in a status referendum should be open to
all persons of Puerto Rican descent, the scenario for the Puerto Rican diaspora
would be different. In this case, anyone who can trace his or her ancestry to the
island would be eligible to vote, and thus included within the boundaries of Puerto
Ricanness. However, this interpretation could open the door for the question of
where to put a limit, or if there should be a limit for such claim of ancestry. This
flexible or limitless perspective could as well affect the way that the Puerto Ricans
in the diaspora produce and reproduce their Puerto Ricanness, as it could stretch
to the limit the particularities that distinguish a migrant group from a diaspora.
Similarly, a congressional decision to restrict the participation on a potential
referendum to only residents of the island could influence some perceptions of
Puerto Ricanness among the Puerto Rican diaspora. In this sense, such exclusion
could draw a boundary of division/distinction that could be interpreted as the
legitimation of the true Puerto Ricans, in opposition to the “others.” Consequently,
this would mean a potential process of identity [re]definition for the Puerto Rican
diaspora.
These different points of view on the status debate over Puerto Rico
demonstrate the complexity behind the consideration of a referendum in Puerto
Rico. Yet, regardless of which direction the debate will finally take, Congress
will be deciding much more than the political future of Puerto Rico; they will
as well be setting the guidelines of Puerto Ricanness. Thus, in the case of Puerto
Rico, Puerto Ricans would not be only voting to decide the political future of the
island, they will also be defining the boundaries of their community, in both the
island and the diaspora. Therefore, a status referendum in Puerto Rico could put
in motion a process of identity construction for Puerto Ricans in Puerto Rico, but
more importantly in the United States.
For Puerto Ricans living in Puerto Rico, considerations of their political status
could lead them to a process of defining their Puerto Ricanness in relation to
the political status of Puerto Rico (e.g., Puerto Rican, Puerto Rican-American,
American, etc.). However, for the Puerto Rican diaspora the answer to the
question of whether they should be included in any consideration regarding the
final decision about the island is particularly important as their identity as Puerto
Ricans could be dramatically [re]defined with the outcome of the congressional
debate over a potential status referendum for Puerto Rico. Therefore, for the
Puerto Rican diaspora, voting in such a referendum would mean much more than
just participating in an electoral event; it could instead signify the way in which
they make sense of themselves individually and collectively, and thus, it could
determine the future of their community.
Moreover, the status debate is making visible notions of difference within the
Puerto Rican nation. While Puerto Rican identities are based on language, race,
ethnicity, religion, sexuality, gender, class, nationalism, they are predominantly
based on place of origin (Sánchez 2009: 7). Therefore, answering the question
of who is Puerto Rican or what makes someone Puerto Rican, to then determine
who would be included for the purpose of participating in any potential status
referendum, has had the effect that terms such as “island-born Puerto Rican,”
“mainland-born Puerto Rican,” “Nuyorican,” “Orlandorican,” and “Diasporican”
are now being used as the main means to define affiliation. For that purpose, Puerto
Ricans are creating (or reviving) labels and concepts of “othering.” The labels
used to differentiate social groups are important for the psychological segregation
and perpetuation of the categories (Lorenzo-Hernández 1999: 991). In this sense,
the consequential [re]definition of Puerto Ricanness is establishing opposites such
as “us versus them.” We must not forget that one of the most powerful ways to
construct an identity is by identifying against (Cornell and Hartmann 1998). First
define the “other” and then we define ourselves as “not the other,” the opposite or
what the other is not. But in the same way we define “others,” they may also define
us, and their definition may not match our own self-identity (Sánchez 2009: 8).
The status referendum debate is making Puerto Ricans aware of the fact that
Puerto Ricanness means different things for different people at different places in
different times. Consequently, such differences are being used to build boundaries
between different notions of Puerto Ricanness, and thus to determine who can vote
or participate in any decision-making process dealing with the island’s political
future. As Barth (1996: 15) puts it:
the dichotomization of others as strangers, as members of another ethnic group,

implies recognition of limitations on shared understandings, differences in
criteria for judgment of value and performance, and a restriction of interaction

to sectors of assumed common understanding and mutual interest.
The status debate is also accentuating the notions of difference among members of
the Puerto Rican diaspora. In response to all the discussion and questions that the
status debate had initiated, the members of the Puerto Rican diaspora are trying to
identify among them who are the “real” Puerto Ricans. Consequently, the Puerto
Rican diaspora, and their notions of difference, is splitting between the “old”
communities (particularly the three historical communities of New York, Chicago,
and Philadelphia) and the “new” communities (particularly central Florida).
Puerto Ricans in central Florida are distancing themselves from the rest of
the diaspora, also adopting the use of terms such as “Nuyoricans” to reinforce
their authenticity claims. They are building an identity based on what they
think is from the island, away from the rest of the Puerto Rican diaspora on the
mainland (Sánchez 2009: 92). On the other hand, Puerto Ricans in the historical
communities perceive themselves as the center of the Puerto Rican diaspora, and
thus, an integral part of the Puerto Rican nation, rejecting any claim of distinction
from those in other parts of the United States.
Equally important in this debate is the fact that a possible vote of the Puerto
Rican diaspora in a status referendum over the political future of the island could
represent the difference between which option ultimately wins. The current
volatility in Puerto Rico regarding popular support of the different political status
options shows that no status option has sufficient support in the island to be
considered representative of the political desires of the people of Puerto Rico as a
whole. The results of all three referenda on the island have shown how supporters
of commonwealth and statehood are merging together, with almost no difference
between the percentage of Puerto Ricans supporting the commonwealth option
and those supporting statehood supporters for the two options. Therefore, since
Puerto Rican residents on the island are nearly evenly divided on their political
desires, the votes of the Puerto Rican diaspora could decide which political status
option would win and become the permanent political status of Puerto Rico.
The Puerto Rican diaspora is a significant element in the discussions over a
potential vote to finally solve the political limbo of Puerto Rico. Yet, they are
significant in the debate not only because their potential vote could affect the ultimate
outcome of such a referendum, but more importantly because the consequences of
the decision of whether or not to include them in such a referendum to begin with
will cause reactions that could [re]define Puerto Ricanness, both in Puerto Rico
and the United States.
Conclusion
The debate over the possible inclusion of members of the Puerto Rican diaspora in
a referendum regarding the political future of the island gives a new significance
to the debate over Puerto Rico’s status situation. The debate has the potential to
put in motion a process of identity definition among the Puerto Rican nation.
Consequently, not only the political situation of the island is being deliberated, but
also the question of who is Puerto Rican or what makes someone Puerto Rican,
particularly among the Puerto Rican diaspora.
Determining who can vote or participate in any decision-making process
dealing with the island’s political status situation implies issues of identity
definition and construction. The outcome of those considerations will potentially
[re]define Puerto Ricanness, particularly for the nearly five million Puerto Ricans
who live in the mainland United States. In this sense, the making of Puerto
Ricanness blends with the definition of the political future of the island.
In the case of the diaspora, the process to find an answer to the question of who
is Puerto Rican or what makes someone Puerto Rican, to then determine who would
be included for the purpose of participating in any potential status referendum, is
initiating a process of self-definition, in which Puerto Ricans are constructing their
identities in an inbetweenness of being “here” but also being “there.” This stage of
in-betweenness, or “third space” (Bhabha 1994; Soja 1996), is a state of belonging
to both and to neither, or as Bhabha (1994: 219) said, “neither one not the other
but something else besides, in-between.” This diversity allows the multiplicity of
identities based on the existence of multiple layers of identity (Sánchez 2009: 72).
Determining who can vote or participate in any decision-making process
dealing with Puerto Rico’s political future not only had initiated a process of
identity definition and construction for Puerto Ricans in Puerto Rican and the
United States, but has also demonstrated that the task of identifying Puerto
Ricanness is particularly difficult.
References
Bhabha, H. 1994. The Location of Culture. London: Routledge.

Bondi, L. 1993. Locating identity politics. In M. Keith and S. Pile (eds) Place and
the Politics of Identity. London: Routledge, 84–101.
Brah, A. 1996. Cartographies of Diaspora: Contesting Identities. London:
Routledge.
Christin, T., S. Hug, and P. Sciarini. 2002. Interests and information in referendum
voting: An analysis of Swiss voters. European Journal of Political Research
41: 759–76.
Cornell, S. and D. Hartmann. 1998. Ethnicity and Race: Making Identities in a
Changing World. Thousand Oaks, CA: Pine Forge Prints.
Gold, J. 1980. An Introduction to Behavioural Geography. Oxford: Oxford
University Press.
Keith, M. and S. Pile. 1993. The place of politics. In M. Keith, and S. Pile (eds)
Place and the Politics of Identity. London: Routledge, 22–40.
LeDuc, L. 2002. Opinion change and voting behaviour in referendums. European

Journal of Political Research 41: 711–32.
Lorenzo-Hernámdez, J. 1999. The Nuyorican’s dilemma: Categorization of
returning migrants in Puerto Rico. International Migration Review 33(4):
988–1013.
Maldonado-Denis, M. 1980. The Emigration Dialectic: Puerto Rico and the USA.
New York: International Publishers.
Márquez, C. 2004. Adios, Puerto Rico; Hola, United States. Caribbean Business
September 16.
Mintz, S. 1955. Puerto Rican migration: A three-fold comparison. Social and
Economic Studies 4: 311–25.
Peet, R. 1998. Modern Geographical Thought. Oxford: Blackwell
Rivera-Batiz, F. and C. Santiago. 1996. Island Paradox: Puerto Rico in the 1990s.
New York: Russell Sage Foundation.
Rosario, C. 1983. Éxodo Puertorriqueño: Las Emigraciones al Caribe y Hawaii,
1900–1915. San Juan: Ramallo Brothers.
Sánchez, L. 2009. The New Puerto Rico?: Identity, Hybridity and Transnationalism
within the Puerto Rican Diaspora in Orlando, Florida. Germany: VDM Verlag.
Soja, E. 1996. Third Space: Journeys to Los Angeles and Other Real-and-Imagined
Places. Cambridge, MA: Blackwell.
Part III
Electoral Geography in the US
Chapter 8
Class, Ethnicity, Religion and Place
in the 2008 US Presidential Election
Barney Warf
The 2008 presidential election was unquestionably “as riveting and historic a
spectacle as modern politics has ever produced” (Heilemann and Halperin 2010,
ix). After eight years of Republican rule under George W. Bush, a period that
witnessed two foreign wars and rising economic dysfunction, the United States
elected its first African-American leader, Barack Obama. By a margin of 53 to
46 percent of the votes cast (69.3 million v. 59.6 million), and 365 to 174 in the
electoral college, Obama won an election that is by any measure historic.
As always, this contest played out unevenly over space, as local constellations
of economic, political, and cultural forces favoring Obama and his GOP opponent,
John McCain, were precipitated in different combinations and to varying
Figure 8.1 Percent of Votes for Obama by County, 2008

Figure 8.2 Percent of Votes for McCain by County, 2008
levels across the face of the nation. Obama’s primary sources of strength lay in
metropolitan areas, the West Coast, the upper Midwest, New England, and a belt
of historically and predominantly African-American counties in the South (Figure
8.1). For the first time in decades, traditionally Republican-leaning states such as
Virginia, North Carolina and Indiana joined the Democratic column. Conversely,
McCain garnered majorities in most rural areas and small towns, notably in much
of the South, the Great Plains, and the intermontane West (Figure 8.2).
However, underlying this geography are several sociospatial categories that
profoundly shape American electoral landscapes, notably class, ethnicity, and
religion. This chapter examines the 2008 presidential election through the lenses
of each of these dimensions. The first three sections offer an overview of how
each of these categories has shaped voting patterns, often in unexpected ways.
Class, for example, traditionally absent in American politics, is not manifested in
a simple dichotomy of wealthy support for Republicans and lower income support
for Democrats, but rather the reverse; ethnic support for different political parties
is widely variable among groups, and often within them; and the influence of
religion varies markedly by denomination. The fourth section offers a quantitative
analysis using geographically weighted regression, a means of examining the
spatially uneven strength of different variables’ influence on votes for Obama while
statistically controlling for the influence of confounding forces. The results are
Class, Ethnicity, Religion and Place in the 2008 US Presidential Election 135
presented in a series of maps. The concluding discussion links these outcomes to

a wider discussion of the geographies of identity, difference, and marginalization.
Class, Income, and Party Alignment: A Counterintuitive Pattern
Simplistic views of American politics tend to assume that the Republican Party
represents vested corporate and business interests, whom they have consistently
and zealously supported when in office, while the Democrats are allegedly
sustained by a wide range of groups in civil society, including unions, teachers,
low income people, and ethnic minorities. In this reading, support for the GOP
should rise monotonically with median family income. Unfortunately, the reality
is more complex, and less hopeful from the perspective of those concerned about
the poor. Contrary to the stereotype, in fact statistically it was wealthier counties
that were more likely to vote for Obama, while McCain garnered a majority in
the poorest counties (Table 8.1). Thus, Obama, on average, won 58 percent of
counties with above-average household incomes, including almost 63 percent
of those with incomes over 120 percent of the national median, while McCain
gained a slight majority of the vote in the poorest counties, i.e., those with
incomes under 70 percent of the national median. In short, support for Obama
rose steadily as median incomes rose, while they declined for McCain.
There are several forces at work that underpin this apparent discrepancy. Since the
1970s, the United States has been in the grip of a tremendous social transformation
wrought by globalization, deindustrialization, stagnant incomes, rising inequality,
growing demographic diversity, and wholesale class war launched by the ruling
class under the banner of neoliberalism. Given its long history of individualism, anti-
intellectualism, nationalist militarism, religiosity, anti-government rhetoric, obsession
with a mythologized “free market,” lack of social democracy, and weakened labor
unions, no major political party effectively represents the interests of the working
class, the poor, or the increasingly beleaguered middle class. Rather than a turn to
Table 8.1 Votes in 2010 Presidential Election by Household Income
Household Income Votes for % of Votes for % of

as % of Average Obama Votes McCain Votes
> 120 7,512,019 62.8 4,450,295 37.2
100–119 8,721,561 55.1 7,118,178 44.9
90.0–99.9 8,556,514 57.8 6,258,803 42.2
80.0–89.9 15,958,416 55.1 13,021,362 44.9
70.0–79.9 14,570,760 51.2 13,869,966 48.8
< 69.9 14,159,650 48.2 15,216,153 51.8
Total 69,478,920 53.7 59,934,757 46.3
Source: Calculated by author.
working-class politics, mounting frustration at stagnant or declining standards of

living, as well as deep seated fears about declining white hegemony, have driven
countless millions of middle-income voters into the hands of conservatives, who
preach a doctrine that substitutes liberals for the ruling class. The rise of conservatives
since the 1970s, pioneered by Ronald Reagan and propelled by right-wing think tanks
and conservative hegemony of the media (e.g., Fox News), has found widespread
popular support among white Americans, particularly blue collar males, who often
blame their economically precarious positions on the modest gains of marginalized
groups such as women, immigrants, ethnic minorities, and gays. As a result, white
Democratic voters began turning to the Republican Party en masse in the 1970s and
1980s, most particularly in the South (Aistrup 1996).
As Thomas Frank (2004) noted in his influential book What’s the Matter
with Kansas?, much of the popularity of American conservativism among the
working class owes its existence to the propagation of a mythologized view of
America as a uniquely blessed country increasingly overwhelmed by satanic
force of moral deviance, as manifested by the rise of social indicators such as
out-of-wedlock births, abortion ostensibly available on demand, multiculturalism,
coddled criminals, and conservatives’ favorite obsession, sexual permissiveness
in various forms. In this reading, the roots of stagnant social mobility lay with the
changes initiated during the 1960s, when evil began, an idea at the heart of many
of the “cultural wars” fought over the last few decades. This notion of increasing
liberal-promulgated decadence at home was accompanied by alarmist fears of
declining US hegemony abroad. Neoconservatives have been highly successful at
representing American politics in terms of a moral contest between good and evil,
in which liberals, feminists, gays, immigrants, and environmentalists are equated
with economic decline and moral decadence. Republicans have long popularized a
mythologized view of a “liberal elite” intent of pushing a politically correct agenda
down the throats of the innocent and young, whether through the corrupting values
of the allegedly “liberal media” (including Hollywood) or the poisonous lectures
of college professors. The real culprits, in this view, are those who have made
modest gains allegedly at the expense of white males. Widespread ignorance
and anti-intellectualism among a frighteningly gullible public contributed to the
propagation of this worldview. Anti-intellectualism is nothing new in American
political history (Hofstadter 1966), and is refreshed anew each generation by an
educational system carefully designed to minimize any sense of criticality. The
celebration of ignorance as a virtue among conservatives, for example, is evident
in the growth of “tea party” movements following Obama’s election, who soundly
denounce government spending so long as it doesn’t affect them.
The rhetoric of the radical neoconservative revolution admirably served the
interests of neoliberalism, obfuscating the economic agenda of conservatives
behind the rhetoric of family values, individualism, religiosity, and the ideology
of the free market. In essence, the Republican Party has used a faux populism to
disguise its deeply elitist politics, which are designed to assist the ruling class in
extracting an ever-larger share of the country’s surplus value. Thus, under the
administrations of Ronald Reagan and George W. Bush, income inequality grew

markedly, in part due to deregulation and tax cuts on unearned income, which is
the primary source for the well-to-do. In this context, what voters expect and what
they receive are often two different things. As Frank (2004, 7) puts it:
Vote to stop abortion; receive a rollback in capital gains taxes. Vote to make our
country strong again; receive deindustrialization;. Vote to screw those politically
correct college professors; receive electricity deregulation. Vote to get government
off our backs; receive conglomeration and monopoly everywhere from media
to meat-packing. Vote to stand tall against terrorists; receive Social Security
privatization. Vote to strike a blow against elitism; receive a social order in which
wealth is more concentrated than ever before in our lifetimes, in which workers
have been stripped of power and CEOs are rewarded in a manner beyond imagining.
Essentially, such a strategy allows Republican candidates such as McCain to

claim to speak on behalf of the middle class without offending their big business
constituency. Many middle class voters, especially whites, are led enthusiastically
to vote against their own economic interests, leading to what Frank labels a
“French Revolution in reverse.” Veterans, for example, eagerly supported McCain
in 2008 even as the GOP voted to cut their benefits. It is important to note that the
neoliberal agenda that has blamed government and liberals as the source of the
country’s problems has, with admirable efficiency, made great strides in rolling
back many of the social and policy gains made during the 1960s and the Great
Society, is now targeting initiatives launched during the New Deal (e.g., Social
Security), and appears eagerly intent on taking the country back to a nineteenth
century form of laissez faire and social Darwinism.
Ethnicity/Race: The Fault Line of American Politics
The influence of race or ethnicity on American voting patterns has been the subject
of considerable scrutiny (Leighley and Vedlitz 1999; Hutchings and Valentino
2004), although its effects are difficult to disentangle from those of class and other
bases of identity. Although there are many nuances to this story, in essence, since
the 1970s Republican supporters, especially the most vehement ones, tend to be
disproportionately white, whereas Democrats have traditionally attracted support
from ethnic minorities. This pattern reflects, in no small part, the success of the
GOP in capitalizing on white resentment against minorities for social and economic
gains (often exaggerated in the minds of conservatives) that ostensibly come at
the expense of working class whites. The GOP’s famous “Southern Strategy”
initiated in the 1970s, for example, led to widespread white abandonment of the
Democrats in the South and the entrenchment of GOP politicians in what was
once solidly Democratic territory (Black and Black 2002). By diverting white
resentment toward minorities, the fundamental role of race and ethnicity has been
to obfuscate the role of class. As Huckfeldt and Kohfeld (1989, 12) note, “race
served to splinter the Democratic coalition” because the federal policies initiated
during the Civil Rights era provoked “[r]acial hostility, particularly on the part
of lower-status whites.” However, the mounting conservativism of American
whites has, to some extent, been offset by the growing minority populations, now
roughly 30 percent of the population, which tend Democratic. The rapid growth
of Latinos, for example, who now form the country’s largest ethnic minority, has
problematized the traditional black-white binary (Segura and Rodriguez 2006).
Ethnicity also played an important role in the 2008 presidential election
(Ansolabehere and Stewart 2009; Mas and Moretti 2009; Pasek et al. 2009; Philpot
et al. 2009). Obama’s racial composition obviously played a hugely important
role. Widely touted as the first “African-American presidential candidate,” Obama
is in fact of mixed race (African father, white American mother), but in the racial
discourses of American politics being of mixed race is tantamount to being equated
with “black,” a nomenclature reminiscent of the old racial stratification system
found in the antebellum South. Race figured prominently in the heated Democratic
primary contest between Obama and Hillary Clinton, who enjoyed more support
among whites. Despite his attempts to appear “post-racially” transcendent on
this issue, Obama suffered repeated and blatantly false conservative allegations
that he was a Muslim, not born in the US, or was a closet Marxist, all of which
reflected a deep racism on the part of a large chunk of the electorate. Obama’s
ethnicity, as well as his foreign-sounding name, helped Republican efforts to
demonize otherness, albeit unsuccessfully, fueling latent white xenophobia and
concerns about their declining proportion of the population. More subtly, even if
they are not explicit racists, most whites exhibit a preference for white candidates
(Bafumi and Herron 2009). Nonetheless, despite such obstacles, and due in part
to widespread disgust with the Bush administration and a collapsing economy,
Obama’s campaign cobbled together a successful coalition of ethnic minorities
and progressive whites, including Democrats, many independents, and even a few
“Obamicans” (Republicans who voted for Obama).
While Obama’s election was undoubtedly a historical achievement in the
struggle for racial equality, the 2008 presidential election lent support to the notion
that most backers of Democrat candidates are minorities and most Republican
supporters are found in predominantly white regions. Table 8.2, for example,
summarizes exit polls on the eve of the election, indicating that whites preferred
McCain by a margin of 55 to 43 percent, whereas Obama enjoyed overwhelming
popularity among blacks (95 percent), who were electrified by his candidacy,
and solid majorities among Latinos and Asian-Americans. Given the numerical
dominance of whites among the electorate, such a pattern would seem to indicate
an inevitable victory for McCain. However, within the complex dynamics of the
electoral college and its winner-take-all rule for amassing electoral votes, white
support for McCain was largely rendered ineffectual, in part because it was so
heavily concentrated in the South. In this sense, the Southern Strategy may have
worked against Republicans. As Table 8.3 reveals, while minority-dominant
Table 8.2 Exit Polls of Voters by Ethnicity on Eve of 2008 Presidential

Election
Ethnicity % of the Electorate % for Obama % for McCain

White 74 43 55
Black 13 95 1
Latino 9 67 31
Asian 2 62 35
Source: New York Times. 2008. November 5. National Exit Polls Table. Available at: http://
elections.nytimes.com/2008/results/president/national-exit-polls.html.
Table 8.3 Votes in 2010 Presidential Election by White Percent of County

Population, 2008
% Population Votes for % of Votes for % of

White Obama Votes McCain Votes
> 90 31,783,205 53.2 27,993,460 46.8
80.0–89.9 11,260,897 52.1 10,365,811 47.9
70.0–79.9 8,056,834 54.5 6,736,108 45.5
60.0–69.9 7,142,757 52.6 6,442,390 47.4
50.0–59.9 5,306,683 53.0 4,697,883 47.0
< 50.0 5,928,544 61.6 3,699,105 38.4
Total 69,478,920 53.7 59,934,757 46.3
counties voted heavily for Obama, in majority-white counties Obama still amassed
a slim majority of votes. In short, many whites did indeed support Obama, and
there are important geographic patterns that are hidden by these national data.
Religion and American Presidential Politics
The role of organized religion in American politics is too important to ignore.

