Garret 2009

Journal of Communication ISSN 0021-9916
ORI GI NAL ARTI CLE

Politically Motivated Reinforcement Seeking:
Reframing the Selective Exposure Debate
R. Kelly Garrett
School of Communication, The Ohio State University, Columbus, OH 43221, USA
This article seeks to reframe the selective exposure debate by demonstrating that people
exhibit a preference for opinion-reinforcing political information without systematically
avoiding opinion challenges. The results are based on data collected in a national random-
digit-dial telephone survey (n = 1,510) conducted prior to the 2004 U.S. presidential
election. Analyses show that Americans use the control afforded by online information
sources to increase their exposure to opinions consistent with their own views without
sacricing contact withother opinions. This observationcontradicts the commonassumption
that reinforcement seeking and challenge avoidance are intrinsically linked aspects of the
selective exposure phenomenon. This distinction is important because the consequences of
challenge avoidance are signicantly more harmful to democratic deliberation than those
of reinforcement seeking.
doi:10.1111/j.1460-2466.2009.01452.x
U.S. Americans preferences regarding which opinions they encounter fundamentally
shape their exposure to political information. This exposure process has been the
subject of inquiry and debate since the 1940s (Lazarsfeld, Berelson, & Gaudet,
1944), and although many ndings point toward some form of selectivity, the
literature remains plagued by discrepant claims. Resolving these contradictions
is important because decisions about what news sources people use and which
articles they read are profoundly important to the future of deliberative democracy.
The contemporary information environment, by virtue of a rapidly expanding
and increasingly specialized news market and sophisticated new computer and
network technologies, allows individuals unprecedented ability to selectively acquire
political information (Bimber & Davis, 2003; Neuman, 1991). If individuals seek
to expand their familiarity with information that supports their beliefs and to limit
their exposure to other perspectives, the prospects for deliberation are dire. In
Republic.com, Sunstein (2001) predicts just such a future. He foresees a fragmented
citizenry, where individuals use the Internet to construct echo chambers in which
Corresponding