From popular beliefs of the electorate and politicians to how issues are often
discursively framed, religion saturates many political debates. From the country’s
founding, religion and politics were intertwined (Thompson 1988), as exemplified
historically by the anti-slavery movement, agrarian socialism, Prohibition, and the
civil rights struggle. In contrast to most of the industrialized world, such as Europe,
in which secularism is more common than religiosity, the US exhibits significantly
higher levels of religious adherence and participation (Hackett 2003). Religion’s
political influence reflects the unique circumstances of the United States – as
claimed by the American exceptionalism thesis – including the peculiarly high
degrees of religiosity found in the US.
A large political science literature has investigated how religion has played
a critical role in shaping the political loyalties of vast swaths of the electorate
(e.g., Green et al. 2002; Leege et al. 2002; Wald and Calhoun-Brown 2007).
Religion enters electoral politics through a variety of channels, including how it
shapes the ideological propensities of the electorate and candidates as well as the
inclination to vote itself. For example, large majorities of Americans indicate they
would not vote for a candidate “who did not believe in God’ (Servin-Gonzalez
and Torres-Reyna 1999). Religious attitudes are also important to politicians,
including presidential candidates and members of Congress, who “are not empty
religious vessels, but indeed bring deeply held religious conceptions to the world”
(Hertzke 1988, 829). Often Congressional religious preferences and voting records
reflect the relative degree of religiosity in their respective districts. For example,
generally, significant local membership in conservative Protestant churches is
inversely related to liberal Congressional voting records (Green and Guth 1991).
The relations between theological and political ideology are complex, however,
and should not be oversimplified. While some maintain that religion inculcates
a passive acceptance of the prevailing social order, others point to the role of
religion in progressive trends such as the civil rights movement.
The rise of politically conservative Protestants is perhaps the most visible
recent face of this phenomenon. In the late twentieth century, evangelical
conservative religious activists surfaced as a highly influential political force
(Guth and Green 1991; Wilcox 1992). In 1976, for example, Jimmy Carter ran as
a “born again” Christian, and religion was openly embraced by Ronald Reagan
and Bill Clinton. In the 1980s, leaders of the Republican Party, recognizing the
size, militancy and potential power of this electorate, began to court such voters
aggressively, leading them to become a pillar in a highly successful political
alliance with business interests. The Moral Majority and religious leaders
such as Jerry Falwell and Pat Robertson became vocal and influential voices
on behalf of conservative Protestantism in particular and the Republican Party
more generally. Today, many GOP state political platforms affirm the status of
the US as a Christian nation, label the separation of church and state a myth,
demand abstinence instead of sex education, reject evolution, and otherwise
mirror the agenda of religious conservatives (Phillips 2006). Thus, Heinemann
(1998) observes, the predominant impact of contemporary religion in the US has
been to inspire, motivate, and produce conservatives. As Jacobs (2006) notes, the
marriage of political and religious conservatives occurred precisely at the historic
moment in which American economic and political hegemony was contested
internationally, a moment in which multiculturalism and the postmodern relativity
of moral values achieved widespread popularity. Indeed, the rise of the Religious
Right fueled not only a series of “culture wars,” but also a significant gender
and ethnic gap in which white males increasingly turned toward the Republican
Party whereas women and ethnic minorities remained predominantly Democratic
in their voting behavior. Issues that particularly animate conservative religious
voters include prayer in the schools, reproductive rights and contraception,
sex education, tolerance of homosexuality, the teaching of evolution, stem cell

research, and a variety of other important topics often loosely lumped under the
phrase “moral values.” As a result, the “religion gap” between Republicans and
Democrats has become one of the most striking and important dimensions of
US politics, rivaling or exceeding the significance of class, race, or gender in
determining local and national electoral turnout and outcomes.
Religion played an important role in the 2008 presidential election (Espinosa
2008). President George W. Bush, a born-again Methodist, personified conservative
Protestantism, with numerous public references to the role of God in his personal and
public life (Mansfield 2003; Aikman 2004). He was arguably the most visibly religious
president in decades, perhaps in national history. However, given widespread voter
revulsion at the Bush presidency, the effects of religion played out differently among
different denominations. For example, counties that were heavily Catholic tended
to favour Obama by wide margins (Table 8.4), in part due to his popularity among
Hispanics, whereas conversely McCain won substantial victories in the counties in
which Baptists comprised 30 percent or more of county adherents (Table 8.5).
Table 8.4 Votes in 2008 Presidential Election by County Percent of Total

Adherents who are Catholic
% Adherents Votes for % of Votes for % of
Catholic Obama Votes McCain Votes
> 70.0 3,447,288 61.1 2,196,309 38.9
60.0–69.9 5,675,259 58.0 4,116,250 42.0
50.0–59.9 10,973,988 59.4 7,487,087 40.6
40.0–49.9 7,494,705 56.5 5,778,437 43.5
30.0–39.9 8,348,764 55.6 6,674,328 44.4
< 30.0 33,538,917 49.9 33,682,344 50.1
Total 69,478,920 53.7 59,934,757 46.3
Table 8.5 Votes in 2008 Presidential Election by County Percent of Total

Adherents who are Baptist
% Adherents Votes for % of Votes for % of
Baptist Obama Votes McCain Votes
> 70.0 941,804 46.6 1,078,921 53.4
60.0–69.9 2,788,220 43.8 3,574,970 56.2
50.0–59.9 3,355,889 42.3 4,574,288 57.7
40.0–49.9 2,851,024 43.8 3,650,743 56.2
30.0–39.9 4,815,032 48.8 5,048,211 51.2
< 30.0 54,726,951 56.6 42,007,624 43.4
Total 69,478,920 56.6 59,934,757 46.3
Table 8.6 Simple Correlation Coefficients between Percent of Votes for

Obama and McCain and Income, Ethnicity, and Religious
Denomination
% Voting for Obama % Voting for McCain

Household Income 0.26 -0.25
% White -0.31 0.28
% Black 0.27 -0.24
% Latino 0.06 -0.05
% Asian 0.25 -0.25
% Catholic 0.36 -0.37
% Baptist -0.41 0.43
% Methodist -0.24 0.25
% Pentacostal -0.13 0.12
% Latter-day Saints -0.15 0.13
% Jewish 0.32 -0.32
% Muslim 0.18 -0.17
The Independent Effects of Class, Race, and Religion: Empirical Assessments
The simple relationships between the variables discussed above and votes for
Obama and McCain can initially be approximated using simple correlation
analysis (Table 8.6; given the large sample size, 3,138 counties, all of these
coefficients are statistically significant). These numbers confirm many of the
observations made above. Thus, average household income was strongly and
positively correlated with support for Obama (.26), and negatively so with votes
for McCain (-.25). Percentage white was strongly associated with opposition
to Obama (-.31) and support for McCain (.28) while percentage black and
Asian-American exhibited exactly the reverse trend; Latinos were not strongly
associated with either candidate. Among religious denominations, Catholics,
Jews, and Muslims supported Obama and opposed McCain, while Baptists,
Methodists, Pentacostalists, amd Latter-day Saints (Mormons) demonstrated the
reverse pattern. Indeed, the correlations for the two candidates are largely mirror
images of one another.
However, simple correlations disguise as much as they reveal. First, the effects
of class, race, and religion on electoral outcomes are hopelessly intertwined with
one another and empirically difficult to segregate. Second, the impacts of even
one variable are likely to vary spatially: For example, household income may have
very different effects on the West Coast compared to the South. For these reasons,
this chapter analyzed votes for Obama using a relatively recent methodology,
geographically weighted regression (GWR), which allows not only independent
effects of variables to be specified, but also allows for spatial variations in the
influence of one variable (Brundson et al. 1998; Fotheringham et al. 2002).
Standard regression analysis assume spatial stationarity, i.e., that the relationship
under analysis is constant over space. In contrast, GWR permits relationships to
vary spatially within a study region (or spatial non-stationarity). Each observation
receives a weight that is a function of its proximity to the center of a given window
(in this case, 20 counties). While standard regression analysis finds correlations
between two variables across an entire data set, GWR yields correlations across
only a geographic subset of the entire data set and then estimates how its statistical
significance varies across the region under study.
Including variables derived from the discussion above (but limiting religious
denominations to only two), the multivariate linear GWR model utilized in this
analysis is
poi = βi + βhi + βwi + βbi + βli + βai + βci + βbi + ε,
where
poi = % voting for Obama in each county i;

βi = constant
βhi = county household income as % of national median;
βwi = percent white population in county i;
βbi = percent black population in county i;
βli = percent Latino population in county i;
βai = percent Asian population in county i;
βci = % of adherents who are Catholic in county i;
βbi = % of adherents who are Baptist in county i;
ε = error term.1
Such an approach allows for a view of the geographically uneven independent

effects of each variable controlling for the influence of the others. The results of
the analysis for each variable are summarized in maps of the partial regression
coefficients, and, when necessary, the results of a simple GWR bivariate model
without controls as well as the results of the broader model with controls.
GWR Class Effects
Although it is an imperfect measure of class, income (in this case, county average
income as a percent of the national median) is an important determinant of voter
preferences. The spatial distribution of household incomes (Figure 8.3) reflect the
uneven development that has long formed a centerpiece of the social and economic
geography of the United States, with the highest levels found in the northeastern
seaboard, the West Coast, much of the old Manufacturing Belt, and metropolitan
Figure 8.3 Household Income as a Percent of the National Average
Figure 8.4 Results of Bivariate GWR of Support for Obama and Average
Household Income without Controls for Ethnicity and Religion
Figure 8.5 Independent Effects of Household Income on Support for

Obama, Controlling for Ethnicity and Religion
areas generally, and lower levels distributed across rural areas, particularly in the
South. A simple bivariate GWR analysis of income and support for Obama yields
surprising results (Figure 8.4): higher income households demonstrated the most
support for him in traditionally conservative regions such as the intermontane West,
notably Utah and Idaho, the West Coast, and pockets widely distributed throughout
the central and eastern thirds of the country. However, the bivariate relationship may
be heavily affected by other variables. Once the controls for ethnicity and religion
are introduced, the GWR nonetheless confirms that Obama’s support tended to rise
steadily with household income in the West more than most parts of the country
(Figure 8.5). In contrast to their counterparts elsewhere in the country, it would
seem that upper income Western voters are relatively more Democratic in their
inclinations, even in regions dominated by conservative religious denominations
such as the Church of Latter-day Saints (Mormons). Such a result may reflect the
particular cultural heritage of the Western United States, in which ideologies such as
individualism and the cult of entrepreneurship are particularly strong (Etulain 1980).
GWR Ethnicity Effects
Although the voting impacts of ethnicity have been well demonstrated among
ethnic groups, very little insight has been shed on their effects within a given
ethnic group. GWR allows for such an analysis. Whites, who comprised roughly
three-quarters of the electorate, demonstrated spatially uneven support for Obama.
Thus, without the controls of income or religion, white support was highest in
northern California and the Pacific Northwest (including southern Idaho!) and
parts of Virginia near Washington, DC (Figure 8.6). Once the effects of income
and religion are removed, however, this pattern changes significantly, with white
support highest in the West Coast and Nevada, the Mississippi Delta, Colorado,
southern New England, and the greater Washington, DC region (Figure 8.7). Such
a pattern points to two possible forces contributing to a liberalized white electorate:
the presence of a large minority population and local economies centered either on
the public sector or professionalized producer services, as with Florida’s (2004)
celebrated “creative class.”
Black support for Obama, which was almost universal, also demonstrated
a geography in its own right. Without controls for income and religion, blacks
voted for Obama most heavily outside of the South (Figure 8.8), particularly in
the Great Plains and Rocky Mountains. Similarly, Latino support for Obama was
heaviest outside of the Southwest, notably in parts of the South with large African-
American populations, Illinois and Wisconsin, and New England (Figure 8.9).
Such geographies, intimately tied up with the politics of identity, hint that in both
cases minority support was accentuated in regions where they were less common
numerically.
Figure 8.6 Results of Bivariate GWR of Support for Obama and White %
of County Population without Controls for Income and Religion
Figure 8.7 Independent Effects of White Support for Obama, Controlling

for Other Ethnicities and Religion
Figure 8.8 Results of Bivariate GWR of Support for Obama and Black %
Figure 8.9 Results of Bivariate GWR of Support for Obama and Latino %
GWR Religion Effects
Data on the distribution of religious denominations by county in 2000 were

obtained from the census published by the Glenmary Research Center (2002).
While the Glenmary data are widely recognized as the best available, they are
not without flaws. Some denominations such as the Jehovah’s Witnesses refused
to participate. Also, the Glenmary database does not identify the religion of those
who live in one county and worship in another. Because the data on religious
adherents is geolocated by place of worship (which may be in a neighboring
county) rather than place of residence, as in the case of voting data, there is the
distinct possibility of a spatial mismatch between religious affiliation and voting
patterns, although for purposes of studying the dynamics of the electoral college,
which is based on states, not counties, this issue is irrelevant.
Catholics, for example, who comprise 37 percent of the electorate, form
significant majorities in most of the Northeast, Wisconsin, and much of the
Southwest, particularly California, New Mexico, and southern Texas (Figure
8.10). All of these regions voted for Obama in the general election. Catholics are an
ethnically and politically diverse lot, of course, ranging from conservative Cuban-
American voters in southern Florida to more liberally minded Puerto Ricans in
Florida and New York, Italians and Irish in the Northeast, and Mexican-Americans
Figure 8.10 Catholics as % of Total Adherents
in the Southwest. GWR reveals that once the effects of income and ethnicity are
removed statistically, Catholic support for Obama consisted of a hodgepodge of
pockets of greater and lesser enthusiasm (Figure 8.11). Catholics in the South
tended to vote him in greater proportions than elsewhere, although Catholics
in southern New Mexico and Utah likewise were also supportive. Surprisingly,
Catholic support in strongholds such as New England and the northern Midwest
was relatively tepid.
Protestant denominations are quite another story. Baptists (14.3 percent of
voters), who comprise the largest Protestant group, generally voted heavily
in favor the Republican John McCain. Concentrated heavily in the South,
particularly interior counties removed from the coasts, Southern Baptists were
located in states that all voted against Obama by large margins (Figure 8.12),
although Florida, North Carolina, and Virginia did narrowly vote Democratic.
Baptist support for Obama, such as it was lay for the most part outside of
the South (Figure 8.13), including islands of support around the New York
metropolitan region, Chicago, the northern Great Plains, the borderlands of
Texas and New Mexico with Mexico, and the Pacific Northwest. Such a pattern
reflects, perhaps, the willingness to consider candidates not actively supported
by one’s denomination when living as a minority outside of the region of
denominational hegemony.
Figure 8.11 Independent Effects of Catholic Support for Obama, Controlling

for Income and Ethnicity
Figure 8.12 Baptists as % of Total Adherents

Figure 8.13 Independent Effects of Baptist Support for Obama, Controlling

for Income and Ethnicity
Concluding Thoughts
The presidential contest of 2008 was a watershed moment in several respects,

including a rejection of the policies of the George W. Bush administration and
the election of the country’s first African-American candidate. Obama’s victory
reflected a constellation of voters of many kinds with varying reasons. In this
race, as always, class, ethnicity, and religion all had profound effects on the
American electoral landscape. (No doubt so too did other markers such as age,
gender, and sexuality). Contrary to much received opinion, among income
groups, the Democrats’ greatest strengths lay among upper income voters rather
than poorer households, who generally supported McCain. Whites, at least
outside of the South, tended to be split between the two parties, whereas blacks,
Latinos, and Asians tended to support Obama whole heartedly. Using the two
largest denominations as a marker of religious influence, Catholics tended to vote
for Obama in large proportions, whereas Baptists were strongly opposed to his
candidacy.
What is critical to note here is that the effects of these variables are not
spatially uniform: contrary to the aspatial wonderland in which many economists
and political scientists operate, there are enormous geographical variations in the
intensity of the relationship of each category. Geographically weighted regression
offers a valuable means with which to interrogate such patterns in a statistically

rigorous manner.
The results of the GWR model deployed here hint at one consistent, albeit
surprising, finding: support for Obama within each economic, ethnic, and
religious group tended to be highest in regions in which supporters were a
relative minority. In most cases, support for Obama within each income, ethnic,
or religious group was greatest when members of that group were relatively
uncommon. Whites, the majority in most regions, did not conform to this
generalization. However, black support for Obama was greatest outside of the
South, and Latino support highest in regions removed from the Southwest or
Florida. Some Baptists did indeed support Obama, but largely in areas outside
of the South. Higher levels of household income tended to be more closely
associated with votes for Obama in the Western US (parts of which are quite
wealthy, belying the observation made above), including the Rocky Mountains,
Nevada, and California. At the least, these results indicate the need to avoid
simplistic, aspatial generalizations that essentialize groups and their political
views.
More deeply, such a trend points to the politics of marginality that played
an important role in electing Obama. If identity is largely constructed and
maintained through difference (Sibley 1995; Natter and Jones 1997; Valentine
2008), then voters belonging to particular income, ethnic, or religious categories
who live in regions in which they find themselves minorities may well have
been more sympathetic to a candidate who was a member of an ethnic minority
himself. Perhaps supporters of Obama who were not surrounded by others of
similar wealth, culture, and beliefs were more willing to elect a candidate whose
own background was so resoundingly non-traditional. In short, a geographical
analysis indicates that the lines of difference so important not only to identity
but to political attitudes are highly susceptible to local and regional variations.
Acknowledgement
The author thanks Aaron Gilbreath for his assistance with the GWR.
References
Aikman, D. 2004. A Man of Faith: The Spiritual Journey of George W. Bush. New
York: W Publishing Group.
Aistrup, J. 1996. The Southern Strategy Revisited: Republican Top-Down
Advancement in the South. Lexington, KY: University of Kentucky Press.
Ansolabehere, S. and C. Stewart. 2009. “State of the Nation: How Post-Racial was
Obama’s Victory?” Boston Review 34(1): January/February.
Bafumi, and Herron. 2009. “Prejudice, Black Threat, and Racist Voters in the 2008
Presidential Election.” Journal of Political Marketing 8(4): 334–48.
Black, E. and M. Black. 2002. The Rise of Southern Republicans. Cambridge,
MA: Harvard University Press.
Brunsdon, C., S. Fotheringham, and M. Charlton. 1998. “Geographically Weighted
Regression.” Journal of the Royal Statistical Society 47(3): 431–43.
Espinosa, G. 2008. Religion, Race, and the American Presidency. Lanham, MD:
Rowman and Littlefield.
Fowler, R., A. Hertzke, L. Olson, and K. den Dulk. 2004. Religion and Politics in
America: Faith, Culture, and Strategic Choices. Boulder, CO: Westview Press.
Fotheringham, A., C. Brunsdon, and M. Charlton, M. 2002. Geographically
Weighted Regression: The Analysis of Spatially Varying Relationships.
Chichester: Wiley.
Etulain, R. 1980. “Frontier, Region and Myth: Changing Interpretations of Western
American Culture.” Journal of American Culture 3(2): 268–84.
Florida, R. 2004. The Rise of the Creative Class. New York: Basic Books.
Frank, T. 2004. What’s the Matter with Kansas? How Conservatives Won the
Heart of America. New York: Holt.
Glenmary Research Center. 2002. Religious Congregations and Membership in
the United States. Atlanta: Glenmary Research Center.
Green, D., B. Palmquist, and E. Schickler. 2002. Partisan Hearts and Minds. New
Haven, CT: Yale University Press.
Green, J. and J. Guth. 1991. “Religion, Representatives, and Roll Calls.”
Legislative Studies Quarterly 16: 571–84.
Green, J., M. Rozell, and C. Wilcox (eds) 2006. The Values Campaign? The
Christian Right and the 2004 Elections. Washington, DC: Georgetown
University Press.
Guth, J. and J. Green (eds) 1991. The Bible and the Ballot Box: Religion and
Politics in the 1988 Election. Boulder, CO: Westview Press.
Gutterman, D. 2005. Prophetic Politics: Christian Social Movements and
American Democracy. Ithaca, NY: Cornell University Press.
Gutterman, D. and A. Murphy (eds) 2006. Religion, Politics, and American
Identity: New Directions, New Controversies. Lanham, MD: Lexington Books.
Hackett, D. 2003. Religion and American Culture. London: Routledge.
Heilemann, J. and M. Halperin. 2010. Game Change: Obama and the Clintons,
McCain and Palin, and the Race of a Lifetime. New York: Harper.
Heineman, K. 1998. God is a Conservative: Religion, Politics, and Morality in
Contemporary America. New York: New York University Press.
Hertzke, A. 1988. “American Religion and Politics: A Review Essay.” The Western
Political Quarterly 41: 825–38.
Hofstadter, R. 1966. Anti-intellectualism in American Life. New York: Vintage.
Huckfeldt, R., and C. Kohfeld. 1989. Race and the Decline of Class in American
Politics. Urbana, IL: University of Illinois Press.
Hutchings, V. and N. Valentino. 2004. “The Centrality of Race in American

Politics.” Annual Review of Political Science 7: 383–408.
Jacobs, A. 2006. “The New Right, Fundamentalism, and Nationalism in
Postmodern America: A Marriage of Heat and Passion.” Social Compass 53:
357–66.
Jelen, T. 2006. “Religion and Politics in the United States: Persistence, Limitations,
and the Prophetic Voice.” Social Compass 53: 329–43.
Kohut, A., J. Green, S. Keeter, and R. Toth. 2000. The Diminishing Divide:
Religion’s Changing Role in American Politics. Washington, DC: Brookings
Institute.
Leege, D., K. Wald, B. Krueger, and P. Mueller. 2002. The Politics of Cultural
Differences: Social Change and Voter Mobilization Strategies in the Post-New
Deal Period. Princeton, NJ: Princeton University Press.
Leighley, J. and A. Vedlitz. 1999. “Race, ethnicity, and political participation:
Competing Models and Contrasting Explanations.” Journal of Politics 61(4):
1092–114.
Mansfield, S. 2003. The Faith of George W. Bush. New York: Charisma House.
Mas, A. and E. Moretti. 2009. “Racial Bias in the 2008 Election.” American
Economic Review 99(2): 323–9.
Natter, W., and J.P. Jones. 1997. “Identity, Space, and Other Uncertainties.” In
G. Benko and U. Strohmayer (eds) Space and Social Theory: Geographical
Interpretations of Postmodernism. Oxford: Blackwell, 141–61.
Olson, L. 2002. “Toward a Contextual Appreciation of Religion and Politics.”
In E. Blumhofer (ed.) Religion, Politics, and the American Experience:
Reflections on Religion and American Public Life. Tuscaloosa: University of
Alabama Press.
Pasek, J., A. Tahk, Y. Lelkes, J. Krosnick, B. Payne, O. Akhtar, and T. Tompson.
2009. “Determinants of Turnout and Candidate Choice in the 2008 US
Presidential Election: Illuminating the Impact of Racial Prejudice and Other
Considerations.” Public Opinion Quarterly 73(5): 943–94.
Philpot, T., D. Shaw, and E. McGowen. 2009. “Winning the Race: Black Voter
Turnout in the 2008 Presidential Election.” Public Opinion Quarterly 73(5):
995–1022.
Phillips, K. 2006. American Theocracy. New York: Viking.
Segura, G. and H. Rodriguez. 2006. “Comparative Ethnic Politics in the United
States: Beyond Black and White.” Annual Review of Political Science 9:
375–95
Servin-Gonzalez, M. and O. Torres-Reyna. 1999. “Trends: Religion and Politics.”
The Public Opinion Quarterly 63: 592–621.
Sibley, D. 1995. Geographies of Exclusion. London: Routledge.
Thompson, K. 1988. “The Religious Transformation of Politics and the Political
Transformation of Religion.” The Review of Politics 50: 545–60.
Valentine, G. 2008. “Living with Difference: Reflections on Geographies of
Encounter.” Progress in Human Geography 32(3): 323–37.
Wald, K. and A. Calhoun-Brown. 2007. Religion and Politics in the United States.
5th edition. Lanham, MD: Rowman and Littlefield.
Wilcox, C. 1992. God’s Warriors: The Christian Right in Twentieth Century
America. Baltimore, MD: Johns Hopkins University Press.
Chapter 9
The Emerging Geography of How
Americans Cast their Ballots
Toby Moore1
The first decade of the 2000s witnessed rapid change in how elections are run in the
United States. Spurred on by the problems exposed by the Florida recount process
after the 2000 presidential election, lawmakers, advocates and the public pressed
for reform in a wide range of voting laws and practices. Bush v. Gore triggered
a renewed sense that American election administration had become moribund
and lax, in need of significant government investment and at least some measure
of increased oversight and scrutiny. An initial focus was on voting technology,
since the notorious “hanging chads” were the most visible sign of election failure
in 2000, but other issues, from voter identification to voter registration to early
voting, would also see new dynamism.
The landscape was actually already in some flux when Bush v. Gore hit, as
a series of traditional concerns was prompting experimentation in the states and
at the local level. Participation rates had long been a focus of political scientists
and progressive reformers, who saw them as a measure of the health of American
democracy, and new methods were adopted to make voting easier. Unsettled
questions over the fairness of American elections, particularly to ethnic and racial
minorities, and the continued salience of the Voting Rights Act of 1965 sustained
efforts by the federal government to intervene in what is traditionally reserved for
the states and local governments. Allegations of electoral fraud persisted, hotly
contested by the political parties and the reform community. Electronic voting
machines had already begun to replace paper ballots in some jurisdictions before
Florida contemplated its hanging chads. The spark from the Florida recount
process, therefore, hit dry tinder.
Reformers, however, confronted a country that stubbornly guards local
prerogatives in determining how elections are run. The US famously lacks
a constitutional right to vote, and the task of administering elections is left to
the states, which in turn have historically devolved much responsibility to local
jurisdictions. Filtered through the thousands of local election jurisdictions, then,
the extreme federalism of US election administration was virtually guaranteed to
1 The author is a project director for the non-partisan, non-profit research institute
RTI International. Generous support for this chapter was provided by RTI through a
professional development award.
produce strong geographical variability as America grappled with how to improve

its election methods.
This chapter describes the emerging spatial patterns of how Americans cast
their ballots. It is “emerging” both because different changes are taking occurring
in different places, and because we are only now collecting the data that makes
seeing these patterns possible. This variance across space and between states
and regions, it is argued, is of increasing importance to American democracy
and the political system; the divergence appears to be growing, with potentially
far-reaching effects. Investigation of this variation at the national level has only
recently become possible due to newly-available data sources, including the
federally-sponsored survey of election jurisdictions that forms the basis for
this chapter.
While electoral geography has long been of interest to geographers, it has
focused largely on the choices voters make or on the way electoral district
boundaries are drawn. Little work has been done on the myriad ways that
modern voters go about casting their ballots. Election administration has not
generally been part of political geography, despite the rich geographic data
available, to the detriment both of the discipline and American elections.
Drawing largely from the results of the 2008 Election Administration and Voting
Survey, sponsored by the US Election Assistance Commission and directed by
the author, this chapter aims to lay the groundwork for a better engagement
between geography and the question of how elections are run in the United
States. Finally, the chapter discusses possible future directions for research, at
scales from the local to the international, and ties the field to larger questions
of democracy and equity.2
Elections in Electoral Geography
Traditionally, electoral geography has been divided into three branches (Taylor
and Johnston, 1979). The first is the pattern of voters’ choices, aggregated to a
certain spatial scale and analyzed through mapping and multivariate analyses.
Often socio-economic or other variables are included in the analysis in the search
for understanding the spatial patterns. The second, “geographical influences on
elections”, was the subject of considerable debate on so-called “context effects”
and their importance and value as a subject of inquiry (Agnew, 1996; King, 1996).
This branch saw geographers defend the impact of local circumstances on voter
choices, with mixed success. Finally, the “geography of representation” is focused
on the drawing of district boundaries and the biases that develop from that process.
2 Unless stated otherwise, all the data used in this article are from the EAC survey
and publicly available through the EAC’s website at www.eac.gov. Summary tables with
statewide aggregations are available in the reports based on the survey issued by the EAC
and also available at the same Web address.
The Emerging Geography of How Americans Cast their Ballots 159
This has been a particular interest in the United States, where little is done to
constrain partisan gerrymandering, and where racial gerrymandering (for both the
benefit and harm of minority voters) has a long tradition, but has proved important
in other countries as well. Border delineation is an important part of the technical
assistance provided to emerging democracies, who must often build constituencies
from scratch.
Within these three branches of electoral geography, little has been written
about the geography of the myriad ways in which people differ in the way they
cast their ballots. The “geography of elections,” to date, has referred almost
exclusively either the spatial variation in the choices that voters make between
candidates, parties or issues, or the division of space into electoral districts.
A “…geography of election administration”, by contrast, would focus on the
way laws and procedures impact the molding of the electorate and their role
in promoting or restricting democratic outcomes, and the variance of such
laws and procedures across the thousands of election jurisdictions in the
United States. A broad array of topics can be studied in such a geography of
election administration; the intention here is to suggest some topics that other
geographers might find of interest for their own work, as well as providing
a high-level description of some of the ways election laws and practices are
diverging across the country. The topics discussed below, as a starting point for
this geography of election administration, include:
• when voters cast their ballots (i.e., on election day or before, or even
afterwards);
• where they cast their ballots (i.e., polling places or remotely)
• how they cast their ballots (i.e., through the mail or in person); and,
• what barriers they face (i.e., ID laws, felony disenfranchisement or
registration requirements);
• what technology they use to record their choices (i.e., electronic voting
machines or paper ballots).
Such issues are of growing interest to political scientists, legal scholars and
election professionals. Much of the literature they have produced focuses on the
impact of specific election reforms either on participation as a whole, or on the
participation of specific parts of the electorate. For example, the move by some
states during the 2000s to impose new identification requirements on voters
(discussed below) slowly gave rise to a small body of scholarly writings over the
second half of the decade. Not surprisingly, it lacked consensus and mirrored the
sharp political divide over the issue, but nonetheless was able to provide policy
makers with some of the basic research needed to consider and debate ID laws.
Much of the work done on the impact of voter ID laws was done by scholars or
advocates with an interest in one side of the debate or the other, an indication of
the shortage in election studies of independent social science.
The 2008 Election Administration and Voting Survey
Under the mandate given it by the 2002 Help America Vote Act (HAVA), the US
Election Assistance Commission began collecting election data following the
2004 election. Initially called the Election Day Survey, and later the Election
Administration and Voting Survey (EAVS), the project is somewhat misnamed;
it is less a survey than an accounting process. Across a number of election
administration topics, the EAVS attempts to collect data at the local level,
generally the county or parish, but in some instances the town or township. The
survey is administered to the individual states, territories and the District of
Columbia, which either generate responses from central databases or collect the
data from their own sub-jurisdictions. Some of the data requested is required by
federal law (to measure compliance with the federal National Voter Registration
Act (NVRA) and the Uniformed and Overseas Citizen Absentee Voting Act
(UOCAVA)), but much of the data requested is voluntary on the part of the states
and local governments. The data collects forms the basis for a series of reports
issued by the EAC.
The EAC’s work coincided with a push by academics and reformers to develop
better data collection on election administration, and took place during what the
journalist James Fallows has called “the decade of big data” across a variety of
public policy fields from education to health to finance (Fallows, 2010). The Pew
Charitable Trust’s Make Voting Work project promoted a “Data for Democracy”
effort, hosting conferences and issuing reports aimed at increasing the quality and
quantity of election data available. A group of scholars, mostly law professors,
argued throughout the decade that a reliance on empirical data on election
administration was a way to overcome partisan and ideological divisions and
avoid the traditional reliance on anecdotal evidence (for example, Tokaji, 2005;
Overton, 2006; Gerken 2009).
Starting with a limited effort in 2004, data collection has improved with each
iteration, netting higher response rates on most data elements and providing greater
levels of disaggregated data. The 2008 EAC survey was the most complete data
collection yet. Its baseline number of jurisdictions (including states or territories
that only submitted aggregate responses; counties or parishes; and townships in
some states) was 4,517, a 45 percent increase from the 3,123 jurisdictions covered
by the 2006 survey. (The number of jurisdictions varies from year to year because
some states aggregate their township data to the county level, or submit only
statewide totals. Numbers of responses to individual items varied considerably).
Data was collected from each of the 50 states, the District of Columbia, Guam,
Puerto Rico, the Virgin Islands and American Samoa. While still incomplete, the
2008 Election Administration and Voting Survey (EAVS) suggests that future data
collections will come yet closer to providing a complete picture of how Americans
cast their ballots.
The 2008 survey was composed of two parts3: a qualitative Statutory Overview,
which sought to gather data on the laws, definitions and procedures used by the
states, and the main, quantitative survey of 52 questions (many questions had
multiple parts). The main questionnaire included questions on voter registration;
registration and voting by overseas civilians and uniformed service members;
polling place operations; voting technology; poll workers; and other election
administration topics.
Data was collected through the state or territorial election offices. Data
collection practices and capacity vary significantly across the country. Many states
maintain central voter registration and voter history databases that can be readily
queried for most of the information requested. In other states, little data is kept
at the state level, and it instead must be collected individually from each local
jurisdiction. Many of these jurisdictions are small and lack the staff or motivation
necessary to complete the survey. In addition, states vary in the ways they define
and code important election terms (such as absentee ballots). For these reasons the
level of response to individual questions on the survey varies. For some items, data
was widely available; for others, there are significant holes. Finally, states differ in
their willingness to cooperate with the EAC on the survey; some states and local
officials feel the survey unduly burdens their offices and constitute an unfunded
mandate on the part of the federal government.
Spatial Variations of Elections in the United States
Results from the 2008 EAVS show the marked differences between states and
regions in the United States in how voters cast their ballots and other aspects
of election administration. While this chapter focuses on state- and regional-
differences, the county-level data available from the EAC can be readily mined to
highlight local differences as well. In addition, the EAVS supplies data on topics
beyond those covered below.
Convenience Voting
The traditional method of voting in the United States has two main components.
The first, and most common, is voting in-person, on election day, usually at a
polling place near one’s place of residence. The second component, discussed
below, entails voting through an absentee ballot.
3 An additional data collection effort related to the survey was the collection of
precinct-level data from five states, as part of a $10 million Congressional program. This
data, not included in this analysis, is also available at the EAC’s website. The five states
participating were Illinois, Ohio, Wisconsin, Pennsylvania and Minnesota.
Figure 9.1 Percentage of Votes Cast In-person
Alongside this model have arisen a series of mechanisms intended to make

it easier for people to cast their ballots, thus saving voters time and aggravation
and encouraging participation. The actual impact on turnout of such measures
is subject to debate; the literature at present suggests that the impact of most so-
called convenience voting practices is minimal, and in some cases may even drive
down overall turnout. States continue to experiment with convenience voting
mechanisms, however, in part to save money through reduced administrative costs
(Gronke, Galanes-Rosenbaum, Miller and Toffey, 2008).
The most effective way of measuring convenience voting is to look at the rates
of traditional voting, since convenience voting comes in many forms. In 2008,
states reported that 60.2 percent of voters cast their ballots in-person on election
day (see Figure 9.1).4 In-person voting on election day remains strongest in the
Northeast and upper Midwest.
4 Not all states provided breakdowns of how their voters cast ballots; thus, another
6.1 percent of the ballots in the EAC survey were “not categorized”. Many of these were
undoubtedly in-person voters. Even within traditional balloting systems, some states and
local jurisdictions have experimented with relaxing the geographic constraints imposed by
polling places. “Vote centers”, in which any registered voter in the jurisdiction can cast a
ballot, have been implemented in some places.
Figure 9.2 Map of Early Voting
The second component of traditional voting is through an absentee ballot, cast

in a variety of means including through the mail or at an election office prior
to election day. Absentee balloting dates to the Civil War, when Union states
sought ways to allow soldiers to vote in the 1864 presidential election. Absentee
ballot laws vary significantly across the country, with some states requiring valid
excuses (such as being out of town on election day) while others allowing “no
excuse” absentee balloting. In 16 of the 49 states that reported absentee voting
figures, absentee voters made up 5.0 percent or less of the total voters. At the
other extreme, three states (Arizona, Colorado and Washington) reported absentee
voting comprised more than half of their total.
One of the more common forms of convenience voting is early voting, which
allows for the casting of ballots before election day. In 2008, 24 states reported
at least some level of early voting, with more than 17 million voters casting their
ballots before election day. Early voting is a prominent part of the election process
in Florida, Georgia, Nevada, New Mexico, North Carolina, Tennessee, and Texas;
in each state, at least 30 percent of voters were reported to have cast an early
ballot. Other states with lower but still significant numbers (more than 10 percent)
of early voters were Alaska, Arkansas, Colorado, Hawaii, Illinois, Louisiana,
North Dakota, Utah, West Virginia, and American Samoa. Early voting is most
common in the southern half of the country.
Early voting changes elections in several ways. Voters who cast their ballots
early forfeit the information that might have become available in the last days
or weeks of a campaign. The extended time that voting is open may privilege
campaigns with the resources and organizational strength to get their voters to
the polls, thus making campaign financing even more important than it already
is. Administratively, early voting may help election officials minimize the risks
of overburdening polling places on election day in high turnout elections, and cut
the time voters must wait in line to vote. Some critics of early voting contend that
casting ballots in this way dilutes the social aspects of voting, eroding an important
community event and thus diminishing its perceived importance. Evidence of the
impact of early voting on turnout has been mixed (Gronke, Galanes-Rosenbaum
and Miller, 2007).
Another convenience voting mechanism is vote-by-mail, in which entire
precincts, jurisdictions or even states do away with traditional balloting entirely,
and replace it with votes cast through the postal system. Voters are mailed a ballot,
fill it out and return it through the mail. Oregon switched to all vote-by-mail in 1998
after the passage of a public referendum, and other states have moved at least some
of their elections in that direction as well. The mode is most popular in the West:
nearly all voters in Washington now vote by mail, as do many voters in California.
(The EAC survey in 2008 conflated vote-by-mail with mail-in absentee voting; in
2010, the survey will separate the two options. Generally, vote-by-mail denotes
the replacement of in-person voting, while mail-in absentee voting augments it).
Like early voting, vote-by-mail has its advocates and its critics. Advocates
argue that it increases the convenience of voting and boosts turnout, while saving
money by reducing the cost of administering elections. Vote-by-mail is also said
to increase uniformity in the voting process, and the iterative process of mailing
out ballots and receiving notices of non-delivery helps keep voter registration
rolls up-to-date (Bradbury, 2005). Critics counter that vote-by-mail leaves voters
at the mercy of a sometimes unreliable US Postal Service; does little if anything to
boost turnout; is vulnerable to fraud; and privileges the affluent over low income
voters (Slater and James, 2007). (On turnout and vote-by-mail, see Southwell and
Burchett, 2000, and Berinsky, Burns and Traugott, 2001). While vote-by-mail has
increased rapidly since Oregon adopted it 12 years ago, it remains to be seen
whether its growth will continue and reach beyond the West Coast.
Same Day/Election Day Registration
While early voting and other forms of convenience voting are aimed principally
at making ballot casting easier, same day registration (SDR) or election day
registration (EDR) aims at lowering another barrier: voter registration. In SDR and
EDR, voters are allowed to register and vote on the same day, either before or on
election day itself. In addition to states that open SDR and EDR to all voters, some
states offer it only to certain kinds voters, such as recently discharged uniformed
Figure 9.3 Map of Same Day/Election Day Voting Rates
services personnel or new citizens. In still others, SDR can occur as a side effect
of the election calendar; for example, when the close of registration comes after
the start of early voting, creating a window in which people can do both on the
same day.
The 2008 EAVS asked about SDR/EDR in a roundabout way, due to refusal
of some states to classify their voting as SDR or EDR, and problems with states
tracking the votes of those who registered at the same time they cast a ballot. The
2008 EAVS asked states to report the number of registration forms received on
days that voters could register and vote on the same day. These figures give some
indication of the prevalence of SDR/EDR.
Nationally, 17 states reported receiving 3,616,874 voter registration applications
on days when the voter could then cast a ballot. At least 963,144 of these applications
resulted in new voters being added to the voter list; however, because most states did
not categorize the applications they received, the actual number was almost certainly
much higher.
Like other non-traditional registration and voting practices, SDR/EDR has
been criticized as creating a vulnerability to election fraud. Without effective
registration lists that can be checked against databases in an effort to detect dead
or ineligible voters, critics argue, non-eligible voters can walk in, register and vote
without the necessary screening mechanism of the traditional registration process.
In response to these concerns, and to facilitate other innovations such as vote
centers (see below), some states have adopted electronic pollbooks that update
continuously during a voting period, enabling election officials to track who has
voted in real-time.
Provisional Voting
Federal law guarantees voters in federal elections the right to cast a provisional
ballot if they are otherwise refused a regular one. Generally, the need to vote
provisionally occurs when a voter’s name is not on the registration list, and the
voter contends that he or she is lawfully registered to vote. Provisional ballots
are used by states for a wide variety of other purposes, including the processing
of change-of-address (in Ohio) and the failure by a voter to produce required ID.
Another source of provisional ballots is voters who request absentee ballots, then
either do not receive them or fail to return them.
In close elections, provisional ballots can be the focus of intense scrutiny,
as their validity may not be determined for days or weeks after election day. In
addition, some have looked to provisional ballots as an overall indicator of how
well an election system works. Gerken has called provisional ballots the “miner’s
canary” for election administration, providing a warning sign that a particular
election system is “under stress” (Gerken, 2009).
More than 2 million provisional ballots were cast during the 2008 Presidential
election. Nearly three-quarters of these ballots were cast in just four states:
California, New York, Ohio and Arizona (Table 9.1). Nationally, 28.2 percent of
provisional ballots, or 609,016 ballots, were rejected, while another 5.5 percent
were counted only in part. The most common reason for rejection, accounting
for slightly more than half the rejections, was the determination that a voter was
not properly registered in the state; another 16.8 percent were rejected for not
being properly registered in the local jurisdiction or in the wrong precinct. Failure
to provide sufficient identification resulted in the rejection of 12,321 ballots.
In spite of Gerken’s assertion, interpreting provisional ballot statistics
remains problematic. States employ provisional ballots in very different ways,
and acceptance rates are difficult to interpret. High rates of provisional ballot
acceptance may mean that voters are being improperly denied a regular ballot;
low rates of acceptance may mean that many voters who think they are properly
Table 9.1 Top Provisional Ballot States, 2008
State Provisional Ballots Cast Provisional Ballots Rejected % Rejected

California 798,332 136,286 17.1
New York 279,319 111,843 40.0
Ohio 204,651 39,390 19.2
Arizona 151,799 44,473 29.3
registered are in fact not. Both suggest problems in the election system, so
establishing a normative basis is a challenge. In any case, provisional ballots
remain a time bomb in American elections, providing fertile ground for recount
controversies.
Voter ID
The introduction of more stringent identification requirements for voters was one
of the most hotly contested election administration issues of the past decade. A
weak ID requirement was included in HAVA, but covered only first-time voters
who registered by mail and did not provide an ID when they registered. The list of
acceptable IDs was also long, and included photo IDs and forms of identification
such as utility bills. Pushed mainly by Republicans as a weapon against voter
impersonation, voter ID laws caused considerable controversy when they were
passed in Georgia and Indiana in 2005, and later in Florida and Arizona. Controversy
also arose when a voter ID proposal was included among the recommendations of
the Commission on Federal Election Reform (the “Carter-Baker Commission”) in
2005. The Supreme Court’s ruling in Crawford v. Marion County Election Board
in 2008 validated the Indiana law, although litigation and public controversy
continue. To opponents, tougher ID requirements are a partisan ploy to discourage
voting among the poor, ethnic and racial minorities and the elderly, all of whom
opponents claim are more likely to lack the required forms of ID.5
Voter ID laws are generally complicated pieces of legislation, making
comparison between states difficult. The number of states requiring photo ID
as a prerequisite to voting, however, remains small. The EAC survey found that
only two – Florida and Indiana – require photo ID of all voters (Georgia did
not respond to the Statutory Overview portion of the survey; three other states
nominally require photo ID but accept an affidavit in its absence). Twenty-four of
the 49 responding states required some form of ID beyond oral self-identification
or provision of a signature; most of these accept non-photo ID.
Voting Technology
The notorious failure of punch cards in the Florida Presidential race in 2000
launched a federal drive to replace voting equipment: some $3 billion was spent on
voting technology through the EAC and HAVA, and there was wide expectation
that electronic voting machines would replace the myriad punch card, optical
5 There has also been an attempt to require proof of citizenship in order to register
or vote. Fewer states have moved on this issue, Georgia and Arizona being the exceptions,
although with illegal immigration a leading national issue, more states are expected to
follow suit.
Figure 9.4 Map of Voting Technology
scan and lever machines in place across the country. Deploying direct recording
electronic voting machines (DREs) turned out to be more difficult than anticipated,
however. Concerns over their reliability and susceptibility to tampering spawned a
cottage industry of electronic voting skeptics, who advocated for more traditional
systems such as optical scan ballots that leave a paper ballot behind for recounting
and auditing purposes (see Warf, 2006). Called voter-verified paper audit trails, or
VVPAT, they were adopted by a series of states such, as Maryland, who had been
early adopters of DREs. The 2008 election, by one estimate, was the first in which
the number of voters casting ballots by DREs actually declined over the previous
election (Election Data Services, 2008).
Legacy technology, local discretion and controversies such as the one
surrounding electronic voting has resulted in a patchwork in the deployment of
voting technology (see Figure 9.4).6 While lever machines and punch cards were
doomed by the requirements of HAVA, surviving were a wide variety of systems,
including those using optically scanned ballots; various makes and models of
DREs; and hybrid systems incorporating multiple technologies. Some jurisdictions
6 Because the EAC survey collects data on the number of machines rather than on
how the technologies are used, I have used data from a nonpartisan advocacy group, Verified
Voting, here rather than the EAC. The map in Figure 9.4 is adapted from Verified Voting’s data.
Available at: http://www.verifiedvoting.org/verifier/index.php?ec=standard&state=&year=.
use one technology for in-person voting, another for absentee ballots, and a third
for voting by disabled persons.
States that rely on DREs without a paper record are clustered along the East
Coast. With New York’s decision to adopt paper ballots, lever machines have
vanished from the country’s electoral landscape, and punch cards are in use only
in some Idaho counties. Many states use combinations of technologies, either
through serving different groups of voters or through variation from county to
county; the state-level map in Figure 9.4 actually understates the spatial variation
in voting technologies deployed.
Emerging Issues
Changes in voting technology, convenience voting and other election administration

issues will continue to be foci of election scholars over the coming decade. In
addition, a number of new issues appear poised for the next decade’s battles
over election reform. Voter identification laws are still being introduced in state
legislatures in the wake of the Supreme Court’s upholding of the Indiana ID law,
and many of those same states will consider changes to the way they maintain their
voter registration lists. In 2010 Georgia won federal clearance for a new database
matching procedure which would compare new voter registration applications
against drivers’ license and Social Security databases, in an effort to confirm
the citizenship of new voters. Arizona has a similar law. Opponents claim the
matching procedure will place an unequal burden on minority, elderly and poor
voters (Keefe, 2010). Other states may consider implementing such procedures on
the heels of the federal clearance.
Voting by overseas citizens and uniformed services members will continue
to be a focus both of research and policy, for a number of reasons. Almost alone
among the major election topics, efforts to ease participation by these voters enjoys
strong bipartisan support. The federal UOCAVA also gives federal lawmakers an
unusual lever over what is traditionally a local and state prerogative. Finally, new
technologies may be more of a boon to distant voters than to domestic ones. State
experimentation in UOCAVA voting, such as allowing voter registration over
the Internet and transmitting ballots electronically, will be watched closely for
potential broader application. In 2009, Congress passed the Military and Overseas
Voter Empowerment Act, which will prompt further changes in how these voters
cast their ballots.
Conclusions
The ways that American cast their ballots shows considerable spatial variation
across the United States, and while reliable historical data is lacking, it appears
that this divergence is increasing. Strong regional patterns appear as well. In the
West, voters increasingly cast their ballots through the mail. In the upper Midwest,
and in selected other states, governments are lowering barriers to registration by
allowing voters to register and cast their ballots on the same day. Early voting
has become popular across the South and Southwest. Finally, some Southern and
Western states have passed new laws that place additional burdens on voters to
prove their identity and citizenship.
This divergence is important in several ways and at several scales. Nationally,
the variation introduces a potentially significant variable in the ways that federal
elections, including presidential ones, are decided. Combined, factors such as
convenience voting, voter ID and registration rules, provisional ballots and voting
technology can help mold the electorate and the number of counted ballots, and
therefore can impact candidates’ vote totals. Lengthening the window in which
ballots are cast through early voting and extended deadlines for some voters
means that people now vote with different sets of political knowledge. Late-
breaking events, such as the disclosure of President George Bush’s 1976 drunk
driving charge, may come before election day but after substantial numbers of
voters have already cast their ballots in some states. In addition, campaigns run
under early voting laws may be more expensive or place a premium on a good
campaign organization, which is needed to leverage the increased time to get
a candidate’s backers to the polls. The impact of all of these variations is still
poorly understood.
At the state level, election laws are vulnerable to manipulation for partisan
advantage. Laws that restrict or expand access to the ballot through raising or
lowering the bar for registration and voting will rarely fall equally on all segments
of the electorate. It is an article of faith among both parties that the marginal voter
– the voter who will vote only in high turnout elections and even then may need
additional motivation through get-out-the-vote techniques – will tend to vote
Democratic in most places. Generally, Republicans, citing fear of electoral fraud,
favor putting additional burdens on voters to ensure that only eligible voters cast
ballots. Democrats, emphasizing the importance of broad participation and leery
of laws that might disenfranchise, argue for making registering and voting easier.
Both parties are well aware of the partisan impacts of changes in election laws.
Locally, changes in election laws inevitably raise questions of equal protection,
particularly if discretion is left to local election officials in implementation and
enforcement. Equal access to the franchise for voters of ethnic or racial minorities
has been guaranteed relatively recently, and deep suspicions remain about the
country’s ability to preserve equal voting rights from discrimination and the
push for partisan advantage. Barriers to voting can be both formal, through the
imposition of laws that disproportionately impact certain groups, or informal,
through the use of administrative policies that discriminate. Formal barriers may
include restrictive registration or polling place laws, while informal techniques
may be such seemingly mundane administrative decisions such as the distribution
of voting machines or the location of polling places or early vote centers. New
types of data, such as information on waiting times at polling places for white
and minority voters, may be needed to measure disparate impact of some of the
voting changes outlined in this chapter. In fact, given the growing divergence in
American voting laws, such issues may be the new battleground for voting rights
as the twenty-first century progresses.
Beyond the impact of election laws and practices on the political process,
scholars interested in American regionalism can glean insights by the ways states
and regions run their elections, and the discourses of reform that are produced.
Debates over voter ID, for example, can help us understand the ways that
meanings of citizenship are contested in different parts of the United States. The
gulf between those who see voting as a right and those who see it primarily as an
obligation or a privilege intersects with national and regional debates over race
and illegal immigration, among other topics.
This chapter has relied upon new and improving datasets, but the push for
improved data collection in still in its early stages. There is still a crying need
for more and better quality data, at lower scales (including, most importantly, the
precinct level), and for better and more widely available maps. Data the county (or
equivalent) and state levels don’t tell us much about the way variation in balloting
laws and practices may discriminate against particular communities. Since few
states maintain voter registration data by race, reference to demographic variables
will generally be accomplished through Census data at the tract- and lower levels.
Calculating these variables will require more accurate and up-to-date precinct
maps, made available through election websites. Few jurisdictions provide such
maps at present.
Geographers, particularly political geographers, should be attuned to the
potential research and service opportunities available in helping improve
election data, cartography and analysis. Research into the geography of election
administration can be a valuable and engaging teaching tool, and an outlet for
student and professor community outreach. Every university, community college
and high school resides in an election jurisdiction, and many of these jurisdictions
are eager for research and service partnerships. On the scholarly side, disputes
over voting rules will demand the expertise of scholars who can evaluate their
potential impact; litigation over voting rights and other election topics shows no
signs of abating. Geographers are well positioned to make valuable contributions
to this field.
The tug-of-war that is embedded in the current landscape of American
voting and described briefly in this chapter is largely the result of two distinctly
American facets of democracy: local control of election administration, and more
fundamentally, longstanding differences in how Americans view democracy and
the right to vote. The extreme federalism of the US election system leaves states
and their local jurisdictions with wide discretion on how to run the voting process;
local preferences and histories are producing different solutions to the challenges
of casting and counting ballots in twenty-first century. These local practices
combine with deep disagreements over the value of the right to vote and the
obligations that can be placed upon a voter by the state. These differences predate
the US Constitution and have never been entirely resolved: it is no accident that
since the Bill of Rights, nearly half of the Constitution’s 17 amendments have
concerned the right to vote. Despite these amendments and the passage of such
landmark federal laws as the Voting Rights Act, Americans have yet to arrive at a
national consensus on who should be allowed to vote and the role of the state in
making voting easier or harder to do. The divergence across the country in how
Americans cast their ballots will be well worth studying in coming years.
References
Agnew, J. 1996. “Mapping Politics: How Context Counts in Electoral Geography.”

Political Geography 15, 2: 129–46.
Berinsky, A., Burns, N., and Traugott, M. 2001. “Who Votes By Mail? A Dynamic
Model of the Individual-Level Consequences of Voting-By-Mail Systems,”
Public Opinion Quarterly.
Bradbury, B. 2005. “Vote-by-mail: The Real Winner is Democracy,” Washington
Post, January 1, A23.
Election Data Services. 2008. “Nation Sees Drop in Use of Electronic
Voting Equipment for 2008 Election – A First.” Available at: http://
www.electiondataservices.com/images/File/NR_VoteEquip_Nov-2008w
Appendix2.pdf.
Fallows, J. 2010. “Fallows on the News: NY Mosque and More.” Commentary on
All Things Considered, National Public Radio, August 14. Transcript available
at: http://www.npr.org/templates/story/story.php?storyId=129201849.
Gerken, H. 2009a. The Democracy Index. Princeton, Princeton University Press.
Gerken, H. 2009b. “Provisional Ballots: The Miner’s Canary for Election
Administration.” In Provisional Ballots: An Imperfect Solution, report
published by the Pew Center on the States. Available at: http://www.
pewcenteronthestates.org/initiatives_detail.aspx?initiativeID=54789.
Gronke, P., E. Galanes-Rosenbaum and P. Miller. 2007. “Early Voting and
Turnout.” Political Science and Politics 40: 639–45.
Gronke, P., E. Galanes-Rosenbaum, P.A. Miller and D. Toffey. 2008. “Convenience
Voting.” Annual Review of Political Science 11: 437–55.
Keefe, B. 2010. “Ga. policy halts voting by illegal immigrants, backers say.”
Atlanta Journal-Constitution, August 30.
King, G. 1996. “Why Context Should Not Count.” Political Geography 15, 2:
159–64.
Overton, S. 2006. “Voter Identification.” Michigan Law Review 105, 631.
Slater, M. and T. James. 2007. “Vote-by-mail Doesn’t Deliver.” Commentary on
Tompaine.com. Available at: http://www.tompaine.com/articles/2007/06/29/
votebymail_doesnt_deliver.php [last accessed: February 18, 2010].
Southwell, P., and J.I. Burchett. 2000. “The Effect of All-Mail Elections on Voter
Turnout.” American Politics Quarterly 28(1): 72–9.
Taylor, P., and R.J. Johnston. 1979. The Geography of Elections. Penguin,
Harmondsworth.
Tokaji, D. 2005. “The Moneyball Approach to Election Reform.” October 18
commentary. Available at: http://moritzlaw.osu.edu/electionlaw/comments
/2005/051018.php.
Warf, B. 2006. “Voting Technologies and Residual Ballots in the 2000 and 2004
Presidential Elections.” Political Geography 27: 664–77.
Chapter 10
The 2008 US Presidential Primary Elections
in Geographic Perspective
Fred M. Shelley and Heather Hollen1
The 2008 US presidential election was one of the most riveting and fascinating
presidential elections in the country’s history. Two months before the general
election, it was all but certain that the voters of the United States would either
elect an African-American president or a female Vice President for the first time
in US history.
The selections of Senators Barack Obama and John McCain as the presidential
nominees of the Democratic and Republican parties, respectively, followed many
months of intensive campaigning among potential nominees in both parties.
Voters throughout the country expressed their preferences among these potential
nominees in primary elections and caucuses in all 50 states. The results of these
primary elections themselves reveal a fascinating portrait of geographic differences
within states as voters expressed their preferences for the Presidential nominees of
the two major parties.
The purpose of this chapter is to examine the geography of the 2008 primary
elections. In doing so, it is possible to describe the history and geography of the
United States’ unique system of selecting major-party Presidential candidates.
Detailed discussion of the 2008 campaigns for the Democratic and Republican
nominations are presented along with the results of cartographic and statistical
analysis of county-level voting outcomes in some key states. Furthermore,
analytical issues associated with using electoral-geographic analysis of US
presidential primary elections are called to attention. A general discussion on the
impacts of the primary election campaign on the general election, and on future
elections concludes the chapter.
US Presidential Primary Elections in Geographical and Historical Context
The process of selecting the president of the United States is a two-step process.
First, both major parties select their presidential nominees. Once these nominees
are chosen, they run against each other in the November general election.
1 The authors thank Trung Vinh Tran for assistance with the graphics.
Methods of selecting nominees for high public offices such as the president
of the United States vary among the world’s democracies. A party’s nominee can
be chosen in one of several ways: by party members who are elected officials, by
the party elite, by party activists, and by ordinary voters (Punnett 1993). In the
United States, all four methods of selecting party nominees to run for President
have been used at various times over the course of American history. Over the
past two centuries, major American political parties have moved from selection
of presidential nominees by members of Congress to selection by party elites to
selection by representatives chosen by ordinary voters. The system of presidential
primary elections used today gives ordinary voters much more influence in the
selection of presidential nominees than was the case in the past, or is the case in
many other countries.
Relative to political parties in many other countries, US political parties are
loose coalitions of often disparate interest groups. Although each party drafts
a platform that expresses the policy positions of the party on major domestic
and international issues, these party platforms are not binding on the party’s
candidates and its members. Frequently, battles within a party for its presidential
nomination reflect deeper divisions within the party over the party’s priorities
and its underlying political philosophy. Both parties attempt, with varying
degrees of success, to unite diverse economic and cultural interests that are
often diametrically opposed to one another. Tensions between these interests
are often especially evident during primary elections. The 2008 election was
no exception.
Prior to the New Deal of the 1930s, two key elements in the Democratic
Party’s uneasy coalition included Southern conservatives and Northern urban
dwellers. The South had been strongly Democratic since the end of the Civil
War and Reconstruction. Many Southern Democratic leaders and officeholders
were nativists, supporters of prohibiting the manufacturing and sale of alcoholic
beverages, supporters of state’s rights, and vehement opponents of civil rights
for African-Americans and other minorities (Key 1949). In the North, however,
many Democrats were blue-collar urban dwellers and labor union members.
Many were European immigrants themselves, or the children or grandchildren
of immigrants. Describing the Democrats of the 1930s, Barone (1990, 28) wrote
that “The Democracy was a party of white southerners and northern Catholics, of
Southern Baptist Prohibitionists and immigrant imbibers, of nativists and those
who spoke no English, of teeming eastern cities and the wastelands of the Great
Basin. Its members had little in common except that most of its members were not
native-born white Protestants.”
Barone went on to describe the constituencies of two leading Democrats in the
House of Representatives at this time: Eugene Cox of Georgia and Adolph Sabath
of Illinois. Barone wrote, “Two such different constituencies as the worn-out cotton
lands of south Georgia and the mostly abandoned slum streets of Chicago’s river
wards could hardly be imagined; little wonder that their representatives, members
of the same party who were yoked together on the Rules Committee, could agree
The 2008 US Presidential Primary Elections in Geographic Perspective 177
on almost nothing” (224). Franklin D. Roosevelt, however, was able to create a

Democratic majority during the 1930s and 1940s by linking these two traditional
Democratic constituencies with African-Americans, ideological liberals, and
academics and other highly educated professionals. This “New Deal” coalition
elected Roosevelt four times. Even after it broke down at the presidential level, the
Democrats controlled the House of Representatives for all but four years between
1931 and 1994. Although the Democrats have held the White House for only 12 of
the previous 40 years, this coalition continues to impact Democratic Party politics.
Tension between old-line, working class Democrats and ideological liberals and
professionals has characterized the party’s politics for decades, and was critical to
the 2008 primary elections (Shelley 2008).
The Republicans have often been as divided as the Democrats over the past
several decades. Prior to the 1960s, the Republicans were split between what
some historians have termed “Wall Street” and “Main Street” Republicans. The
“Wall Street” Republicans were based in the large metropolitan areas of the East
and Midwest. They were internationalist in outlook, oriented to big business,
and supportive of an assertive foreign policy; after World War II, they strongly
supported an aggressive stance against the Soviets in the Cold War. “Main Street”
Republicans, on the other hand, tended to come from small towns and rural areas
west of the Appalachians. They were much more skeptical of foreign involvement
and supported agrarian interests.
During the 1960s and 1970s, the Republicans drifted to the conservative side
of the political spectrum. The nomination of Barry Goldwater, whose candidacy
was strongly opposed by Northeastern liberal Republicans, for President in 1964
signaled the rise of conservatives, who often opposed the continued expansion of
government, within the Republican Party. As conservatives became increasingly
prominent within the Republican Party, they became divided into several distinct
factions. Economic conservatives favored restricting government influence on the
economy. Social conservatives favored increased government control over such
issues as abortion, gay marriage, and immigration. Neoconservatives, who support
an aggressive foreign policy stance, represent a third major interest group in the
Republican Party.
Just as Roosevelt succeeded in uniting the Democratic factions in the
1930s, Ronald Reagan was successful in uniting Republicans, both economic
conservatives and social conservatives, in the 1980s. Since Reagan’s time,
however, the Republicans have faced continuing tension between economic,
social, and foreign policy conservatives. These tensions were evident in the early
primaries in 2008 as well.
The task of resolving intraparty tensions and attempting to unite the disparate
interests that comprise major political parties has been undertaken in different
ways over the course of US political history. During the early nineteenth century,
members of Congress selected their parties’ nominees for President and Vice
President. In 1831, however, the Anti-Masonic Party held the first national party
convention to select the party’s 1832 nominees. The Democratic Party soon
followed suit, and the Whig Party held its first national convention in 1839.
Since that time, US political parties have chosen their nominees at national party
conventions.
The transition from selecting nominees by members of Congress to selecting
them at national party conventions was facilitated by the development of the
railroad, which enabled delegates from most parts of the country to attend them
after at most a few days of travel.
Nineteenth-century nominating conventions were held at locations easily
accessible by rail. The 1831 Anti-Masonic convention was held in Baltimore, as
were the first six Democratic national conventions between 1832 and 1852 and five
of the six Whig national conventions between 1839 and 1860. After the Civil War,
Chicago emerged as the center of the nation’s railroad network. Between 1860
and 1916, nine of the Republican Party’s 15 nominating conventions took place in
Chicago. Today, delegates from throughout the country can reach the convention
site within a day. Since transportation time and cost are no longer critical factors
in selecting convention sites, political or symbolic factors have taken precedence
in site selection. For example, the Republicans held their 2004 convention in New
York in part to remind voters of Republican President George W. Bush’s response
to the September 11, 2001 terrorist attack on New York’s World Trade Center.
During the nineteenth and early twentieth centuries, Republican and
Democratic Party conventions were dominated by party elites, including political
bosses whose decisions were made in “smoke-filled rooms.” The public had little
input in the selection of convention delegates or party nominees. In 1910, however,
Oregon established the first presidential preference primary. Oregon’s delegates to
the parties’ national conventions were required to support the winner of the state’s
primary. Between 1920 and 1968, 13 to 20 states held primary elections each
presidential election year (Ware 2002). However, many of these primaries were
“beauty contests.” In many states, delegates were selected independently of the
presidential primaries, and they were under no obligation to support the candidate
who got the most votes in their state’s primary election.
In 1968, the Democratic Party held a bitterly contested nomination battle. In
March of that year, President Lyndon Johnson almost lost the first-in-the-nation
New Hampshire primary to his challenger, Senator Eugene McCarthy, who opposed
continuing US involvement in the Vietnam War (White 1969; Shelley 2008).
Johnson withdrew from the race soon afterwards and Senator Robert Kennedy and
Vice President Hubert Humphrey entered the race for the Democratic nomination.
Kennedy, who also opposed continuing US involvement in the Vietnam War,
ran head-to-head in several primaries against McCarthy before Kennedy’s tragic
assassination after the California primary in June. Most of the delegates selected in
primary elections were pledged to either Kennedy or McCarthy. Humphrey did not
enter the primaries. Nevertheless, he was supported by enough delegates and party
leaders from non-primary states to win the nomination. He lost the subsequent
general election to the Republican nominee, Richard Nixon.
After the bitter intraparty battle of 1968, the Democrats initiated party reforms
that gave greater input to ordinary voters in the selection of their party’s presidential
nominees. These reforms were used for the first time in the 1972 election. The
Republicans initiated similar reforms in the 1970s. These new rules mandated
more public participation in the selection of convention delegates. All states now
hold either primary elections or caucuses in order to select convention delegates.
Primary election procedures vary between parties and from state to state.
Some states hold “closed” primaries in which a voter is required to be a registered
party member in advance in order to participate in that party’s primary. Others use
“open” primaries, in which a voter can choose either party’s primary ballot at the
polls and is not required to be a registered party member. Delegates chosen through
primary elections or precinct caucuses are pledged to particular candidates and are
required to vote for their candidate in the first round of balloting at the national
convention. The methods by which the parties allocate pledged delegates between
candidates for the nomination also vary between parties. The Democrats use a
system of proportional representation. The number of delegates pledged to each
candidate in a state’s primary is proportional to that candidate’s percentage of vote
in the primary. In contrast, in many states the Republicans use a “winner-take-all”
system in which all the elected delegates are pledged to the candidate who wins
the most votes in that state’s primary. About 20 percent of Democratic delegates
are so-called “superdelegates.” Superdelegates, who include Democratic members
of Congress, state governors, and leading party professionals, are free to vote for
whomever they wish and are not bound by primary results.
In 2008, the winner-take-all system worked to McCain’s advantage in his quest
for the Republican nomination, because McCain won several primaries over his
leading opponents by narrow margins. For example, in the Missouri Republican
primary on February 5 McCain got 194,053 votes (33.0 percent), Mike Huckabee
got 185,642 (31.5 percent), and Mitt Romney got 172,329 (29.3 percent) (Figure
10.1). However, in accordance with the winner-take-all system McCain got all 58
of Missouri’s pledged delegates. In Oklahoma, McCain got 122,722 votes (36.6
percent) as compared to 111,899 for Huckabee (33.4 percent) and 83,030 for
Romney (24.8 percent) (Figure 10.2). Under Oklahoma’s rules, three of the state’s
25 delegates go to the winner in each of the state’s five House of Representatives
districts, with the other ten going to the statewide winner. Huckabee won pluralities
in two of the five districts and gained six delegates, while McCain got 19 delegates
by winning pluralities in the other three districts and a statewide plurality. Romney,
despite finishing a close third in both Missouri and Oklahoma, won no delegates
from either of these states.
Furthermore, when compared to the Democratic party, the proportional system
helped Obama, who lost most of the larger states to Clinton but won a majority of
primaries and caucuses in the smaller states. In Missouri, Obama had 406,917 votes
(49.2 percent) with 395,185 (47.9 percent) for Clinton. Each got 36 of the state’s
72 delegates. In Oklahoma, Clinton won a substantial majority, with 228,480 votes
(54.8 percent) to 130,130 (31.2 percent) for Obama and 42,725 (10.2 percent)
Figure 10.1 Map Showing the Results of Missouri Republican Primary, 2008
for the third-place finisher, John Edwards. Clinton won 24 of the 38 Oklahoma
delegates, with 14 going to Obama. Because Oklahoma Democratic Party policy
requires a candidate to achieve at least 15 percent of the vote to get any delegates,
Edwards got no pledged delegates. Under Republican rules, Obama would have
won all of Missouri’s delegates and Clinton would have won all of Oklahoma’s
Figure 10.2 Map Showing the Results of Oklahoma Republican Primary, 2008
delegates. However, if all states had used winner-take-all systems, Obama would
have lost the Democratic Party nomination to Clinton, who won many large-state
primary elections including those in California, Texas, and New York.
The Primary Calendar and its Impact
The United States’ primary election system is also unique in that primaries are not
held simultaneously. Rather, various states hold their primary elections at different
times over a period of several months. In 2008, the first primary elections took
place in January; the final ones occurred in June.
The order in which the states hold primary elections is critical to the eventual
outcome of the process. Candidates who do poorly in early primaries often drop
out of the race. Thus, voters in states that hold later primaries may be unable to
vote for their initial candidates of choice. In January 2008, for example, voters
in New Hampshire, which has a monopoly over the first primary each election
year, could choose among seven Republicans and six Democrats. By March, all
of the viable candidates except for McCain, Obama, and Clinton had dropped
out of the race. McCain had clinched the Republican nomination by early March,
and Republican voters in late-voting states had no input into the selection of their
party’s nominee. As a general rule, the number of viable alternatives available to

voters declines from earlier to later primaries (Shugart 2004).
Because states holding early primaries tend to have more influence on the
outcome, there was considerable controversy about what states would be permitted
to hold the earliest primaries in the 2008 cycle. Since 1952, New Hampshire has
held the first primary election in the United States. New Hampshire has held a
primary election every four years since 1916 (Gregg and Gardner 2000). The New
Hampshire primary proved to be an early and important gauge of voter support for
the Democrats in 1952 and 1968, and for the Republicans in 1952, 1960, 1964, and
1968. By the late 1960s, the New Hampshire primary had become recognized as
the first important test of voter support each presidential election year.
After the two parties reformed their nominee selection processes in the
1970s, the state of Iowa initiated a system of precinct caucuses, which were held
prior to the New Hampshire primary. Democratic and Republican voters in each
precinct across the state meet to elect delegates to county conventions (Hull 2007).
Delegates pledged to that potential nominee at the precinct level are allocated
proportionally on the basis of support levels within the precinct (Squire 1989). The
county conventions select delegates to Congressional district conventions, which
in turn select delegates to statewide conventions. The Iowa caucus system gained
nationwide attention in 1976, when then little-known Jimmy Carter, the former
governor of Georgia, won a plurality. Carter’s success in Iowa helped propel him
to victory in the New Hampshire primary and eventually to the Democratic Party
nomination and the Presidency.
Since 1976, Iowa and New Hampshire have held the first caucuses and first
primary, respectively, in the country each presidential election year. In 1977, the
New Hampshire legislature passed a law requiring New Hampshire’s primary
to precede those of any other state (Gregg and Gardner 2000). The parties were
authorized to move the New Hampshire primary to an earlier date if another state
announced plans to hold its primary earlier than that of New Hampshire. On
several occasions, New Hampshire has moved its primary earlier and earlier as
other states have scheduled early primaries also. In 1968, the New Hampshire
primary was held on March 12; in 2008, it took place on January 8. Critics have
pointed out that Iowa and New Hampshire with their small populations, few
minority voters, and lack of large metropolitan areas are unrepresentative of the
US electorate as a whole. It is also argued that the early scheduling of Iowa’s
caucuses and New Hampshire primary gives these states a disproportionate
influence in the outcome of presidential elections. Since 1952, only Bill Clinton in
1992 has won the Presidency without winning either the Iowa precinct caucuses or
the New Hampshire primary. Moreover, Clinton did not contest the Iowa precinct
caucuses that year because one of his rivals for the Democratic Party’s presidential
nomination that year was an Iowan, Senator Tom Harkin. Clinton and George W.
Bush in 2000 were the only Presidents since 1960 to have won the general election
without winning the New Hampshire primary. In 2008, Obama and Huckabee won
Iowa, while Clinton and McCain won New Hampshire.
In response to the perceived disproportionate influence of Iowa and New

Hampshire in the electoral process over the years, party leaders in other states
considered holding their primaries earlier, or working with leaders in neighboring
states to hold their primaries on the same day in order to increase the influence of
their regions on the nomination process. In the 1980s, some southern Democratic
leaders became concerned that Southern candidates would be uncompetitive in
the Northern states of Iowa and New Hampshire. In response, they decided to hold
their primaries on the same day with the idea that a Southern-oriented candidate
who did well in several states at the same time would get a boost toward the
Democratic Party’s presidential nomination. Thus the “Super Tuesday” concept
was born, and was used for the first time in 1988. In that year, the Southern
primaries were divided among several candidates and a Northern candidate,
Governor Michael Dukakis of Massachusetts, won the Democratic nomination.
Four years later, however, Clinton swept the Super Tuesday states in 1992 en route
to the Democratic nomination and victory in the general election.
Before the 2008 election cycle began, party leaders in other states also
explored the idea of adjusting the primary calendar in order to give voters in their
states more influence in the nomination process. In August 2006, the Democratic
National Committee agreed to allow South Carolina and Nevada, along with Iowa
and New Hampshire, to hold “early” primaries prior to February 5. South Carolina
and Nevada are Southern and Western states, respectively, and both contain larger
minority populations than do Iowa or New Hampshire. Between them, the four
“early” states were seen as representative of the country as a whole in terms of
both geography and demography. The Committee ruled that these four states could
hold their primaries prior to February 5, but that no other state could do so.
In 2007, this ruling was challenged in Florida and Michigan. The Michigan
Legislature, without the support of the Democratic National Committee,
passed a bill scheduling that state’s primary for January 15. Florida scheduled
its primary for January 29. In response, the Committee stripped the Michigan
and Florida Democratic Parties of their delegates, rendering the primaries
meaningless. The Democratic candidates agreed not to campaign in these states,
and Obama and several other candidates had their names taken off Michigan’s
Democratic ballot. Nevertheless, Democratic voters in both states participated
in the primaries, both of which were won by Hillary Clinton. Later in the
campaign, Clinton would argue for seating delegates from these states. As the
nomination campaign drew to a close and Obama’s nomination was assured, the
Committee eventually agreed to seat the delegates from the two states.
As the controversy associated with the Florida and Michigan primaries
continued, other states scheduled their primaries. Twenty states including
California, New York, and Illinois chose February 5, the first day that the
Democratic National Committee set aside for primary elections. Thus, February 5
became the 2008 version of Super Tuesday. The remaining 26 states held primaries
at various times between February 9 and June 3.
The 2008 Candidates and Outcomes
As the 2008 primary season approached, potential nominees in both parties

announced their intentions to run for the Presidency. The Republicans anticipated
a protracted battle between several candidates, each of whom brought very
different strengths. Meanwhile, Clinton was seen as the front-runner for the
Democratic nomination throughout 2007. Most experts expected her to clinch
the Democratic nomination early, whereas they expected a drawn-out battle for
the Republican nomination. As events turned out, the Republican nomination
was decided quickly whereas the Democratic nomination was not decided until
the end of the primary season in June.
This uncertainty occurred in part because neither the incumbent President nor
the incumbent Vice President sought the Presidency. For only the second time
since 1928, both parties knew going into the 2008 presidential election campaign
that neither President Bush nor Vice President Dick Cheney would be on the 2008
general election ballot. In 1951, the 22nd Amendment, which limits Presidents
to two terms, was adopted. Between 1951 and 2000, only three Presidents –
Dwight D. Eisenhower in 1960, Ronald Reagan in 1988, and Bill Clinton in 2000
– completed two four-year terms. Vice Presidents Richard Nixon, George H.W.
Bush, and Al Gore, respectively, won their parties’ presidential nominations. Bush
was elected in 1988, while Nixon and Gore narrowly lost to John F. Kennedy and
George W. Bush respectively. The younger Bush’s Vice President, Dick Cheney,
had low popularity ratings and had faced a long history of health problems. Cheney
said repeatedly during Bush’s Presidency that he had no interest in running for
President (Barnes 2005), and he made no effort to contest the 2008 Republican
nomination.
With neither George W. Bush nor Cheney on the ballot, the race for the
Republican nomination was wide open. During 2007, several candidates announced
plans to run. These included Senator John McCain of Arizona, former New York
City Mayor Rudolph Giuliani, former Governor Mike Huckabee of Arkansas,
former Governor Mitt Romney of Massachusetts, former Senator Fred Thompson
of Tennessee, and Representative Ron Paul of Texas. In addition to these potential
nominees, several other candidates including Senator Sam Brownback of Kansas
and Representatives Duncan Hunter of California and Tom Tancredo of Colorado
also initiated campaigns. None of these campaigns generated momentum, however,
and all three dropped out early in the process.
Each of these candidates appealed to a very different constituency, reflecting
the long-standing tension between economic, social, and foreign policy
conservatives within the Republican Party. McCain, who had contested and lost
the Republican nomination to Bush in 2000, emphasized his role as a maverick and
an independent voice. McCain was a strong supporter of the war in Iraq, but took
a more moderate position on many social and environmental issues than did many
Republican voters. Giuliani, who had been Mayor of New York at the time of the
9/11 terrorist attack, emphasized national security issues. Huckabee, an ordained
Southern Baptist minister, emphasized social conservatism. Romney emphasized

his business experience and the economy in his campaign, while Thompson tried
to appeal to the Republican Party’s conservative base. Paul, who had been the
Libertarian Party’s presidential nominee in 1988, argued against the war in Iraq
and against government interference in the economy and in social issues.
Despite this vigorous competition within the Republican Party, many observers
expected that the Democrats would win the 2008 election. Several factors appeared
to boost the Democrats’ chances. Since the adoption of the 22nd Amendment, a
party had won three consecutive elections only once – in 1988, when George H.W.
Bush succeeded Reagan. However, Reagan had won landslide victories in 1980
and 1984 and he left office four years later as a highly popular incumbent. The
younger Bush, however, had been elected only narrowly in 2000 and 2004 and
his popularity among voters eroded throughout his second term. By the beginning
of 2008, the unpopular war in Iraq had been going on for nearly five years. Many
Americans believed that the economy was weakening and expected it to continue
going downhill. Corruption scandals forced the resignation or humiliation of
several Republican members of Congress. In part because of these issues, Bush’s
popularity rating among the American electorate remained low throughout most
of his second term.
Throughout most of 2007, many commentators and party professionals
expected Senator Hillary Clinton to receive the Democratic nomination. However,
the perceived likelihood of a Democratic victory in 2008 induced several other
candidates to contest the Democratic Party’s presidential nomination. The
Democratic field included Senators Joseph Biden of Delaware, Christopher Dodd
of Connecticut, and Barack Obama of Illinois, former Senator John Edwards
of North Carolina, Governor Bill Richardson of New Mexico, and two fringe
candidates, Representative Dennis Kucinich of Ohio and former Governor Mike
Gravel of Alaska.
Biden, Dodd, and Richardson brought extensive governmental experience to
their campaigns. Biden, who would eventually be selected as Obama’s running
mate, had served in the Senate for 35 years and was Chair of the Senate Foreign
Relations Committee. Dodd had served in the Senate since 1980 and had previously
served in the House of Representatives. Richardson had also served in the House
of Representatives for 14 years, and had also served as the US representative to the
United Nations and as Secretary of Energy under President Bill Clinton.
Clinton, Obama, and Edwards had had much less formal governmental
experience. Although Clinton emphasized her experience during her campaign,
her formal governmental experience was limited to seven years in the Senate.
Edwards had held public office for only six years, serving a single term in the
Senate before running unsuccessfully for Vice President in 2004. Obama had
been elected to the Senate in 2004, and his previous political experience had
been limited to eight years in the Illinois State Senate. Nevertheless, as 2007
ended polls showed that Clinton, Obama, and Edwards were most popular among
Democratic voters. Obama and Edwards in particular emphasized the need for
change in their campaigns, and it became clear as the campaign continued that
the desire for change resonated more with many voters than did the experience
factor. Obama also emphasized his opposition to the war in Iraq, criticizing
Biden, Clinton, Dodd, and Edwards for having voted in favor of authorizing
expenditures to support it in 2003. (At that time, Obama had not yet been elected
to the Senate.)
The Iowa caucuses and the New Hampshire primary began to winnow down the
field. Huckabee, with strong support from evangelical voters, won the most votes
and delegates in the Republican caucuses in Iowa on January 3, 2008. Huckabee
won 34 percent of the vote and carried 74 of Iowa’s 99 counties. Romney finished
second with 25 percent of the vote and carried 24 counties. On the Democratic
side, Obama won with 38 percent of the vote. Edwards with 29.7 percent edged
Clinton with 29.4 percent for second place. Richardson, Biden, and Dodd split less
than three percent of the vote and dropped out of the race.
The New Hampshire primary took place five days later, on January 8. On
the Democratic side, a poll taken shortly after Obama won the Iowa caucuses
showed him with a 13-point lead over Clinton (Page 2008). However, Clinton
won a narrow victory over Obama. Clinton won 39 percent of the vote, Obama
won 36 percent, and Edwards won 16 percent (Shelley 2008). Later in January,
Obama easily defeated Clinton and Edwards in South Carolina, where Edwards
had been born. He subsequently left the race, leaving Clinton and Obama as the
last contenders for the Democratic nomination. Meanwhile, McCain defeated
Romney, 37 percent to 32 percent, in the Republican primary. Huckabee followed
with 11 percent, and Giuliani finished fourth with 9 percent. A week later, the
Republicans held a primary in Michigan. Romney, a Michigan native whose father
had once served as that state’s Governor, defeated McCain by a 39 percent–30
percent margin. McCain also won the primary in Florida, where Giuliani had
campaigned extensively but got few votes. Shortly after the Florida primary,
Giuliani dropped out of the race and endorsed McCain.
Twenty states held primaries on “Super Tuesday” on February 5. McCain won
several key Republican primaries and took a commanding lead in the delegate
count. After Super Tuesday, McCain’s nomination was a foregone conclusion.
However, Huckabee remained in the race for a few more weeks until McCain had
won enough delegates to formally clinch the nomination. Attention then shifted
to the Democratic Party. Clinton and Obama divided the Super Tuesday states.
Clinton won primaries in California, New Jersey, Massachusetts and her home
state of New York, while Obama won in Missouri, Georgia, Connecticut and his
home state of Illinois. Smaller states that also voted on Super Tuesday divided
evenly between Clinton and Obama, and political commentators declared Super
Tuesday a draw.
After Super Tuesday, Obama picked up momentum by winning primaries in
Louisiana, Maryland, Virginia, the District of Columbia, and Wisconsin throughout
the month of February. Obama’s success encouraged some superdelegates who had
originally supported Clinton to reconsider their commitments. Clinton rebounded
with victories in Ohio and Texas on March 4, while Obama won Vermont and
Rhode Island on the same day. By this time, Obama had a narrow but significant
lead in the Democratic delegate count. Clinton won Pennsylvania in late April and
narrowly won Indiana two weeks later, but Obama won a solid victory in North
Carolina on the same day. By this time, a majority of the Democratic superdelegates
had declared support for Obama, increasing his lead in the Democratic delegate
count. Even though Clinton won late primaries in West Virginia, Kentucky, and
South Dakota, Obama was too far ahead in the delegate count for Clinton to catch
up. Obama sewed up the nomination on June 3 with a victory in Montana, and
Clinton formally suspended her campaign the following day.
Analysis
Numerous studies have been devoted to cartographic and statistical analysis of

mapped results of individual elections or electoral sequences. Over many years,
this line of research has provided valuable insights into understanding the political
economies and cultures of the United States and other countries. Of course,
the shortcomings of electoral geography have long been recognized, including
the possibility of the ecological fallacy, associated difficulties in assuming or
determining motivations of voters on the basis of electoral data, and the assumption
that explanatory variables represent meaningful surrogates for the characteristics
of individual voters.
These difficulties associated with interpretation of geographically
disaggregated election results may be even more problematic when applied
to US primary elections, for several reasons. Part of the problem lies with the
nature of the primary election process itself. As we have seen, the 2008 primary
season lasted five months. Issues of concern to voters in early primaries receded
by later primaries, and vice versa. Those candidates who did not do well in
early primaries dropped out of the race, reducing the set of choices available to
voters who happened to live in states holding later primaries. The Republican
race was essentially over after Super Tuesday, after which it was clear that
McCain would be the Republican nominee regardless of the outcomes of
later primaries. Thus, few insights can be gained from analysis of the later
Republican primaries because their outcomes had no effect on the overall race
for the Republican presidential nomination.
Another problem associated particularly with primary elections is that the
electorate participating in each primary represents only a subset of a place’s total
electorate. A voter can participate in the Democratic primary or the Republican
primary, but not both. The demographic, ethnic, and economic characteristics of
voters participating in each party’s primary can vary considerably. For example,
according to the 2000 US census about 11.5 percent of Missouri’s population was
African-American. Because a large majority of African-Americans in Missouri
and other states are Democrats, according to exit polls African-Americans
represented nearly 20 percent of the Democratic Party’s primary electorate in

Missouri. However, less than 3 percent of the Republican voters of Missouri
were African-Americans. Because the overwhelming majority of Missouri’s
Republican primary voters were Euro-Americans, using race as an explanatory
variable to explain levels of support for Republican candidates is useless. Data
from the US Census and other government sources describe the characteristics of
a place’s population but do not break these data down between Democratic and
Republican primaries.
Recognizing these pitfalls, we nevertheless conducted county-level analysis
of Republican and Democratic primary elections in various states. Statistical
analyses, comparison of counties that were carried by different candidates, and
state-level exit polls were used to shed light on geographic differences within
and between states in primary election outcomes. Through detailed analysis of
the Super Tuesday primary elections in Missouri and Oklahoma it was possible
to make connections between the geographical make-up of the voters and their
constituent selections. These states were chosen because both primaries were
very closely contested between three leading Republican contenders – McCain,
Huckabee, and Romney.
Although Missouri and Oklahoma share a boundary, Missouri has long
been a bellwether. Its electoral votes have gone to the national winner of every
presidential election since 1904, with the exceptions of 1956 and 2008, in which
McCain won Missouri’s electoral votes by a margin of less than 4,000 popular
votes. Missouri’s demography closely parallels that of the US as a whole. Missouri
is 11.5 percent African-American. Its per capita income in 1999 was $19,936;
13.3 percent of its residents were over 65, and 13.0 percent were living in poverty
in 2000. These figures for the US as a whole were $21,387, 12.4 percent, and
12.7 percent respectively. Missouri’s population density of 81.2 persons per
square mile is similar to the 79.6 persons per square mile across the United States.
Thus, the demography and economic status of Missouri closely parallel that of
the US as a whole. Oklahoma, on the other hand, is a strongly “red” state that has
supported the Republican nominee for President in every election since 1952, with
the exception of 1964. Relative to Missouri, Oklahoma is more rural, has fewer
minorities, and has a lower per capita income.
Although McCain won both states’ Republican primaries narrowly, in both
cases Huckabee won a majority of counties. Huckabee won a plurality in 72
counties in Missouri, with 30 for McCain and 13 for Romney (Figure 10.1).
Counties carried by all three candidates in Missouri were geographically
concentrated. Nine of Romney’s 13 counties were located in the Kansas City
metropolitan area. McCain was strongest in more rural counties around Kansas
City, and in the St. Louis metropolitan area. Huckabee swept the southern portion
of Missouri (which borders his home state of Arkansas) and in northeastern
Missouri, a region sometimes known as “Little Dixie” because of the Southern
origins of many of the region’s early settlers (Hurt 1992).
Counties carried by the three candidates in Missouri show very different

characteristics. The counties carried by Romney tended to grow more rapidly
and contain younger, better-educated, and higher-income voters. The median per-
capita income of Romney’s counties was $19,844 as compared to $16,197 for
McCain and $14,885 for Huckabee. The median percentage of college graduates
was 23.4 percent for Romney counties as compared to 12.1 percent for McCain
counties and 10.8 percent for Huckabee counties. The median population density
was 175.7 persons per square mile in Romney counties compared to 29.5 for
McCain counties and 29.0 for Huckabee counties. Thus, Romney did better in
urbanized as opposed to rural areas. These data suggest that Romney’s appeal
in Missouri was greatest among younger, more upscale, and better-educated
Republicans. This is corroborated by positive correlations between county
percentages for Romney with per capita income (r=0.68) and percentage of
residents holding college degrees (r=0.57). Levels of support for Bush in 2004
were similar among all three groups of counties; the median Bush percentage in
2004 was 64 percent for Huckabee counties, 62 percent for McCain counties, and
59 percent for Romney counties.
In Oklahoma, Huckabee outpolled McCain in 39 of the state’s 77 counties,
with McCain winning the remaining 38 (Figure 10.2). Romney, who came in
third, won no counties in Oklahoma. Huckabee was strongest in the poorer
southern and eastern portion of the state, while McCain did best in the wealthier
urban areas and in the rural western part of the state. As in Missouri, McCain’s
counties tended to have higher incomes and levels of education. McCain also
did better among elderly Republican voters, as demonstrated in exit polls.
However, the counties carried by McCain were more strongly pro-Bush in 2004,
in contrast to Missouri where Huckabee’s support was correlated positively with
support for Bush.
On the Democratic side, the closest Super Tuesday primary was in Missouri,
which Obama carried by a margin of only 11,000 out of more than 800,000
votes cast. Interestingly, Obama won his majority in Missouri despite carrying
only five of the state’s 114 counties along with the independent city of St. Louis.
The five counties carried by Obama included St. Louis County, which includes
many of the inner suburbs of St. Louis; Jackson County (Kansas City); Boone
County (Columbia and the University of Missouri); Cole County (the state capital
of Jefferson City), and Nodaway County (Maryville and Northwest Missouri
State University). Obama’s margins in the City of St. Louis, St. Louis County,
and Jackson County – the three most populous local jurisdictions in the state –
overcame Clinton’s majority elsewhere.
The Obama counties contain substantial populations of African-Americans,
students, and professionals who made up the core of Obama’s support. The City
of St Louis, which Obama won by a 71–29 margin, is 50.5 percent African-
American, and St Louis and Jackson Counties are both over 20 percent African-
American. St Louis County is Missouri’s highest-income county, with a per
capita income of $27,595. Statewide, the county-level correlation between the
Obama vote and per capita income was 0.63, and the correlation with population
density was 0.54. In Missouri, Clinton did best in lower-income counties with
less educated populations – a pattern that would become even more clear in later
primaries.
As we have seen, Obama surged ahead in the weeks after Super Tuesday,
winning primaries in several states later in February. In our analysis, we focused
especially on the states of Virginia and Wisconsin. Obama won the Virginia
primaries, along with primaries in neighboring Maryland and the District of
Columbia, on February 12. He won the Wisconsin primary a week later, on
February 19. In Virginia, Obama got 627,820 votes (64 percent) as compared
with 349,766 for Clinton (35 percent) (Figure 10.3). Obama won Wisconsin over
Clinton by a vote of 646,851 (58 percent) to 453,954 (41 percent) (Figure 10.4).
Although Obama won both Virginia and Wisconsin decisively, the geographic
patterns of support for Obama and Clinton in these two states were strikingly
different. Virginia’s Democratic electorate was highly polarized. Obama won the
primary by sweeping the state’s most populous areas including the suburbs of
Washington, DC, and the Richmond and Norfolk/Virginia Beach metropolitan
areas. Clinton won large majorities in rural areas of western Virginia. Statewide,
Virginia has 114 counties and independent cities. Obama won 75 percent or more
Figure 10.3 Map Showing the Results of Virginia Democratic Primary, 2008
Figure 10.4 Map Showing the Results of Wisconsin Democratic Primary, 2008
of the vote in 14 of these jurisdictions, and he won less than 25 percent of the vote
in 11 others. Most of the 14 Obama jurisdictions have large African-American
populations. Virginia’s overall population was 19.9 percent African-American in
2000. All but one of the 14 Obama jurisdictions were at least 43 percent African-
American. The exception was the city of Charlottesville, home of the University of
Virginia. All but one had fewer elderly persons than the statewide average. Eight of
the 14 were central cities with high population densities (Charlottesville, Danville,
Emporia, Hampton, Newport News, Petersburg, Portsmouth, and Richmond).
Only two of the 14 were carried by Bush in 2004.
The 11 Clinton jurisdictions included nine counties and two small independent
cities, Norton and Wise. All of these are located in the Appalachian area of
western Virginia. The population densities of the nine counties are all less than
half the statewide density of 178.8 persons per square mile. All 11 Clinton
jurisdictions had at least 14 percent of their populations over 65, as compared
to 11.6 percent statewide. Education levels in these places were also strikingly
low. Statewide, 29.5 percent of Virginia residents over 25 held college degrees;
the highest total in the 11 jurisdictions was 14 percent and less than 10 percent
held college degrees in seven of the remaining ten jurisdictions. All 11 had per-
capita incomes less than $18,000, as compared to a statewide per capita income
of $23,975. Relative to the rest of Virginia, the Clinton jurisdictions were low-
income, elderly, and rural with low education levels and, in most cases, economic
and population stagnation.
We also compared the Obama and Clinton jurisdictions to the four counties
(Arlington, Fairfax, Loudoun, and Prince William) that include most of the suburbs
of Washington, DC. Obama won all four easily, with more than 60 percent of the
vote in each despite the fact that none are more than 10 percent African-American
except for Prince William, the most exurban of the four. Relative to the rest of the
state, these places have high per capita incomes and high levels of education. More
than half of the adults in Arlington and Fairfax Counties held college degrees, and
per capita income in these counties was more than $37,000 per year. Loudoun
County, which has grown by 58 percent since 2000, is one of the fastest-growing
counties in the United States. Not surprisingly, only 5.6 percent of Loudoun’s
residents are over 65. These data, along with data from the strongest Obama and
Clinton jurisdictions, confirmed that Obama’s base of support in Virginia came
from upscale, well-educated voters along with African-Americans. These upscale
voters along with African-Americans helped Obama become the first Democrat
since Lyndon Johnson in 1964 to win Virginia’s electoral votes in the November
general election.
The Wisconsin primary, held a week later, yielded a strikingly different outcome
when examined geographically. In contrast to Virginia, there was relatively little
difference between counties in the percentage of Obama’s vote. Obama’s strongest
counties were Dane County, home of the state capital of Madison and the University
of Wisconsin-Madison, and Milwaukee County, whose population includes nearly
two-thirds of the state’s approximately 300,000 African-Americans (who make
up about a quarter of the county’s population). Obama won 68 percent of the vote
in Dane County and 64 percent in Milwaukee County. His weakest counties were
rural Douglas County (42 percent) and Forest County (43 percent). Obama carried
62 of Wisconsin’s 72 counties, but won more than 60 percent of the vote in only
seven of them.
Despite much smaller differences in vote percentage between counties in

Wisconsin relative to Virginia, there were similar patterns of demography between
the two states. All ten counties carried by Clinton had lower population densities,
per capita incomes, and percentages of college graduates, and higher percentages
of elderly residents, than statewide totals. The five counties carried by Obama with
more than 60 percent of the vote, in addition to Dane and Milwaukee, included
counties containing three other branches of the University of Wisconsin system
(Eau Claire, Oshkosh, and Whitewater). Except for the large campuses in Madison
and Milwaukee, these are three of the four largest campuses in the state university
system. All but one of the seven Obama counties had fewer elderly people than
the statewide total and all but two had higher per capita incomes. In a much more
subtle way, the Wisconsin data reflect Obama’s strength among younger voters,
African-Americans, and in urban and upscale communities. Clinton did better
among less upscale, less educated, less wealthy, and older voters.
This pattern was even clearer in Pennsylvania, whose Democratic primary took
place on April 22 after six weeks without any primary elections at all. Clinton won
Pennsylvania easily, getting 1,275,039 votes (55 percent) to 1,061,441 (45 percent)
for Obama. Clinton won 60 of Pennsylvania’s 67 counties. The seven Obama-
majority counties included Philadelphia County, suburban Delaware and Chester
Counties, Centre County (State College and Pennsylvania State University), and
Dauphin County (the state capital of Harrisburg). Obama’s strength among young
voters, African-Americans, and upscale voters was evident in these places as well.
In contrast, Clinton won 72 percent or more of the vote in 12 counties, many of
which were located in rural central Pennsylvania or in the Pittsburgh metropolitan
area. All but one of these counties had more elderly people than the statewide
total of 15.2 percent. All but one have lost population since 2000, and all 12 have
less than the statewide African-American percentage of 10.7 percent. All 12 also
have lower per capita incomes than the state per capita income of $20,880. Thus,
Clinton’s strength was greater among more rural, less educated, and older voters
in Pennsylvania (York 2008). This was reinforced later in the campaign when she
won the more rural states of West Virginia and Kentucky easily.
Conclusion
Primary elections provide a useful and fascinating but little explored source
of information about the geography of the American electorate, although the
peculiarities associated with the American primary election system can limit their
usefulness. Nevertheless, state-by-state comparison of factors associated with
primary election outcomes can shed valuable light on voter preferences among
candidates for major-party nominations. The results of analyses of primaries in
Missouri for both major parties, Oklahoma for the Republicans, and Virginia and
Wisconsin for the Democrats demonstrated the appeal of Hillary Clinton and
Mike Huckabee to rural, downscale voters whereas more upscale and urbane
voters tended to support Barack Obama and John McCain, who eventually won
their respective parties’ nominations. In future elections, primary elections should
continue to provide electoral geographers with valuable information about voter
preferences.
References
Barnes, F. 2005. President Cheney? The Weekly Standard, 10:23. Available at:
http://www.weeklystandard.com/Content/Public/Articles/000/000/005
/295zkwpw.asp.
Barone, M. 1990. Our Country: The Shaping of America from Roosevelt to
Reagan. New York: Free Press.
Gregg, H., and W. Gardner. 2000. Why New Hampshire? The First-in-the-Nation
Primary State. Enfield, NH: Enfield Publishing and Distribution Company.
Hull, C.C. 2007. Grassroots Rules: How the Iowa Caucus Helps Elect American
Presidents. Palo Alto, CA: Stanford University Press.
Hurt, R.D. 1992. Agriculture and Slavery in Missouri’s Little Dixie. Columbia,
MO: University of Missouri Press.
Key, V.O., Jr. 1949. Southern Politics in State and Nation. New York: Vintage
Books.
Page, S. 2008. Obama, McCain Leap Ahead in New Hampshire. USA Today,
January 6.
Punnett, R. 1993. Selecting the Party Leader in Britain: A Limited Participatory
Revolution. European Journal of Political Research, 24, 257–76
Shelley, F.M. 2008. Sections and Elections: Reflections on the Early 2008 US
Presidential Primaries. Political Geography, 27, 505–8
Shugart, M.S. 2004. The American Process of Selecting a President: A Comparative
Perspective. Presidential Studies Quarterly, 34, 632–55.
Squire, P. (ed.) 1989. The Iowa Caucuses and the 1988 Nominating Process.
Boulder, CO: Westview Press.
Ware, A. 2002. The American Direct Primary. New York: Cambridge University
Press.
White, T.H. 1969. The Making of the President 1968. New York: Atheneum.
York, B. 2008. Hillary in Bitterland: Clinton Finds Her People in Pennsylvania.
National Review Online. Available at: http://article.nationalreview.com/355
050/hillary-in-bitterland/byron-york.
Chapter 11
Electoral Alignments and Place-Based
Cleavages in Statewide Votes in Alabama
Nicholas Quinton and Gerald R. Webster
Geographers have fixated upon contextualized accounts of space as leaders in the

broader spatial turn in the social sciences (Warf and Arias 2008). This change in
approach has been enlivened by calls to link various and potentially contradictory
approaches to social inquiry. In particular, geographers have shown interest in the
connection between modern hegemonic approaches to social inquiry and critical
(though modern) interpretations of hegemony, and what post-structural social
investigation means to both. Barnes (2009) argued that there is no fundamental
reason why quantitative methodologies cannot be applied to critical geography. The
argument was simply that decidedly hegemonic modern scientific approaches can
(and should) be applied with critical interpretations of the world in a conciliatory
approach to social inquiry.
The turn to conciliatory approaches has been embraced by recent changes
in the practice of electoral geography. Electoral geographers have combined
traditional statistical methodologies with qualitative discussions of context in
mixed-methods approaches (Brown, Knopp, and Morrill 2005; Chapman, Leib,
and Webster 2007; Morrill, Knopp, and Brown 2007; Shin and Agnew 2002, 2008;
Webster, Chapman, and Leib 2010; Webster and Leib 2001, 2002). The general
application of a conciliatory mode of social inquiry is central to the present work.
In particular the connection between quantitative approaches and context centered
investigation is addressed in the linkage between particular elections and their
spatial vote patterns.
Accounting for Context
During the past two decades electoral geographers have largely concentrated their
research efforts on partisan elections and voting systems, with comparatively little
work on non-partisan votes including referenda and initiatives. Through study of
referenda new insights on the juncture of context and politics can be made. As
stated by Toal and Shelley (2003: 175):
Analysis of the geographic distribution of initiatives, referenda, and other

direct democratic processes has often proven a particularly valuable source of
information to political geographers, because under direct democracy voters

are expressing opinions on individual policy issues. Direct democracy provides
especially valuable information about cultural and identity politics, and is
therefore critical to the understanding of social and cultural linkages to political
processes.
In the United States, for example, state legislatures can put referenda on the
ballot for constitutional amendments in all states, and in two dozen states citizens
may bypass the legislature and place changes to state statutes or amendments to
state constitutions directly on the ballot. In 2010, 155 such measures were on
the ballot in 36 different states. While 42 of these measures were placed on the
ballot by citizens through the initiative process, 113 were placed on the ballot by
state legislatures as referenda (National Conference of State Legislatures 2010).
These measures pertained to a wide array of topics including taxes, state budgets,
immigration, redistricting, abortion, stem cell research, marijuana and alcohol
policy, among others. Such ballot measures ask citizens for a simple yes or no
vote, and in conjunction with partisan elections, offer a more contextualized view
of electoral geographies.
Over the past three decades “place” as an ontological unit tailored to electoral
geography has also been extended (Agnew 1987, 1994, 1995, 1996, 1997;
Agnew and Duncan 1989; Shin 2001; Shin and Agnew 2002, 2008). In these
discussions, ballot preferences are attributed to the voter’s social milieu rather
than their demographic characteristics. Thus, voters from the Wire Grass region
in southeastern Alabama may be considered an important contextual distinction
in place rather than as simply a pocket of white middle class voters. Context,
therefore, is engaged through the application of place as an ontological unit.
These developments in place-based electoral geography have largely been
left unaddressed with quantitative methods, though recent publications indicate
a shift in focus. Several authors (Chapman, Leib, and Webster 2007; Webster,
Chapman, and Leib 2010; Webster and Leib 2001, 2002; Shin and Agnew 2002,
2008; Brown, Knopp, and Morrill 2005; Morrill, Knopp, and Brown 2007)
demonstrate the move to a more contextualized quantitative electoral geography.
These studies incorporate context inspired discussions of place-based electoral
geographies alongside quantitative methods. Context, however, is often excluded
from the quantitative models employed in these studies. The statistics in question
have been used to extract spatial similarities associated with demographic
profiles of selected electorates. In these methods place becomes an addendum to
quantitative analysis rather than a fundamental component of it. The consequence
is a normative stance in favor of quantitatively defined social similarity relative to
place-based social difference, and the sacrifice of context as a central tenet within
quantitative electoral geographies. This chapter seeks to address this bias with the
incorporation of quantitatively defined place-based difference utilizing the local
indicator of spatial autocorrelation (LISA) statistic. We follow the lead of Shin
Electoral Alignments and Place-Based Cleavages in Statewide Votes in Alabama 197
and Agnew (2002, 2008), Darmofal (2006), and (Seabrook 2009) in the use of this
statistical approach in analysis of election returns.
The purpose of this chapter, then, is to extend the incorporation of context in
electoral geography along three planes: the contrasts between partisan elections
and referenda, in places, and with quantitative methodology. This chapter continues
with a brief discussion of the 12 statewide votes examined, and is followed by a
section detailing how we quantitatively defined Alabama’s place-based electoral
geographies. We follow this section with an overview of the political temperament
associated with each place by relating their roles in the statewide votes. We
conclude with a commentary relating what this effort adds to discussions regarding
the nexus of context and quantitative methods as it applies to Alabama’s electoral
geographies.
Statewide Votes
We examined six statewide referenda and six statewide partisan elections in

Alabama during the past dozen years. The referenda were legislatively placed
on the statewide ballot as amendments to the state constitution between 1999
and 2006. They involved gambling, interracial marriage, tax reform, segregated
schools, and the prohibition of same-sex marriage. All of these measures may
be viewed as pertaining to culture war debates, at least from the lens of many in
the Alabama electorate, with issues of morality being central to the decisions of
many voters. The six recent statewide partisan elections included the presidential
elections of 2000, 2004, and 2008, and the gubernatorial races of 1998, 2002 and
2006 (see Figure 11.1). A brief discussion of each of these statewide votes follows.
1998 Gubernatorial Election
The 1998 Alabama gubernatorial election pitted incumbent Republican Fob

James against Democrat and sitting Lieutenant Governor Don Siegelman. James
had won the 1994 gubernatorial election against the incumbent Democrat James
Folsom, Jr. by less than 1 percent of the vote. Folsom’s opposition to replacing the
Confederate Battle flag on top of the state capitol played a role in James’ narrow
victory (Webster and Leib 2002). James’ term had been controversial on several
accounts, particularly in his treatment of education. He had also entered litigation
arguing that student preachers had the right to engage in religious activity in the
state’s public schools, going so far as to submit a brief to the Supreme Court
holding that only the federal government, and not the states, were bound by the
Bill of Rights and the separation of church and state (Shelley and Webster 2000).
While such efforts ingratiated James to the state’s large fundamentalist religious
community, they led to growing dissatisfaction in the business wing of the
Republican Party.
Figure 11.1 Partisan Elections

Siegelman’s centerpiece campaign issue was the creation of an education

lottery which used the proceeds to provide accomplished high school students
with scholarships to Alabama colleges and universities. Dissatisfaction with
James and interest in the lottery issue led Siegelman to victory with nearly 59
percent of the vote (Beyerle 1998b). Notably, there was record turnout in the
election with 57 percent of registered voters casting ballots. Nearly 120,000 more
Alabamians voted in the 1998 gubernatorial election than in 1994. Turnout in
the state’s predominately African-American counties was exceptionally high. In
Sumter County which is 73 percent African-American, turnout was 74 percent.
Similarly in Greene County which is 80 percent African-American, turnout was
78 percent (Beyerle 1998a).
Democrat Siegelman ran for re-election in 2002. His Republican opponent was
former Congressman Bob Riley. Polls prior to the November 5 election indicated
a close contest. On election night Siegelman led and called on his opponent to
concede. But after most returns were in, Republican stronghold Baldwin County
reported a computer error and moved 7,000 votes from Siegelman to Riley. The
result was that Siegelman fell behind by 3,117 votes out of the 1.3 million cast
statewide, or 0.23 percent (Associated Press 2002). Alabama has no automatic
recount statute, but Siegelman and his supporters pursued a statewide recount for
two weeks without success. The Baldwin County district attorney, a Republican,
“threatened to put everyone in jail” who tried to conduct a recount (Associated
Press 2002). Finding few if any options to carry out a recount, Siegelman conceded
on November 18 saying he did not wish to put the state through months of legal
wrangling. As a result, Riley won the election with slightly over 50 percent of
the vote.
The 2006 gubernatorial election pitted incumbent Republican Riley against

Democratic Lieutenant Governor Lucy Baxley. Both candidates faced notable
opposition in the primaries with Riley winning the nomination against the
former Chief Justice of the Alabama Supreme Court, and Baxley besting former
Democratic Governor Don Siegelman. During the campaign Siegelman was being
tried on racketeering charges making an active run all but impossible. Although
Riley came under fire by his own party early in his first term due to his support
for tax reform (Webster and Webster 2004), he slowly repaired his relationships
with both social and business conservatives in the Alabama Republican Party. He
further elevated his status due to his handling of the Hurricane Katrina relief effort
in 2005. As a result, Riley easily won re-election with 55 percent of the vote.
2000 and 2004 Presidential Elections
During the past half century, Alabama has become a reliably Republican state in
presidential elections with Georgian Jimmy Carter the only Democrat to win the
state since John F. Kennedy in 1960. In 2000, George W. Bush won 58 percent
of the state’s vote, and increased his proportion to 63 percent in 2004. In 2000,
Al Gore won only 17 of the state’s 67 counties receiving large vote margins in
the state’s majority African-American counties. In Macon County, which is 85
percent African-American, Gore received over 87 percent of the vote. But in
overwhelming white and suburban Shelby County, Gore received less than 22
percent of the 2000 presidential vote. In 2004, only 11 Alabama counties provided
a majority of their votes for John Kerry, and all were in the Alabama Black Belt.
Overwhelmingly African-American Macon County provided Kerry 83 percent
of its vote, but overwhelmingly white Shelby County cast less than 20 percent
of its ballots for Democrat Kerry. Thus, the geographic patterns of support and
opposition to each Bush run were similar.
2008 Presidential Election
In the 2008 presidential election, Alabama again provided overwhelming support

for the Republican candidate, John McCain. Although McCain garnered 45.6
percent of the nationwide vote, 60.3 percent of Alabama voters cast ballots for the
Republican nominee. Barack Obama won 38.7 percent of the vote in Alabama in
2008, 2 percent better than John Kerry in 2004, but nearly 3 percent less than Al
Gore in 2000. Obama won only 15 of Alabama’s 67 counties with all but Jefferson
County (Birmingham) being in the Alabama Black Belt – Kerry had won 11
counties in 2004, and Gore 17 in 2000. John McCain won 18 of the state’s counties
with more than 75 percent of the popular vote, with most of these counties being
located in overwhelmingly white suburbs of Birmingham and the rural, principally
white, counties in north central Alabama.
Exit polling after the 2008 presidential election indicated that Barack Obama
won 43 percent of the white vote nationwide, equal to Bill Clinton’s proportion in
1996, and the highest proportion for a Democratic nominee since 1976 (Edge 2010:
429). In comparison, Democrat Al Gore received 42 percent of the white vote in
2000, and John Kerry 41 percent in 2004 (Edge 2010: 430). In spite of Obama’s
comparatively successful results with white voters nationwide, he received support
from only 10 percent of Alabama white voters in 2008, while garnering 98 percent
support from the state’s African-American voters (CNN 2008). White voter
support in Alabama for Obama was the lowest in the country (Mississippi whites
provided 11 percent of their support for Obama). In comparison, McCain received
the support of 88 percent of white voters and only 2 percent support among black
voters. Calling into question the pronouncements of some commentators about
Figure 11.2 Amendments

the development of a “post-racial” electorate, Obama actually did worse among

Alabama whites than Massachusetts Senator John Kerry in 2004 (Edge 2010: 437).
Education Lottery, 1999
As noted earlier, Don Siegelman won the 1998 gubernatorial election in part due
to enthusiasm for his campaign to create an education lottery that would provide
the state’s high school students scholarships to attend Alabama colleges and
universities. Students with B averages or above would be eligible for funding
to attend any of the state’s four year institutions, and all high school graduates
would receive support to attend one of the state’s community colleges regardless
of their grade point averages. Proceeds from the lottery were also to be used for
voluntary pre-kindergarten programs and to purchase computers for the public
schools. Due to the funding problems faced by Alabama schools, initial support
for the lottery was substantial. A poll in August 1999, only six week before the
referenda was held, found 61 percent of the respondents were supportive of the
proposal (DeMonia 1999). As the vote neared, the state’s religious community
mounted a vigorous campaign against the referendum’s passage. The Alabama
Christian Coalition used telephone messages and radio ads to oppose the lottery,
and distributed anti-gambling pamphlets to the state’s churches. The day before
the vote ministers opposing the lottery held a 24-hour prayer vigil outside the state
Capitol, and reiterated that “gambling goes against God’s will” (CNN.com 1999).
The efforts of religious leaders and organizations were successful and the lottery
went down to defeat with over 54 percent of the Alabama electorate voting “no”
on October 12, 1999.
Interracial Marriage, 2000
Alabama continues to operate under a constitution that was passed in 1901.

The central goal of the century old constitution was to suppress African-
American political power, and to insulate the state’s large plantation owners
and industrialists from taxation. The constitution severely limited property
taxes, mandated segregated schools, implemented poll taxes and literacy tests,
and outlawed interracial marriage. In 1967 the US Supreme Court found that
prohibitions against interracial marriage were unconstitutional, making this
section of Alabama’s constitution a legal relict. In 1998, South Carolina deleted
a similar section in its state constitution leaving Alabama as the sole remaining
state with such a prohibition. In 1999, the Alabama legislature passed a bill to
place a referendum on the ballot in 2000 that would strike the prohibition from
the state’s constitution. There was uniform support for passage of the referendum
by the state’s political leaders, and the only public opposition came from neo-
Confederate groups. On November 8 the amendment passed with over 59 percent
of the vote. But it is notable that a majority of voters in 25 of the state’s 67 counties
voted against excising the passage from the state’s constitution. A Birmingham-
Southern College political scientist estimated that only 44 percent of the state’s
white voters cast ballots in favor of the amendment (Webster and Quinton 2010).
Tax Reform, 2003
Upon his election to the governor’s mansion in November 2002, Republican Bob
Riley was faced with an extreme financial emergency in the state’s budget. Riley,
who was known as a strict fiscal conservative, proposed that a broad tax package
to be submitted to the voters in September 2003. Alabama’s tax system is one of
the most antiquated in the United States, and its property tax rates are the lowest
in the country. As a result, public education is funded with sales taxes which are
famously unpredictable from year to year. Riley’s tax reform package altered sales
tax rates by eradicating the levy on food and drugs, reduced income taxes on low
income earners while raising them modestly on upper income households, and
raised property taxes on all categories of holdings. The plan would have raised
over $200 million in new revenue, but it was estimated that as many as two-thirds
of Alabama households would have experienced either no change or a reduction in
their annual tax payments to the state (Webster and Webster 2004).
Republican responses to Riley’s proposals were rapid and harsh, with the
Alabama Republican Party Steering Committee voting overwhelmingly to oppose
the Republican Governor’s proposal. Farmers’ groups, business groups, anti-tax
groups and religious groups also strongly opposed the passage of the tax reform
package. Notably, the National Christian Coalition came out in support of Riley
because it reduced the regressive character of Alabama’s tax system but the
Alabama Christian Coalition vigorously opposed the package as a tax increase
(Webster and Webster 2004). Education groups, African-American leaders and
some business groups supported the tax reform package. On September 9, 2003,
68 percent of Alabama voters cast “no” votes on Riley’s tax reform package. Only
13 of the state’s 67 counties voted in favor of the amendment, with all being in or
adjacent to the Alabama Black Belt.
Vote to Delete Racist Language from the State Constitution, 2004
Alabama continues to operate under its 1901 state constitution that was passed
largely to limit the political power of African-American voters. The state’s
constitution included sections requiring poll taxes and segregated schools. The
state constitution was also amended in 1956 in the wake of the Brown vs. the
Board of Education decision stating that there was no “right” to a public education
in Alabama at taxpayer expense. This amendment was passed to underscore the
state’s intention to shut down the public schools should integration be ordered by
the federal courts. Due to these unenforceable sections in the state constitution,
a bill was passed in the Alabama legislature to delete such passages through a
referenda held in November 2004.
The primary opposition to passage came from several of the state’s conservative
religious leaders. They argued that deleting the passage that there is no right to a
public education in Alabama would result in lawsuits to increase taxes to more
fully support the state’s schools. The President of the Alabama Christian Coalition
went so far as to characterize education as a “gift” from the state as opposed to a
“right”. Notably, litigation in the early 1990s had already determined that there is
a right to a public education in Alabama (Shah 2005).
In November 2004 the amendment was rejected by less than 2,000 votes
out of the 1.4 million cast, or by 0.13 percent. Less than half of the state’s 67
counties supported passage of the amendment. These counties included those
with major cities and those with large African-American populations. The
largest cluster of counties rejecting passage was located in the northern part
of the state. The population in this part of Alabama is overwhelmingly white
and includes large proportions of religious conservatives (Webster and Quinton
2010).
“Motor Vehicle Taxes Act,” 2004
The “Motor Vehicle Taxes Act” would have amended Amendment 93 of the state
Constitution of 1901 to authorize the state legislature to repeal state property
taxes on vehicles over 26,000 pounds in weight, except those used for farm and
forest production. In lieu of the ad valorem tax the legislature could instead levy
an excise tax on such vehicles. The Alabama state constitution requires that
motor vehicle excise taxes be spent on public highways, and ad valorem taxes be
allocated to the local governments in the jurisdiction in which they are collected.
Amendment 8 would have created an exception for excise taxes collected on
vehicles over 26,000 pounds to allow them to replace the ad valorem taxes lost
as a result of its passage. The excise taxes were to be apportioned based on miles
traveled inside and outside the state of Alabama.
Amendment 8 received only limited attention by the news media. Arguably,
its purpose was confusing to many. A tax lobby group reported that the Alabama
Department of Revenue and Alabama Trucking Association supported the
Amendment (Arant 2004). In opposition, the Conservative Christians of Alabama
recommended a no vote, arguing that if Amendment 8 was passed the legislature
would “have the authority to set the rate of collection and direct how this money
is spent. In times past, this has been called a slush fund” (Conservative Christians
of Alabama 2004).
Constitutional amendments in Alabama pertaining to taxes generally face
strong opposition, with many voters assuming that any change will result in a
tax increase (Webster and Webster 2004). Amendment 8 was defeated with 68.5
percent of the voters casting “no” votes. Only two counties voted in favor of
the Amendment – Greene and Sumter – both in the Alabama Black Belt. Other
surrounding counties with large African-American populations rejected the
Amendment, but with much closer results than the state as a whole. The counties
most in opposition were generally located in the northern portion of the state.
Nearly 86 percent of Limestone County and 83 percent of Fayette County voters
cast ballots in opposition.
Same-Sex Marriage Vote, 2006
The final referendum examined pertains to a constitutional prohibition against

same-sex marriage. This referendum is one of over two dozen to have been held
since November 2003 when the Massachusetts Supreme Court ruled that barring
same-sex marriage was unconstitutional. It is also notable that many of these
states, including Alabama, already had statutes defining marriage in exclusively
heterosexual terms. Although many states in the South are less than receptive
to gay and lesbian issues, Alabama has a particularly vicious record. In July,
2004 alone, two gay men were murdered because of their homosexuality. As a
result, Alabama has been labeled the “worst place to be a gay person” in the
United States, and the “homophobic center of the universe” (Webster, Chapman
and Leib 2010).
The Alabama legislature passed a bill to place an anti-same sex marriage
referendum on the ballot in 2005. Support for the bill was overwhelming, with only
limited opposition from a handful of African-American legislators. Because there
was little doubt about the vote’s outcome, there was only limited campaigning
for and against the amendment. The most visible support for the measure came
from the Alabama Christian Coalition which distributed 1.2 million voting guides
around the state urging support for the amendment. The limited opposition came
from the ACLU and Equality Alabama, the latter being a gay rights group. The
amendment passed in every Alabama county, garnering 81 percent of the statewide
vote. The counties with the lowest levels of support were found in the Alabama
Black Belt, and the amendment found its greatest support in the northern part of
the state. In both Marion and Winston counties, for example, voters cast over 90
percent of the support for the amendment.
Alabama’s Place-Based Electoral Geographies Quantitatively Defining

Context: LISA Analysis
Local indicators of spatial autocorrelation (LISA) were used to delineate the

geographies associated with our suite of statewide returns. LISA statistics were
calculated with the univariate LISA option in GeoDa. In this case the features
identified as constituent parts of LISA clusters were Alabama’s counties. Though
we do not assume that county electorates in Alabama are homogeneous for the
purposes of this study, county level vote returns provide an appropriate scale for
analysis of statewide votes. The LISA statistic indicated clusters of counties in
terms of support (high-high) or lack of support (low-low) for each Republican
candidate in the partisan elections and the conservative position for each
referendum. The LISA statistic also identified counties that are distinct outliers
from local patterns. These outliers were singled out as either highly supportive of
the Republican candidate or conservative position in an area of tepid support (high-
low), or less-supportive of the Republican candidate or conservative position in an
area of intense support (low-high) (Anselin 2005).
The calculations yielded LISA clusters that were statistically significant at
.05. Anselin (2005) notes, however, that LISA clusters often extend beyond the
statistically significant features produced by GeoDa into surrounding features.
Thus, we chose to forego the simplistic threshold of statistical significance
produced in GeoDa with an alternative cartographic representation of statistical
significance. To complete this task we assigned LISA cluster values to each
county regardless of statistical significance utilizing the Moran scatter plot utility
in GeoDa. This data was transferred into ArcGIS where the dot density tool was
utilized to display the significance measures as an element of the cartographic
presentation. Each dot represents .001 of the total significance as measured by a
P-value. As the P-value increases and the LISA value becomes less statistically
significant, the number of dots in the county increases. As a result, counties with
the most statistically significant values have few dots, while those with less
statistically significant values have high dot densities. In this way, we moved
beyond cartographic representation of statistical significance with a threshold
of .05 or better to a representation of continuous statistical significance. This
allowed us to acknowledge that highly significant features are only the core of
larger LISA clusters.
The results of the LISA analysis are included in Figures 11.3 (partisan
elections) and 11.4 (referenda). There are reoccurring spatial patterns associated
with the returns from both the partisan elections and the referenda. There are
also spatial consistencies related to support for Republican candidates that are
distinct from those found relating to the conservative positions in six referenda.
In particular the Black Belt stood out in its resistance to both Republicans and
the conservative position expressed in the referenda. Alternatively, the partisan
elections and referenda had different areas of high support, the Wiregrass and
north Alabama, respectively. Further, Jefferson, Madison, and Lee counties all
stood out as consistent outliers from local patterns. We hypothesize these spatial
cleavages are the durable underlying electoral geographies associated with the
varied contexts of recent statewide elections in Alabama.
Figure 11.3 Results of the LISA Analysis for Partisan Elections

Figure 11.4 Results of the LISA Analysis for Referenda

Table 11.1 Principal Axis Factor Analysis with an Oblique Rotation (N=67)
Variable Factor I Factor II Extraction

Loadings Loadings Communalities
Presidential Election 2000,
.952 .626 .906
% Bush
.995 .703 .994
% Bush
.905 .862 .935
% McCain
Gubernatorial Election 1998,
.974 .563 .964
% James
Gubernatorial Election 2002,
.976 .783 .980
% Riley
Gubernatorial Election 2006
.949 .556 .913
% Riley,
Education Lottery 1999,
.813 .832 .809
% No
Delete Interracial Marriage Ban 2000,
.580 .872 .761
% No
Tax Reform,
.677 .897 .814
% No
Delete Racist Language 2004,
.549 .934 .885
% No
Truck Excise Tax 2004,
.487 .868 .772
% No
Ban Same-sex Marriage 2006,
.711 .876 .794
% Yes
Extraction Sum of Squared Loadings 8.993 1.533
% of Variance 74.938 12.774
Cumulative % of Variance 74.938 87.711
Quantitatively Gauging Spatial Alignment: Factor Analysis
Given the clear geographic patterns of the LISA clusters we employed a factor
analysis as another test for the presence of durable geographies amongst Alabama’s
statewide elections (Table 11.1). We utilized an oblique rotation since there was no
expectation that two uncorrelated factors would be realistic given the similarities
in the spatial extents of the LISA clusters for the two groups of elections. Two
factors with eigenvalues above 1.0 were generated accounting for 89.7 percent of
the variance. The first factor accounted for 75.9 percent of the common variance
and was primarily associated with partisan elections, with all six having loadings
exceeding .90. Notably, the lowest loading for the six elections was for the 2008
presidential election at .905, while the remaining five partisan contests had loadings
ranging from .949 (2006 gubernatorial election) to .995 (2004 presidential election).
All six referenda had significant but lower loadings on Factor I ranging from .487
(truck excise tax) to .813 (education lottery). Thus, among the six referenda, the
lottery vote was most similar to the partisan elections.
Factor II accounted for 12.8 percent of the common variance, with generally
higher loadings by the six referenda ranging from .832 to .934. Not surprisingly given
its alignment with partisan election on Factor I, the lowest loading by the six referenda
was yielded by the education lottery vote. The six partisan elections had lower loadings
on Factor II with the exception of the 2008 presidential election. Although the other
partisan contests had loadings on Factor II ranging from .556 (1998 gubernatorial
election) to .783 (2002 gubernatorial election), the 2008 presidential election yielded
a loading of .862, higher than two of the loadings for the six referenda on Factor
II. Thus, the 2008 election had the characteristics of both a partisan contest and a
culture war related referenda. Arguably, race played a role since only one in ten white
Alabama voters cast their ballots in support of Barack Obama.
The two factors were inter-correlated at .67, indicating a high level of consistency
not only within the referenda and partisan elections as distinct categorizations, but
also between the two sets of statewide votes. Two statewide votes demonstrate this
relationship: the lottery referenda from 1999 and the presidential election of 2008.
The percent of “no” votes for the lottery amendment produced high loadings on
both Factor I (.813) and Factor II (.832). The percent of votes cast for John McCain
in 2008 also yielded high loadings on both Factor I (.905) and Factor II (.862).
The 1999 lottery referendum and 2008 presidential election returns yielded similar
geographic patterns, which included a persistent effect from the Black Belt and a
general agreement between north Alabama and the Wiregrass. These two statewide
votes were the only contests in which both north Alabama and the Wiregrass returned
strong support for the same issue or candidate. Thus the durability of the Black Belt
coupled with the shifting importance of the Wiregrass and north Alabama in the
statewide votes contributed to the high level of concurrent similarity and difference
identified in the factor analysis. These consistent differences and similarities within
the returns from both the referenda and partisan elections further confirmed our
conclusions from the LISA analysis.
Discussion of Results
Even though the context of each statewide vote was different, coherent spatial
cleavages exist amongst the election returns analyzed. We attribute this spatial
durability to identifiable differences associated with different places in Alabama –
place as the site where people live their lives and where their views of the world
are shaped (Agnew 1987, 1994, 1995, 1996, 1997; Agnew and Duncan 1989; Shin
and Agnew 2002, 2008). These places are enlivened by the context of peoples’
daily routines and reflect collective decision making in regards to the contexts
for each statewide election. Though the entirety of the electorates in each place
most certainly did not vote for the same candidates or issues, enough did so to
identify distinct spatial differences in Alabama’s electoral geographies. These
places included a major urban locale (Birmingham), the Black Belt, the Wiregrass,
counties with concentrations of creative class employment (Madison and Lee
Figure 11.5 Important Places in Alabama

counties), and the counties in the northern third of Alabama (Figure 11.5). Each of
these places has a distinct electoral character associated with past levels of support
in statewide votes (Webster 1996; Webster and Quinton 2010; Webster and Leib
2002; Webster and Webster 2004). The roles played by each, whether in support
or opposition, are discussed below.
Jefferson County – Birmingham
Jefferson County is the site of Alabama’s most populous city, Birmingham. It

played a distinct role in all of the statewide partisan elections analyzed, and in four
of the referenda included in the analysis. Jefferson County stood out in terms of its
lack of support for Republican candidates and lack of support for the conservative
positions in the lottery, interracial marriage, racist language, excise tax, and
same sex marriage referenda. Conversely, we identified Jefferson County as a
significant addition to northern Alabama in its opposition to tax reform in 2003.
While conceivably out of character given the bulk of the statewide elections we
analyzed, we attribute Jefferson County’s opposition to tax reform to the creative
class (Florida 2004). Thirty-two percent of the county’s workforce are employed
in creative class occupations. Members of the creative class generally hold more
liberal views than the population as a whole (Chapman, Leib, and Webster 2007).
However, Webster and Quinton (2010) found that the creative class in Alabama can
be influenced by economic arguments, particularly those suggesting higher taxes.
Thus, Jefferson County’s vacillation between lack of support for the conservative
position in the majority of the referenda and support of the conservative position
in the tax reform referendum can likely be attributed to the creative class concern
over tax increases.
The Black Belt
The Alabama Black Belt is a soil region running across the lower half of the state
that served as the site of the state’s Antebellum plantation economy (Webster
and Samson 1992). The Black Belt was predominantly populated by African-
Americans (Figure 11.6) in 2000, and it is largely rural. This region provided
very limited support for the Republican candidate in all of the partisan elections
examined. In so doing the Black Belt continued its longstanding support of the
Democratic Party (Webster and Bowman 2008). The Black Belt also opposed the
conservative position in all six of the referenda analyzed. While undoubtedly some
African-American residents of the Black Belt have supported the conservative
position on culture wars referenda in Alabama (Webster, Chapman, and Leib
2010), the region as a whole was demonstrably less supportive of the conservative
positions in these referenda than the state as a whole.
Figure 11.6 Percentage of African-Americans in Alabama by County
North Alabama
Race continues to be a major fault line in Alabama’s electoral geographies with

white and black voters having great contrasts in their voting behaviors. Strongly
associated with this division is religious conservatism, which characterizes many

white rural places in the state (Webster and Webster 2004). The counties located in
the northern third of Alabama are both substantially rural, largely white and stood
out in our LISA analysis as a noteworthy place in the 2008 presidential election
and all of the referenda votes. North Alabama, then, is at the forefront of the culture
wars in the state as the most noteworthy place associated with support for the
conservative position in the referenda. Conversely, north Alabama rarely showed
coherent support for partisan candidates. Our lone exception to this trend was the
2008 presidential election. The implication is that the Obama/McCain contest was
interpreted similarly to culture war issues rather than a partisan election in north
Alabama. North Alabama, then, supports the thesis that race played a substantial
role in the 2008 presidential election.
The “Wiregrass”
The far southeastern corner of the state, typically referred to as the “Wiregrass,”
is similar to northern Alabama with a largely white population residing in small
towns and rural settings. Though similar in many ways to north Alabama, the
electoral importance of the Wiregrass is different. The Wiregrass was an important
place identified in each partisan election. As such it was the only continuously
significant Republican stronghold in the state. With that said, the Wiregrass was
relatively ambivalent as a place regarding almost all of the referenda. There were
two notable exceptions to this pattern, however. First was support for passage of the
same-sex marriage ban in 2006, bringing the Wiregrass into alignment with north
Alabama on this culture war issue. Second, the Wiregrass stood out in its opposition
to a state education lottery. It is important to note that the education lottery was
the signature issue in Democrat Don Siegelman’s 1998 gubernatorial campaign,
and the fact that the Wiregrass consistently supported Republican candidates in
both gubernatorial and presidential elections. These facts notwithstanding, the
Wiregrass did align with north Alabama on these culture war issues.
Madison and Lee Counties
Recently, the creative class (Florida 2004) has become a part of electoral research
in the broader South (Chapman, Leib, and Webster 2007) including Alabama
(Webster, Chapman, and Leib 2010; Webster and Quinton 2010). Two counties
consistently stood out as distinct outliers from broader regional vote returns due to
concentrations of the creative class employment: Lee and Madison. Lee County is
the site of Auburn University and is located on the edge of the Black Belt. Madison
County is the location of the University of Alabama-Huntsville, and is also distinct
in its employment base which is associated with federally funded rocket science
(Rogers et al. 1994).
In terms of partisanship, Madison County was a significant outlier in the 2008

presidential election as a place offering relatively limited support for Senator
John McCain. Alternatively, in both the 1998 and 2006 gubernatorial elections,
Madison County was part of a small cluster of counties that provided support for
the Republican nominees. Further, Madison County was identified as a regional
outlier for its lack of support for the conservative position in the tax reform, the
deletion of racist language from the state constitution, and the large truck excise
tax referenda. These results reveal little coherence regarding Madison County’s
political temperament. We can, however, definitively conclude that Madison
County is distinct from the rest of north Alabama, and indeed is distinct from
much of Alabama more broadly speaking.
Lee County was identified as a noteworthy outlier in five of the partisan
contests: the 2000 and 2004 presidential contests and the 1998, 2002, and 2006
gubernatorial elections. In each of these contests, Lee County was significant in its
support for the Republican candidate. Interestingly, in 2008 Lee County was also a
significant place, though in this case its vote returns were similar to the Black Belt
in its lack of support for Senator McCain. When considering the six referenda,
Lee County was noteworthy because of its lack of support for the conservative
position. In these cases its vote returns were similar to those of the Black Belt.
Thus while Lee County offers more support for Republicans than the surrounding
counties of the Black Belt, it aligns with the Black Belt on the culture wars.
Conclusions
This analysis identified durable geographies associated with both the referenda and
partisan elections. These included strong support for Republican candidates in the
Wiregrass and consistent support for the conservative position in the referenda in
north Alabama. Further, Jefferson, Madison, and Lee counties were all consistently
distinct from broader regional vote patterns. These differences were intertwined
with strong similarities in Alabama’s electoral geographies evident throughout all
of the elections. In particular the Black Belt consistently opposed both Republican
candidates and the conservative positions on all six of the referenda. Thus, there
were patterned spatial alignments associated with both the partisan elections and
referenda.
With that said, these spatial patterns changed depending on both the context of
the election and that of the place. Vote returns from the Wiregrass, north Alabama,
Jefferson, Madison, and Lee counties all reflect contingent outcomes dependent
upon context. Indeed, only the Black Belt demonstrated firm durability throughout
all of the elections analyzed. But even in this case, if we go far enough back in time
we can see electoral change in the Black Belt (Webster 1992). Contingency, then,
is the dominant shape of electoral outcomes in Alabama. In this sense a focus on
context in place and associated with specific elections is necessary for a thorough
understanding of electoral processes and geographies in Alabama. Within this
process quantitatively defined place-based electoral cleavages have proven to be

helpful tools in accomplishing this task. We finish then with a reaffirmation of
the engagements already conducted with both context and quantitative methods
in electoral geography and forward the call for the use of context sensitive
quantitative methods.
References
Agnew, J. 1987. Place and Politics: The Geographical Mediation of State and
Society. Boston: Allen & Unwin.
Agnew, J. 1994. “The national versus the contextual – The controversy over
measuring electoral change in Italy using Goodman flow-of-vote estimates.”
Agnew, J. 1995. “The rhetoric of regionalism – The Northern League in Italian
politics, 1983–94.” Transactions of the Institute of British Geographers 20:
156–72.
Agnew, J. 1996. “Mapping politics: How context counts in electoral geography.”
Political Geography 15:129–46.
Agnew, J. 1997. “The dramaturgy of horizons: Geographical scale in the
‘reconstruction of Italy’ by the new Italian political parties, 1992–95.” Political
Agnew, J. and J. Duncan. 1989. The Power of Place: Bringing Together
Geographical and Sociological Imaginations. Boston: Unwin Hyman.
Anselin, L. 2005. “Exploring Spatial Data with GeoDa: A Workbook.” Urbana:
University of Illinois, Spatial Analysis Laboratory, Department of Geography.
Arant, B. 2004. “An overview of the proposed Constitution amendments affecting
Alabama taxes or economic development.” State and Local Tax Bulletin, 28
October.
Associated Press 2002. “Black turnout made election close, NAACP says,”
Birmingham News, 9 December: 4B.
Barnes, T. 2009. “‘Not only … but also’: Quantitative and critical geography.”
Professional Geographer 61: 292–300.
Beyerle, D. 1998a. “Election turnout called highest in history,” Tuscaloosa News,
19 November: 1B, 3B.
Beyerle, D. 1998b. “GOP Right says voters anti-fob, not pro-lottery,” Tuscaloosa
News, 2 December: 2B.
Brown, M., L. Knopp, and R. Morrill. 2005. “The culture wars and urban
electoral politics: Sexuality, race, and class in Tacoma, Washington.” Political
Chapman, T., J. Leib, and G. Webster. 2007. “Race, the creative class, and political
geographies of same sex marriage in Georgia.” Southeastern Geographer 47:
27–54.
Conservative Christians of Alabama 2004. “November 2nd vote on Alabama

Constitutional Amendments.” Available at: http://ccofal.org/alabama/2004/
amendment_recommendations.phtml [last accessed: March 31, 2009].
CNN.com 1998. “Alabama voters to decide on lottery,” 12 October. Available at:
http://www.cnn.com/ALLPOLITICS/stories/1999/10/12/bama.lottery/ [last
accessed: December 15, 2003].
CNN.com. 2008. “Local Exit Polls – Election Center 2008.” Available at: http://
www.cnn.com/ELECTION/2008/results/polls [last accessed: September 27,
2010].
Darmofal, D. 2006. “The political geography of macro-level turnout in American
political development,” Political Geography 25: 123–50.
DeMonia, R. 1999. “Lottery gets strong support.” Birmingham News 29 August:
1A, 8A.
Edge, T. 2010. “Southern strategy 2.0: Conservatives, white voters and the election
of Barack Obama,” Journal of Black Studies, 40(3): 426–444.
Florida, R. 2004. The Rise of the Creative Class: And How it’s Transforming Work,
Leisure, Community and Everyday Life. New York, NY: Basic Books.
Morrill, R., L. Knopp, and M. Brown. 2007. “Anomalies in red and blue:
Exceptionalism in American electoral geography.” Political Geography 26:
525–53.
National Conference of State Legislatures 2010. “2010 Ballot Measures Preview,”
22 September. Available at: http://www.ncsl.org/default.aspx?tabid=20114.
Rogers, W., R. Ward, L. Atkins, and W. Flynt 1994. Alabama: The History of a
Deep South State. Tuscaloosa, AL: University of Alabama Press.
Seabrook, N. 2009. “The Obama effect: Patterns of geographic clustering in the
2004 and 2008 presidential elections.” Forum-a Journal of Applied Research
in Contemporary Politics 7: Article 6
Shah, P. 2005. “The hypothetical impact of Alabama’s failed Amendment 2 on
public school funding,” Alabama Law Review 56: 863–72.
Shelley, F. and G. Webster 1998. “Population, settlement, race, and ethnicity in the
South.” Journal of Geography 97(4/5): 163–76.
Shin, M. 2001. “The politicization of place in Italy.” Political Geography 20:
331–52.
Shin, M. and J. Agnew. 2002. “The geography of party replacement in Italy, 1987–
1996.” Political Geography 21: 221–42.
Politics. Philadelphia, PA: Temple University Press.
Toal, G. and F. Shelley 2003. “Political geography,” in Gary L. Gaile and Cort
J. Willmott (eds), Geography in America at the Dawn of the 21st Century.
Oxford: Oxford University Press, 164–84.
Warf, B. and S. Arias. 2008. The Spatial Turn: Interdisciplinary Perspectives.
London; New York: Routledge.
Webster, G. 1993. “Congressional redistricting and African-American representation
in the 1990s – An example from Alabama.” Political Geography 12: 549–64.
Webster, G. 1996. “Partisan shifts in presidential and gubernatorial elections in

Alabama, 1932–1994.” Professional Geographer 48: 379–91.
Webster, G. and J. Bowman. 2008. “Quantitatively delineating the Black Belt
geographic region.” Southeastern Geographer 48: 3–18.
Webster, G., T. Chapman and J. Leib. 2010. “Sustaining the ‘societal and scriptural
fence’: Cultural, social, and political topographies of same-sex marriage in
Alabama.” Professional Geographer 62: 211–29.
Webster, G. and J. Leib. 2001. “Whose South is it anyway? Race and the
Confederate battle flag in South Carolina.” Political Geography 20: 271–99.
Webster, G. and J. Leib. 2002. “Political culture, religion, and the Confederate
battle flag debate in Alabama.” Journal of Cultural Geography 20: 1–26.
Webster, G. and N. Quinton. 2010. “The electoral geographies of two segregationist
(“Jim Crow”) referenda in Alabama.” Political Geography 29: 370–80.
Webster, G. and S. Samson. 1992. “On defining the Alabama Black Belt: Historical
changes and variations.” Southeastern Geographer 32: 163–72.
Webster, G. and R. Webster. 2004. “Taxing issues: Geography, politics and the
2003 Tax Reform Referendum in Alabama.” Southeastern Geographer 44:
190–215.
Chapter 12
Sexual Politics and the Sunshine State:
Electoral Geographies of Florida’s Marriage
Protection Amendment
Thomas Chapman
In 2008, Florida voters approved the Florida Marriage Protection Amendment,1

also known as Proposition 2, by a margin of 62.1 percent to 37.9 percent, making
Florida the 27th state to prohibit legal recognition of same-sex marriage by
amending its state constitution. This chapter explores the electoral geographies
of Proposition 2 by examining the spatial distribution of the vote through a lens
of the cultural politics of sexuality. I begin with a brief overview of the scalar
implications of the debate on same sex marriage in the United Sates, followed
by a discussion of Florida’s “place” in the national debate. I then examine the
socio-cultural implications of the debate that lead up to the actual vote, taking a
cue from the recent subgenre of work that invokes a critical and post-positivist
approach towards the cultural politics of sexuality and electoral outcomes (see, for
example Brown, Knopp, and Morrill, 2005; Rasmussen, 2006; Webster, Chapman,
and Leib, 2010). I conclude with a discussion of the spatial outcome of the vote,
revealing a multiplicity of cultural, social, and political geographies at the local
scale.
The Scalar Politics of Same-Sex Marriage Referenda
In 1957, Wyoming was the first state to insert a statutory prohibition on same-
sex marriage, followed by Maryland in 1973 and New Hampshire in 1983. But
the notion of extending marriage rights to same-sex couples in the United States
did not become a significant political issue until the mid-1990s, when the Hawaii
Supreme Court struck down that state’s constitutional ban on same sex marriage.2
By most measures, the Hawaii case was the springboard that set the stage for a
national debate on the issue, reacting to fears that the Hawaii case would spur
1 The official ballot language read as follows: “Marriage as the legal union of only
one man and one woman as husband and wife. No other legal union that is treated as
marriage or the substantial equivalent thereof shall be valid or recognized.”
2 Baehr v. Lewin 74 Haw. 530; 852 P.2d 44 (1993).
legal challenges elsewhere. Indeed, the Hawaii decision motivated few state
legislatures to pass their own Defense of Marriage statutes. At the federal level,
Congress passed, and President Clinton signed into law the Defense of Marriage
Act (DOMA)3 in 1996, limiting the definition of legal marriage to opposite sex
couples (for federal purposes), and allowing individual states to reject recognition
of same sex marriages performed in other states. But by all accounts, the 2003
landmark decision by the Massachusetts Supreme Court granting that state’s
same-sex couples the legal right to marry set the stage for a flurry of legislative
action to pass constitutional amendments banning same-sex marriage (Table
12.1). Several states, including Florida, had existing statutory laws banning
same-sex marriage on the books. But those most opposed to granting marriage
rights to same-sex couples argued that a constitutional amendment was necessary
in order to provide a stronger legal remedy that could withstand future judicial
scrutiny (Webster, Chapman and Leib, 2010). The national debate reached a fever
pitch in 2004, when President Bush proposed an amendment the US Constitution
that would limit legal marriage to opposite-sex couples. The notion of a federal
Table 12.1 Defense of Marriage Legislative and Court Actions*
2003 • Massachusetts Supreme Judicial Court rules state constitution guarantees

equal marriage rights for same sex couples.
2004 • President Bush announces support for federal constitutional amendment.
• Same sex couples begin marrying in Massachusetts.
• US Senate votes to block constitutional amendment to ban same sex
marriage.
• US House of Representatives votes against federal constitutional
amendment banning same sex marriage.
• Arkansas, Georgia, Kentucky, Louisiana, Missouri, Michigan, Mississippi,
Montana, North Dakota, Ohio, Oklahoma, Oregon and Utah voters approve
constitutional amendments banning same sex marriage in those states.
2005 • New York Supreme Court justice rules state law prohibiting same sex
marriage is unconstitutional.
• California Superior Court judge rules ban on same sex marriage is
unconstitutional.
• Kansas voters approve constitutional amendment banning same sex
marriage.
• Navajo Nation Tribal Council votes to ban same sex marriage; Navajo
President subsequently vetoes the measure to ban same sex marriage.
• California legislature becomes first legislative body in US to pass a bill
legalizing same-sex marriage without a court order. Governor Arnold
Schwarzenegger vetoes the legislation.
• Texas approves a constitutional amendment prohibiting same-sex marriage.
• Federal Court strikes down Nebraska’s same-sex marriage ban in the state
constitution.
3 Pub. L. 104-199, 100 Stat. 2419 (Sept. 21, 1996). See http://www.domawatch.org.
Sexual Politics and the Sunshine State 221
Table 12.1 Continued
2006 • Alabama passes a constitutional amendment banning same sex marriage.

• Supporters of a constitutional ban on same sex marriage in Florida fail
to collect enough signatures for the issue to appear on that year’s general
election ballot.
• The New Jersey Supreme Court orders the legislature to redefine marriage
to include same-sex couples or to establish a separate legal structure to
provide same-sex couples the same benefits as opposite sex couples.
• South Dakota, South Carolina, Virginia, Wisconsin, Tennessee, Colorado
and Idaho pass constitutional amendments defining marriage between a
man and a woman.
• Arizona becomes the only state in which a constitutional amendment to
define marriage between a man and woman fails.
2007 • Massachusetts legislators pass legislation preventing a proposed
constitutional amendment banning same-sex marriage from being placed on
the ballot.
• The Maryland Supreme Court overturns a lower court decision ruling that
same-sex couples do not have the right to marry.
2008 • The California Supreme Court rules that banning same-sex marriage is
unconstitutional.
• Florida, Arizona, and California voters pass constitutional amendments
defining marriage between a man and a woman.
2009 • The Iowa Supreme Court rules that banning same-sex marriage is
unconstitutional. The ruling takes effect immediately.
• The Vermont legislature passes a statute allowing same-sex marriage.
Vermont is the first state in which legislature, not the courts, authorized
same-sex marriage.
2010 • Washington DC City Council approves same sex marriage.
Note: * does not include legislative action on civil unions or court decisions addressing
local actions on same sex marriage.
Source: National Conference of State Legislatures. Available at: http://www.ncsl.org.
constitutional amendment, popular with many socially conservative politicians,

was also criticized as a transparent campaign strategy designed to energize the
Republican Party voter base in the months leading up to the 2004 presidential
election. The proposal failed to garner sufficient support in Congress that year.
But this unsuccessful attempt at the federal level inspired 11 states to place their
own defense of marriage amendments on the 2004 election ballot. After the dust
cleared on election night, all 11 referenda had passed by healthy margins. By 2010,
several other states had also adopted constitutional amendments banning same-sex
marriage, bringing the total to 31 states (Figure 12.1).
As Webster, Chapman and Leib (2010) point out, a cursory examination
of the spatial distribution of the voting results by state on such referenda point
towards a regional geography that indicates higher concentrations of voter support
for constitutional amendments in the American South than other regions in the
US, particularly the ‘deep south’ states (Figure 12.2). Part of the reason for this
Figure 12.1 The Statutory Landscape of Same Sex Marriage in 2010
Figure 12.2 States with Voter Approved Constitutional Language Prohibiting

Same Sex Marriage
is because southern states as a whole are disproportionately dominated by large

numbers of fundamentalist evangelical protestants, helping to partly explain the
lopsided support in favor of legislation banning same sex marriage. As a southern
state, Florida’s religious landscape is no exception. The state is also home to
some of the most influential and well financed Christian Right political action
lobbies in the nation, particularly in the central and northern parts of the state
(Chapman, 2007). So it comes as no surprise that the debate leading up to the
vote on Proposition 2 was steeped with various religious jeremiads encouraging
support for Proposition 2. But Florida’s cultural and political geographies are also
quite different from most other southern states in terms of its sheer population size,
rates of in-migration from other regions of the country, and its complex ethnic
and racial diversity. Indeed, the spatial outcome of the vote suggests multiple and
complex political, cultural, and economic interpretations of the implications in
instituting a constitutional ban on same-sex-marriage in Florida.
The Balkanization of the Marriage Rights Landscape in Florida
Florida is like Italy in the 15th century. What you have here is a cluster of city-
states, each with their own history and issues.
Elizabeth Birch, former Director of the Human Rights Campaign,
the nation’s largest LGBT4 rights organization
Long-time LGBT civil rights activists and those groups that actively oppose civil
rights for sexual minorities are well aware of the countless culture war battles
for sexual minority rights playing out in Florida for more than 30 years. These
parochial struggles, initiated and debated at the local scale, also had major
implications for the national LGBT rights movement (Stonewall Library and
Archives, 2007). Of course any discussion of LGBT rights in Florida must begin
with nationally known singer and orange juice spokeswoman Anita Bryant. Her (in)
famous culture war crusade opposing the City of Miami’s 1977 anti-discrimination
ordinance, amended to protect sexual minorities from discrimination, was a very
early extension of LGBT civil rights. Bryant’s successful effort to repeal the
ordinance though voter referendum less than a year later propelled Miami into the
national spotlight, placing the city as ground zero for the many culture war battles
on LGBT civil rights legislation that lie ahead, including the fight for marriage
rights (Stonewall Library and Archives, 2007). As a direct result of the Miami
debate, Florida’s Governor Reuben Askew was inspired to take immediate action.
Just one day after the ordinance’s repeal he signed a bill into law prohibiting same-
sex marriage in Florida (Boca Raton News, 1977). These early contributions to the
culture war debates on same-sex marriage paved the way for other states to use
Florida’s legal, social, and cultural framework in opposing LGBT rights. Ten years
4 Lesbian, gay, bisexual, transgender.

later in 1997, the Florida Legislature took up the marriage rights issue once again
by passing The Florida Defense of Marriage Act by an overwhelming margin.
The bill’s senate sponsor, John Grant, stated he thought it was ‘great that [the
Act] takes effect on June 4, right smack dab in the middle of Gay Pride Week’
(quoted in Kanotz, 1998). The legislation was specific in its intent to strengthen
the 1977 law, and was widely seen as a preemptive strike to fend off possible court
challenges at the national level.
The vast majority of debates and legislative action granting civil rights
for sexual minorities in the United States have taken place within local
jurisdictions (Bailey, 1999). Collectively, these localized actions had major
implications for public debate and lawmakers at the state and federal level.
But a more accurate picture of support for or opposition to marriage rights
for same-sex couples in Florida is revealed when place specific historical
and geographical contingencies are considered. In other words, the same-sex
marriage debate within Florida varied widely across space and time. On the
surface it would appear that the citizenry of the sunshine state (and the political
leaders that represent them) were mostly on the same page, firmly opposing
same-sex marriage. But Florida’s cultural and political geographies are much
more complex and diverse than a first glance would indicate. Amidst the high
profile public efforts by state and federal political leaders to legislate bans on
same-sex marriage, LGBT Floridians began taking action on their own, and in
their own communities.
As mentioned earlier, 2004 was the year when state and national debates on
marriage rights hit a fever pitch. President Bush’s efforts to federalize a ban on
same-sex marriage were met with many local protests throughout the country by
those opposed to legislative bans on same-sex marriage. The most high profile of
all was not in Florida but in San Francisco, when Mayor Gavin Newsome defied
California law by marrying hundreds of same-sex couples in the public corridors
of city hall. Watching the spectacle from afar, same-sex couples in Florida began
appearing at city and county courthouses all over the state, demanding marriage
licenses. Ten same-sex couples filed suit in Ft. Lauderdale in March of 2004,
seeking to overturn the 1997 Florida law and gain the right to legally marry.
The lawsuit was the first formal legal challenge to state laws barring same
sex marriage (Clary, 2004). That same week, seven same-sex couples showed
up at Orlando City Hall in Orange County, also demanding marriage licenses
(Santich, 2004). The deputy county clerk refused, handing the couples a copy
of the state statute. The couples were not deterred, and were married later that
day in a public religious ceremony sponsored by the First Unitarian Church of
Orlando. A month later, hundreds of same-sex couples in Gainesville rallied at
the Alachua County courthouse, also demanding the right to marry. And in a
show of solidarity with LGBT Floridians, the city of Key West unanimously
passed a resolution in support of same-sex marriage in Florida (Unitarian
Universalist Association, 2004).
Florida4Marriage versus Florida Red and Blue
As these public spectacles in Florida continued to capture media attention,

legislators in Tallahassee were again inspired to join in the fight by considering
a state-wide voter referendum that would ban same-sex marriage by amending
the state’s constitution. Interestingly, the President’s brother Jeb Bush, who
was the governor of Florida at the time, was a bit more ambivalent, saying he
would not rule out supporting such an amendment, but also added that it might
be unnecessary since existing state laws accomplished the same purpose (Nevins
and Davis, 2004). Without the active support of the governor, the proposal was
never acted upon in the legislature. But behind the scenes, powerful factions of
Florida’s religious right were gearing up for a signature petition drive in order to
get the initiative on the ballot in time for the 2006 election. A political action group
calling themselves Florida4Marriage5 was formed, lead by a cadre of activists
from nationally known religious right organizations with regional headquarters
in central Florida. Bankrolled by the state’s Republican Party (Smith, 2006), their
mission came up short when the group failed to collect enough petition signatures
in time to get the referendum on the 2006 ballot (Florida Division of Elections,
2006). Florida4Marriage was not deterred, eventually succeeding in getting the
required signatures to place the referendum on the 2008 ballot.
Lions, Tigers, and ‘Activist’ Judges from Massachusetts … Oh My!
Local culture war issues often stem from how belief systems shape geographical
imaginations, particularly those self-proclaimed ‘gatekeepers’ of community
values that sense a threat to local prevailing values (Mitchell, 2000; Castells,
2004). These threats can come from inside the community or they can come from
somewhere else. When the Florida4Marriage Chairman John Stemberger was
asked by a Tampa Bay television reporter why he was pursuing a constitutional
amendment to ban same-sex marriage when it’s already illegal in Florida, he
responded:
Well, for the same reason Massachusetts needed one. The state had valid
marriage laws, yet their judges just seemed to, with a stroke of the pen, erased
this fundamental human institution that has served us since the beginning of time.
So it’s important, as 28 other states have done, that we enshrined the definition
of marriage as a union of a man and a woman into our state constitution so a
5 Florida4marriage changed their name to strongfloridamarriages in 2009, focusing

their attention on the divorce rate in the state by pushing for a “Marriage Preparation
Act”, which would require a state mandated waiting period and counseling for any couple
wishing to get married in the state of Florida. For more information, see: http://www.
strongmarriagesflorida.org.
judge can’t look to the state constitution to find some reason to overturn our
valid marriage laws (WTVT Tampa Bay, 2007).
In this case it is ‘activist judges in Massachusetts’ who are most feared,

playing the role of outsiders that pose a serious threat to marriage in Florida.
Stemberger’s othering of Massachusetts judges also depends on cloaking the
issue with right vocabulary, in order to emphasize the gravity of the situation.
Marriage, according to Stemberger, is a ‘fundamental human institution’ that has
served us ‘from the beginning of time’. Therefore it is necessary to ‘enshrine’ the
very definition of marriage into a fundamental and timeless legal document that
symbolizes the bedrock of American civilization (the constitution). If marriage
ends up being defined by the tenets of civil law, or for that matter connected to
love, then the seriousness of the ‘threat’ is minimized, and it becomes, according
to Stemberger, easier to modify the rules for something that is just too sacred to
change. Even more incendiary is the possibility that a judge, especially one that
is not a Floridian, seems to be looking for a reason to use a ‘stroke of a pen’ as the
weapon of choice to ‘erase’ all that is holy and good in Florida. Once this point
is made (and it was made countless times during the lead up to the referendum
vote), the implication is that marriage in Florida is exceptional and serious, while
in Massachusetts it is a mockery and a sham. Indeed, the language written into
Amendment 2 specifically prevents state judges from overturning the law to
allow for same-sex marriage.
Supporters of Amendment 2 also saw an internal threat to marriage in Florida,
one that existed within the borders of the sunshine state. Places such as Miami
Beach, Broward County, and Gainesville had previously instituted domestic
partnership registries as a legal alternative to prohibitions on legal marriage,
viewing the definition of marriage as a matter of civil law that affords financial
benefits, protects private property, and guarantees taken-for-granted privileges
such as hospital visitation rights. The language that was written into Amendment
2 also addressed this ‘threat’, going further than most other state amendments in
the sense that its passage would render any form of legal domestic partnership
unlawful (including heterosexual partnerships). So the culture war battles over
Florida’s constitutional amendment redefining marriage would be fought on two
fronts: fending off threats external to the borders of the state, as well those threats
that lie within the state. And it was the voter referendum that would settle the
matter in Florida once and for all.
Opposition to the measure was spearheaded by the political advocacy group
FloridaRedandBlue (American Civil Liberties Union 2008). Through a well
financed and savvy media campaign called “No on 2”, FloridaRedandBlue’s
primary argument against passage of the marriage protection amendment focused
on the restriction of civil and economic rights. For example, the amendment’s
passage would run counter to Article I of Florida’s constitution (known as the
Declaration of Human Rights),6 which sets a precedent for establishing civil rights
rather than limiting them. In other words, it was argued that the amendment’s
passage would codify discrimination into the state’s constitution by erecting a
legal barrier that excludes an entire class of people from a state-sanctioned
social institution, solely because of their sexual orientation. FloridaRedandBlue
extended their argument beyond Florida’s borders by stating that the amendment’s
passage would ‘run counter to the proud US tradition of expanding, not denying,
human rights via the Constitution’ (Miami Herald Tribune 2008). This allowed
the group to ‘jump scale’, taking advantage of the extra-scalar linkages inherent in
local political debates by overcoming the constraints of their own geographically
anchored struggles (Marston 2000).
Because the language of the amendment included a ban on legalizing other
forms of domestic unions equivalent to marriage, the “No on 2” campaign
argued that unmarried heterosexual couples would be in danger of losing their
economic and legal benefits as domestic partners, such as hospital visitation
privileges, funeral arrangement rights and health insurance policies that
allow domestic partner benefits. The campaign emphasized that thousands of
heterosexual unmarried senior citizen couples residing in the state take advantage
of local domestic partnership benefits, and they would be especially hurts by the
amendment’s passage. In a series of widely debated television ads sponsored
by Fairness4AllFamilies,7 Floridians are introduced to senior citizens ‘Ed and
Clarice’, long term domestic partners of over 30 years:
Ed: We have a relationship based on love, trust, and mutual respect [Clarice
leans over and kisses Ed]. In our domestic partnership it’s very important that
we are able to take care of each other.
Clarice [her arm draped over Ed’s shoulder]: As we get older we want to be able
to take care of each other. We want to be able to visit each other in the hospital
if it comes to that, and this amendment would take it away, those rights away.
Ed [sitting close to Clarice on the sofa, holding her hand]: It’s so important to
vote against the amendment because it hurts real families, like our family. Please
vote no on the so-called marriage protection amendment (Fairness4AllFamilies,
2008).
Opponents of the amendment also argued that domestic partnership registries in

many Florida cities would also be threatened with expensive tax payer funded
legal action, forcing them to shut their doors and cancel domestic partner benefits
already in place. voicing her concern that the amendment’s passage would be a
6 Available at: http://www.leg.state.fl.us/Statutes/index.cfm?Mode=Constitution&Su

bmenu=3&Tab=statutes&CFID=176016536&CFTOKEN=57814895#A01.
7 Available at: http://www.myspace.com/fairness4allfamilies.
serious cultural setback for open-minded communities that currently have domestic
partnership registries, University of Florida’s women’s studies professor Florence
Babb stated that cities like Gainesville “would be losing a bit of its progressive
orientation and its humanity (Dudash, 2008).”
Spatial Analysis of the Electoral Outcome
The overall spatial pattern of the vote on Amendment 2 broadly aligns with the
socio-economic alignments found in most other work on the cultural politics of
sexuality (see, for example, Brown, Knopp and Morrill, 2005; Chapman, Leib and
Webster, 2007; Webster, Chapman, and Leib, 2010). The highest concentrations
of the vote in favor of passage were in the panhandle counties, by far the most
socially conservative and deeply religious portion of the state. Sixteen of the 17
counties where Amendment 2 passed by more than 80 percent of the vote were
in this part of the state (Figure 12.3). The next highest concentration of votes in
favor of Amendment 2 were generally found in the westernmost counties in the
panhandle (collectively known as “The Redneck Riviera”), and the rural interior
counties along Florida’s peninsular ‘spine’. With the exception of Gadsden
County (which is majority African-American), these areas are overwhelmingly
white and socially conservative, with large concentrations of evangelical
Protestants.
Since approval of voter referenda in the state of Florida requires at least a 60
percent majority vote, the amendment was defeated in all of the southern coastal
counties with large urban populations, particularly the South Florida megalopolis
of Palm Beach, Broward, Miami-Dade, by far the most racially and ethnically
diverse in the state.
The amendment also went down to defeat in the Tampa–St. Petersburg
metro area, but only by a thin margin. The socio-demographics of these areas
indicate a younger, more educated population than most other places in Florida.
Additionally, these areas also have very large numbers of same-sex couples,
particularly Ft. Lauderdale and Miami (US Census Bureau 2010). Surprisingly,
voters in Sarasota County in Southwest Florida also voted down the amendment, a
county that has been traditionally supportive of socially conservative Republicans
and their causes.8 Voters in Monroe County, which covers all of the Florida Keys,
had the highest concentration of voters voting ‘no’, 52 to 48 percent. This is not
surprising, since the City of Key West is one of the gay and lesbian “epicenters” in
North America for both residents and tourists alike. Key West is also well known
8 Sarasota is the home of former Congresswoman Katherine Harris. Most recall

Ms. Harris’ controversial role as Secretary of State and Florida Republican co-chair in the
immediate aftermath of the 2000 presidential election. Ms. Harris is nationally known for
aligning herself with the most far right leaders and their causes, including the infamous
Terry Shiavo controversy in 2005.
Figure 12.3 Distribution of Votes for Florida’s Amendment 2 Banning Same

Sex Marriage (2008)
for its long history of support for progressive political causes (Steinberg and
Chapman 2009). Of particular note are the results in Leon and Alachua Counties,
home to the state’s two largest public universities, Florida State University and the
University of Florida, respectively. Though both counties are located within those
regions within Florida that supported the amendment by healthy margins, voters
in these counties voted down the amendment. In this regard, the voting results
in counties are very similar to other large state university communities in other
states that had similar constitutional amendments on the ballot banning same-sex
marriage.
Conclusion
The cultural politics of sexuality is perhaps the number one culture war issue
in America today. Debates over granting civil, legal and social rights to sexual
minorities continue to hold the attention of the public, particularly the issue of
same-sex marriage. Along with Florida, both Arizona and California had similar
referenda on the 2008 ballot, and both referenda were passed by voters. The issue
in California is significant in that same-sex marriage rights were granted to couples
in early 2008 by the state’s highest court, and then taken away by voters at the polls
later that year. Various lawsuits challenging the constitutionality of California’s
same-sex marriage amendment have paved the way for the US Supreme Court
to hear the case sometime in the next year or so, promising yet another highly
salient national debate on granting marriage rights to same sex couples where the
outcome is uncertain.
Florida is not unlike other states that have codified marriage protection clauses
into state constitutions, in that support for or against same-sex marriage are
thoroughly grounded in place-based community identities, where culture, politics,
economics, morality, and values overlap with questions of citizenship and legal
rights. The electoral results of Amendment 2 reveal a balkanized landscape of
marriage rights within the state, revealing a mosaic of historical and geographical
contingencies that are all about the politics-of-place. Indeed, Elizabeth Birch’s
apt description of Florida as a cluster of city-states similar to the geographies of
fifthteenth century Italy continues to be an appropriate metaphor … Benvenuti lo
stato del sole (translation: welcome to the sunshine state).
References
American Civil Liberties Union. 2008. “Florida Red and Blue launches ‘SAY
NO 2’ campaign to fight Amendment 2, the Marriage Protection Amendment”.
Available at: http://www.aclufl.org/news_events/?action=viewRelease&ema
ilAlertID=3388 [last accessed: October 14, 2010].
Bailey, R. 1999. Gay Politics, Urban Politics: Identity and Economics in the
Urban Setting. New York: Columbia University Press.
Benjamin, J. and B. Nevins. 1998. Gay rights group seeks Florida’s help: Tour
building support for anti-hate law. Florida Sun-Sentinel, November 21, 1998.
1B.
Boca Raton News. 1977. Askew signs bill to ban gay marriage. June 9, 1977. 5A.
Brown, M., L. Knopp, and R. Morrill. 2005. The culture wars and urban electoral
politics: Sexuality, race, and class in Tacoma, Washington. Political Geography
24: 267–91.
Castells, M. 2004. The Power of Identity. 2nd edition. Malden, MA and Oxford:
Blackwell.
Christian Newswire. 2007. Available at: http://christiannewswire.com/news/
811905071.html [last accessed: September 22, 2010].
Chapman, T. 2007. Constructing the Moral Landscape through Anti-Discrimination
Law: Discourse, Debate, and Dialogue of Sexual Citizenship in Three Florida
Communities. Ph.D. Dissertation, Florida State University. December 2007.
Chapman, T., J. Leib and G. Webster. 2007. Race, the creative class, and political
geographies of same sex marriage in Georgia. Southeastern Geographer 47(1):
27–54.
Clary, M. 2004. 10 couples launch lawsuit over state ban on gay marriage. South
Florida Sun-Sentinel. July 2, 2004. 1A.
Dudash, A. 2008. Opponents say Amendment 2 unfair, takes away benefits. The
Independent Florida Alligator. University of Florida, October 31, 2008.
Available at: http://www.alligator.org/news/local/article_2e87b84a-a50f-
5661-a7c1-ebaeeba8769.html [last accessed: October 18, 2010].
Fairness4AllFamiles. 2008. Ed and Clarice Television Ad Campaign. Available at:
http://www.youtube.com/watch?v=XMuHNIqU3G4 [last accessed: October
13, 2010].
Florida Division of Elections. 2006. Available at: http://election.dos.state.fl.us/.
Kanotz, M. 1998. For better or for worse: A critical analysis of Florida’s Defense
of Marriage Act. Florida State University Law Review 25: 2 (Winter).
Marston, S. 2000. The social construction of scale. Progress in Human Geography
24(2): 219–42.
Mitchell, D. 2000. Cultural Geography: A Critical Introduction. Oxford:
Blackwell.
Scott, A. 2007. Amendment banning same-sex marriages closing in on ballot spot
in November 2008. Herald Tribune. November 5, 2007. Available at: http://
www.heraldtribune.com/article/20071105/NEWS/711050540# [last accessed
April 10, 2010].
Miami Herald Tribune. 2008. Amendment 2: No. Editorial, October 11, 2008.
A20.
Nevins, B. and J. Davis. 2004. Same-sex marriage ban gains in Florida. Florida
Sun-Sentinel. November 10, 2004.
Rasmussen, C. 2006. We’re no metrosexuals: Identity, place and sexuality in the
struggle over gay marriage. Social & Cultural Geography 7: 807–25.
Santich, K. 2004. Gays say ‘I do’ after clerks say they can’t: Seven same-sex
couples denied marriage certificates tied the knot anyway. Orlando Sentinel,
March 20, 2004.
Steinberg, P. and T. Chapman. 2009. Key West’s Conch Republic: Building
sovereignties of connection. Political Geography 28: 283–95.
Smith, A. 2006. GOP fuels gay marriage ban. St. Petersburg Times. January 12,
2006. Available at: http://www.sptimes.com/2006/01/12/State/GOP_fuels_
gay_marriag.shtml [last accessed: August 2, 2010].
Stonewall Library and Archives. 2007. Days without sunshine: Anita Bryant’s
anti-gay crusade. Available at: http://www.stonewall-library.org [last accessed:
April 4, 2010].
Unitarian Universalist Association. 2004. Available at: http://archive.uua.org/
news/2004/freedomtomarry/florida.html [last accessed: October 15, 2010].
US Census Bureau. 2010. Available at: http://www.census.gov/ [last accessed:
October 21, 2010].
Webster, G., T. Chapman, J. Leib. 2010. Sustaining the ‘societal and scriptural
fence’: Cultural, social, and political topographies of same-sex marriage in
Alabama. The Professional Geographer 62(2): 211–29.
WTVT Tampa Bay. Fox News 13 ‘Your Turn’. Aired December 12, 2007.
Index
228 incident 99–100, 103 Bryant, Anita 223

Bush, George H.W. 121, 184
Absentee ballot 161, 163, 169 Bush, George W. 133, 136, 138, 141, 151,
A Coruna province (Galicia, Spain) 84 157, 170, 178, 182, 184–5, 220,
Actor network theory 19 224
Adams, Gerry 48 Bush, Jeb 225
Africa 12, 33
African-American 6, 19, 133, 134, 138, California 123, 125, 146, 148, 152, 164,
146, 151, 175–6, 187–9, 191–3, 166, 181, 183, 186, 224, 230
199–205 Campaigning, Geography of 3, 36, 80
Agnew, John 10, 17, 19–20, 113, 197 Canada 16, 35
Alabama 6, 195–215 Capitalist regulation 77, 79–80, 93
Alabama ‘Black Belt’ 200, 203, 205–6, Carter, Jimmy 140, 182, 200
210–12, 214–15 Catalonia 59
Alabama Christian Coalition 202–5 Chapman, Thomas 20, 221
Alabama ‘Wiregrass’ 206, 210–11, 214–15 Charlottesville, Virginia 192
Alaska 163 Chen Shui–bian 97, 102–3, 107–10, 113
American Samoa 163 Cheney, Dick 184
Anti-Masonic Party (US) 177–8 Chiang Ching-kuo 97, 101–3, 106
Apportionment 44 Chiang Kai-shek 98–9, 101
Arizona 163, 166, 169, 230 Chicago 128, 149, 176, 178
Arkansas 163 China 98–100, 104, 108, 110, 112
Askew, Reuben 223 Civil rights 138–40, 176, 223–4, 227
Australia 41–2 Class 4–6, 77, 84, 88, 106, 127, 134–8,
141–3, 151, 177, 196
Baltimore, Maryland 178 Clinton, Bill 140, 182–4, 200, 220
Baxley, Lucy 199 Clinton, Hillary 138, 179–87, 189–90,
Bayesian statistics 18 192–4
Beijing 110 Cold War 177
Beiras, Xose Manuel 75 Colorado 146, 163
Bergamo, Lombardy, Italy 65, 69–70 Common Agricultural Policy, EU 89
Berlusconi, Silvio 65 Common Fisheries Policy, EU 88–9
Biden, Joe 185–6 Confederate battle flag 197
Birmingham, Alabama 200, 211–12 Connecticut 186
Bloque Nacionalista Galego (Galicia, Spain) Conservative Party (UK) 34–8, 42–3, 45,
5, 75–6, 80–93 49, 51, 53
Brescia, Lombardy, Italy 65, 68–9 Conservative-Unionists Party (UK) 34
British Nationalist Party 35, 37, 45, 50 Convenience voting 161–4, 169
Brown, Michael 20 Creative class 146, 211–2, 214
Cremona, Lombardy, Italy 65, 68 Proportional representation 31–3, 38,

Critical geography 7, 16, 19, 76–7, 195, 219 42, 47–51, 54
Cultural political economy 5, 75–80, 82, Single Transferrable Vote 32, 42, 46–8
90–93 Supplementary vote 5, 38–40, 42
Cultural studies 5, 76 Electronic voting machines 157, 159, 168–9
‘Culture Wars’ 6, 140, 197, 210, 212, Empiricism 3, 9, 20
214–5, 223, 225–6, 230 Employment 65–70, 72, 74, 86–7, 89,
211–2, 214
Dail Eireann (Irish Parliament) 46–7 England 31–2, 38, 44–5, 53
Dangwai (Taiwan) 101–3, 106–7 Ethnicity 100, 104, 127, 134, 137–140, 142,
Defense of Marriage Act (DOMA) 220 145, 151–2, 223
Deindustrialization 135, 137 Europe 3, 5, 12, 31, 45, 65, 75–6, 80–5,
Democratic Party (US) 6, 134–8, 140–41, 87–93,139, 176
145, 149, 151, 170, 175–90, 192–4, European Parliament 31–2, 44–46
197, 199–200, 212, 214 European Union 81–2, 86–93
Democratic People’s Party (Taiwan) 97,
102–113 Fairness4AllFamilies (Florida) 227
Democratic Unionist Party (Northern Falwell, Jerry 140
Ireland) 48 Fiji 18
D’Hondt/Jefferson electoral formula 45 Florida 7, 15, 123, 125, 128, 148–9, 152,
Developments in Electoral Geography 157, 163, 167, 183, 186, 219–230
9–10, 19–20 Florida Marriage Protection Amendment,
Diaspora 117–18, 122–9 2008 (Proposition 2) 219–30
Dodd, Christopher 185–6 Florida4Marriage 225
Dukakis, Michael 183 FloridaRedandBlue 226–7
Florida, Richard 146
Early voting 157, 163–4, 170 Fort Lauderdale, Florida 224, 228
Ecological fallacy 17, 187 Forza Italia 65
Education 5, 13–14, 16–17, 106, 136, 140, Frank, Thomas 136–7
160, 189, 192, 197, 199, 202–5 Friends and neighbors effects 17, 48
Edwards, John 180, 185–6
Eisenhower, Dwight 184 Gainesville, Florida 224, 226, 228
Elections 3–6, 10, 12–18, 31–52, 59, 69, Galicia 5, 75–6, 80–93
75–93, 97–113, 121, 157–216 Galician Coalition 81
Electoral College, US 13, 15, 17, 133, 138, Gender 3–5, 11, 14, 18–19, 76–7, 88, 127,
148 136, 140–41, 151
Electoral Commission, UK 44 Geographic information systems 4–5, 16,
Electoral geographers 3, 5, 9–10, 14, 60, 68, 71, 206
19–20, 98, 194–5 Geographically weighted regresion (GWR)
Electoral geography 3–5, 7, 9–21, 31, 34, 5–6, 64–74, 134, 142–3, 145–6,
75–6, 80, 98, 158–9, 187, 195–7, 148–9, 151–2
216 Georgia 163, 167, 169, 176, 182, 186
Electoral systems 5, 14–15, 31–54, 195 Germany 18, 49, 79
Alternative vote 5, 31, 41–3, 47, 53–4 Gerrymandering 13, 16, 31, 35–6, 159
First-past-the-post (FPTP) 5, 31, 33–54 Giuliani, Rudolph 184, 186
List systems 31, 43–4, 47, 49, 51, 53–4 Globalization 65, 135
Preferential voting 38 Goldwater, Barry 177
Index 235
Gore, Al 157, 184, 200 Kennedy, Robert 178

Great Society 137 Kentucky 187, 193
Green Party (UK) 35, 37, 45, 50 Kerry, John 200, 202
Gudgin, Graham 31, 34, 36 Key West, Florida 224, 228
Gutierrez, Luis 126 Kuomintang (Taiwan) 5–6, 97–113
Knopp, Larry 20
Harkin, Tom 182
Hawaii 123, 163, 219–20 Labour Party (UK) 34–8, 42–3, 48, 50–52,
Help America Vote Act 160, 167 54
House of Commons, UK 31, 33–4, 54 Latinos 138, 141–2, 146, 151–2
House of Representatives, Australia 41 Lee Teng-hui 97, 100, 103, 106–9
House of Representatives, New Zealand 49, Lee Yuan-tseh 109
51 Lega Nord (Italy) 5, 59–74
House of Representatives, US 44, 122, Legislative Yuan (Taiwan) 101–2, 107
176–7, 179, 185 Leib, Jonathan 18, 20, 221
Hsieh Chang-ting (Frank) 108, 110 Liberal Party (UK) 34–5
Hsu Wen-lung 109 Liberal Democrat Party (UK) 35–8, 42–3,
Huckabee, Mike 179, 182, 184, 186 45–6, 50–51, 53–4
Humphrey, Hubert 178 Libertarian Party (US) 185
Lien Chan 107–10
Idaho 145–6, 169 Lin Yang–gang 107
Identity 6, 18–20, 65, 76–7, 79–81, 87, Livingstone, Ken 39–40
89–90, 93, 97–113, 117–129, 135, Local area statistics (LISA) 18, 196, 205–6,
137, 152, 196 209–10, 214
Illinois 124, 146, 163, 183, 185–6 Lombardy 64, 67–8, 70, 73–4
Immigration 65–67, 69, 101, 167, 171, London 31–2, 38–40, 44, 49–51
176–7, 196 London Assembly 49–50
India 16–17 Louisiana 163, 186
Indiana 134, 167, 169, 187
Iowa 182–3, 186 Maastricht Treaty 82, 87
Iraq 185–6 McCain, John 6, 133–5, 137–8, 141–2,
Ireland 48 149, 151, 175, 179, 181–2, 184,
ISI Journal Citation Reports 9, 13 186–9, 194, 200, 10, 214–15
Israel 12, 15 McCarthy, Eugene 178
Italy 4–5, 59–74 Madison, Wisconsin 192–3
Mantova, Lombardy, Italy 69–70
James, Fob 197, 199 Malapportionment 16, 35–6
Japan 99 Maori party (New Zealand) 49
Johnson, Boris 39 Maryland 186, 190, 219
Johnson, Lyndon 178, 192 Massachusetts 186, 205, 220, 225–6
Johnston, Bennett 121 Ma Ying-jeou 97, 108, 110
Johnston, R.J. 9–10, 15–16, 31–57 Mexico 12, 18
Miami, Florida 223, 226, 228
Kang Ning-hsiang 101–2 Michigan 183, 186
Kaohsiung (Meilidao) incident 102–3, 113 Milano, Lombardy, Italy 65, 68–70, 72
Kao Yu-shu (Henry Kao) 101 Milwaukee, Wisconsin 192–3
Kennedy, John 184, 200 Mississippi 200
Missouri 179, 186–90, 193 Pan-Blue coalition (Taiwan) 104

Mixed-methods approaches 4, 195 Pan-Green coalition (Taiwan) 104
Montana 187 Paper ballots 159
Moral Majority 140 Partito Deomcratico (Italy) 60–62
Moran’s I 60–61, 63 Paul, Ron 184–5
Morrill, Richard 20 Pavia, Lombardy, Italy 65
Peng Ming-min 109
National Assembly (Taiwan) 97, 105, 109 Pennsylvania 124, 187, 193
National Assembly (Wales) 50–51 People First Party (Taiwan) 104, 107
National party conventions, US 177–9 People’s Party (Galicia, Spain) 75, 84, 88,
Nationalism 14, 75, 79, 81, 113, 127 92
Nationalist Party (Northern Ireland) 35 Philadelphia 128, 193
Nebraska 18 Pittsburgh 193
Neocoservative 136, 177 Place 14, 18–20, 59–60, 75, 77–80, 82,
Neoliberalism 135–7 84–5, 93, 113, 127, 196–7, 214–15,
Nevada 146, 152, 163, 183 219, 224, 230
New Deal, The 137, 176 Plaid Cymru (Wales) 35–6, 51, 53
New Hampshire 178, 181–3, 186, 219 Political economy 3–5, 17–19, 75–93
New Jersey 124, 186 Political geographers 4, 9–11, 16, 78, 93,
New Mexico 148–9, 163 171, 196
New Party (Taiwan) 104 Political geography 9–10, 17–18, 20–21,
New Progressive Party (Puerto Rico) 62, 76–7, 79–82, 90, 93
118–19 Political Geography (Journal) 11–13
New York 124–5, 148–9, 166, 169, 181, Political Geography (Textbook) 10
183, 186 Political science 4, 10–11, 13, 31, 72, 140
New York City 123, 128, 178, 184 Political scientists 12–3, 16, 151, 157,
New Zealand 5, 33–34, 49, 51 159, 203
Newsome, Gavin 224 Pontevedra province (Galicia, Spain) 84
Nixon, Richard 178, 184 Popolo della Liberta (Italy) 59–62, 65
Norfolk/Virginia Beach, Virginia 190 Popular Democratic Party (Puerto Rico)
North Carolina 134, 149, 163, 187 118–20
North Dakota 163 Post-modernism 4
Northern Ireland 32, 35, 44–5, 48, 53 Post-positivism 7, 219
Northern Ireland Assembly 48 Post-Soviet/Eurasian Geography 11
‘Nuyorican’ 127–8 Post-structuralism 4–5, 16–21, 76–7, 93, 195
Presidential elections, Taiwan 97–113
Obama, Barack 6, 133–52, 175, 179–94, Presidential elections, US
200, 202, 210, 214 2000 4, 10, 15, 157, 167, 182, 184–5,
Ohio 124, 166, 187 197, 200, 215, 228
Oklahoma 179–80, 193 2004 15, 160, 185, 189, 192, 197, 200,
Optical scan ballots 167–8 202, 210, 215, 221
Ordinary Least Squares (OLS) regression 2008 6, 133–52, 166, 168, 175–94,
63–4, 67–8, 70–73 197, 200–202, 209–10, 214–15
Oregon 164, 178 Presidential Task Force on Puerto Rico’s
Orlando, Florida 224 Status 122
‘Orlandorican’ 127 Primary elections 5–6, 47, 175–94
Ourense province (Galicia, Spain) 84–5 Provisional voting 6, 166–7, 170
Index 237
Puerto Rico 6, 117–129 Shelley, Fred 9

Puerto Rican Independence Party 118–19 Shin, Michael 20, 196
Siegelman, Don 197, 199, 202
Quinton, Nicholas 20 Sinn Fein (Northern Ireland) 48
Social Democrat Party (UK) 35
Race 6, 10, 14, 16, 18–19, 76–7, 127, 137–8, Social theory 3, 9
141–3, 152, 171, 188, 197, 202–3, Socialist Party of Galicia (Galicia, Spain)
210, 212–15, 223 75, 81
Reagan, Ronald 136–7, 140, 177, 184–5 Societal regulation 77
Redistricting 3, 5, 10, 13–17, 19, 196 Sondrio, Lombardy, Italy 65, 69, 72
Referenda 5–7, 14, 18, 33, 38, 43, 53, Soong, James 107–10
109–10, 117–29, 164, 195–7, South Africa 33
202–6, 210, 212, 214–15, 219, 221, South Carolina 18, 183, 186, 202
223, 225–6, 228, 230 South Dakota 187
Religion 6, 127, 134, 139–43, 145, South Korea 98, 112
148–52, 176, 185, 197, 202–3, 214, Southeastern Geographer 9, 11
223, 225, 228 ‘Southern Strategy’ 137–8
Republican Party (US) 6, 121, 133–8, Soviet Union 99, 177
140–41, 149, 167, 170, 175, 177–82, Spain 4–5, 75–6, 80–82, 84–5, 87, 90–93
184–9, 193, 197, 199–200, 203, 206, Spanish Workers’ Socialist Party 75, 81, 92
212, 214–15, 221, 225, 228 Spatial analysis 5, 17, 20, 59–60, 62–3, 71,
Republic of China See Taiwan 74, 228
Rhode Island 187 Spatial autocorrelation 6, 59–62, 71, 73,
Richardson, Bill 185–6 196, 205
Richmond, Virginia 190 Spatial econometrics 63, 70–74
Riley, Bob 199, 203 Spatial heterogeneity 59, 63, 68, 73
Robertson, Pat 140 Spatial lag 61, 63, 67–8, 70–71, 73
Roll-call voting 14 Spatial regression 59–60, 62–3, 70–71, 73
Romney, Mitt 179, 184–6 Spatial structure 63, 68
Roosevelt, Franklin 177 Spatial turn 195
Russia 12–13 Statehood Republican Party (Puerto Rico)
119
Sainte-Lague/Webster electoral formula 44 Stemberger, John 225–6
Same-sex marriage 7, 18, 177, 197, 205, Stoke-on-Trent, UK 39
212, 214, 219–30 Sun Yat-sen 99
Sánchez-Ayala, Luis 5 ‘Super Tuesday’ Primaries, US 183, 186–90
San Francisco 224 Supreme Court, US 10, 167, 169, 197,
San Juan, Puerto Rico 117 199, 202, 205, 219–221, 230
Sauer, Carl 16
Scotland 17, 32, 35–6, 37, 44, 48, 50–51, Tactical Voting 37–8, 41–2, 51, 53–4
53, 59 Taipei 99, 101–3, 107
Scottish Nationalist Party 35–6 Taiwan 4–5, 97–113
Scottish Parliament 50 Taiwan Independence Party 104, 109
Senate, Australia 46 Taiwan Solidarity Union party 104
Senate, US 121–2, 185–6 Tallahassee, Florida 225
Serrano, Jose 126 Tampa-St. Petersburg, Florida 228
Sexuality 7, 14, 18–19, 76, 127, 136, 151, Taylor, Peter 9, 31, 34, 36
177, 205, 219–30 Tennessee 163
Texas 123, 125, 148–9, 163, 181, 187 Vote studies 5, 14–15
‘Tea Party’ movement (US) 59, 136 Voter identification 6, 157, 159, 167,
Thompson, Fred 184–5 169–71
Turkey 45 Voter qualifications 125–29
Turnout 3, 14, 36, 120, 141, 162–164, 170, Voter registration 5–6, 157, 161, 164–6,
199 170
Voting behavior 14, 59–60, 73
Unionist Party (Northern Ireland) 35 Voting machines 167–9
United Kingdom 4–5, 12, 15–16, 31–54 Voting Rights Act, US 157, 172
United Kingdom Independence Party 35, Voting technology 6, 15, 157, 161, 167–70
37, 45
United States 3–6, 10, 12, 15–16, 19, 31, Waisheng 99
35, 44, 59, 85, 108, 117–9, 121–9, Washington 163–4, 186
133, 135–6, 138–41, 143, 145, 152, Washington, DC 117, 146, 190, 192
157–9, 161, 169, 171, 175–6, 181, Wales 31–2, 35–7, 44, 50–54
182, 187–8, 192, 196, 203, 205, Webster, Gerald 6, 18, 20, 108–9, 221
219, 224 West Virginia 163, 187, 193
US Election Assistance Commission 158, What’s the Matter with Kansas? 136
160 Whig party (US) 178
Utah 145, 149, 163 Wisconsin 146, 148, 186, 190, 192–4
Wyoming 219
Vermont 187
Virginia 134, 146, 149, 186, 190–3 Young, Don 121
Vote-by-mail 164, 170

Cogito

Uploaded by

Copyright:

Available Formats

Cogito

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cogito

Uploaded by

Copyright:

Available Formats

Revitalizing Electoral Geography

British Library Cataloguing in Publication Data

Library of Congress Cataloging-in-Publication Data

ISBN 9781409410720 (hbk)

II and bound in Great Britain by

List of Figures  vii

Part I Conceptual Foundations

2 On the Shores of the “Moribund Backwater”?:

Part II Electoral Geography Outside of the US

3 Electoral Systems, Geography, and Political Behaviour:

4 Spatial Regression for Electoral Studies:

5 Elections and Cultural Political Economy:

6 Place, Identity, and Taiwan’s Presidential Elections, 1996–2008  97

7 The Power of Definition:

Part III Electoral Geography in the US

8 Class, Ethnicity, Religion and Place in the 2008

9 The Emerging Geography of How Americans Cast their Ballots  157

10 The 2008 US Presidential Primary Elections in Geographic

11 Electoral Alignments and Place-Based Cleavages in Statewide

12 Sexual Politics and the Sunshine State:

2.1 Number of Journal Articles on Electoral Geography Published,

5.1 BNG Election Support Map, 1993 (%)  83

6.1 Map of the Regions of Taiwan  111

7.1 Status Referendums in Puerto Rico  120

8.1 Percent of Votes for Obama by County, 2008  133

8.4 Results of Bivariate GWR of Support for Obama and Average

9.1 Percentage of Votes Cast In-person  162

11.1 Partisan Elections  198

12.1 The Statutory Landscape of Same Sex Marriage in 2010  222

4.1 Descriptive Statistics for Variables in Lombardy Data Set (N = 1,546)  67

8.1 Votes in 2010 Presidential Election by Household Income  135

9.1 Top Provisional Ballot States, 2008  166

12.1 Defense of Marriage Legislative and Court Actions  220

John Agnew, Department of Geography, University of California at Los Angeles,

Thomas Chapman, Department of Political Science and Geography, Old

Heather Hollen, Department of Geography, University of Oklahoma, USA.

Ron Johnston, School of Geographical Sciences, University of Bristol, UK.

Jonathan Leib, Department of Political Science and Geography, Old Dominion

Daniel McGowin, Department of Humanities, Alabama State University, USA.

Toby Moore, Research Triangle Institute International, USA.

Erinn P. Nicley, Department of Geography, University of Illinois at Champaign-

Charles Pattie, Department of Geography, University of Sheffield, UK.

Nicholas Quinton, Department of Geography, Florida State University, USA.

Luis D. Sánchez-Ayala, Department of History, Universidad de los Andes,

Fred M. Shelley, Department of Geography, University of Oklahoma, USA.

Michael Shin, Department of Geography, University of California at Los Angeles,

Barney Warf, Department of Geography, University of Kansas, USA.

Gerald R. Webster, Department of Geography, University of Wyoming, USA.

More recently, however, a renaissance of sorts has begun to emerge in electoral

different types of electoral geography topics, including general and primary

Finally, Thomas Chapman analyzes the geographical contours of Florida’s

Adams, P. 2007. Atlantic Reverberations: French Representations of an American

Johnston, R., F. Shelley and P. Taylor (eds) 1990. Developments in Electoral

we recognize that it is nearly impossible to track down every non-ISI journal

Has Electoral Geography Declined as an Important Subfield?

The first issue we address is whether electoral geography has declined as an

Figure 2.1 Number of Journal Articles on Electoral Geography Published,

In terms of journals, we found at least one electoral geography article in 28

Figure 2.2 Number of Articles on Electoral Geography Published by

Figure 2.3 Country of Study in Journal Articles on Electoral Geography,

List of Figures vii

6 Place, Identity, and Taiwan’s Presidential Elections, 1996–2008 97

9 The Emerging Geography of How Americans Cast their Ballots 157

5.1 BNG Election Support Map, 1993 (%) 83

6.1 Map of the Regions of Taiwan 111

7.1 Status Referendums in Puerto Rico 120

8.1 Percent of Votes for Obama by County, 2008 133

9.1 Percentage of Votes Cast In-person 162

11.1 Partisan Elections 198

12.1 The Statutory Landscape of Same Sex Marriage in 2010 222

4.1 Descriptive Statistics for Variables in Lombardy Data Set (N = 1,546) 67

8.1 Votes in 2010 Presidential Election by Household Income 135

9.1 Top Provisional Ballot States, 2008 166

12.1 Defense of Marriage Legislative and Court Actions 220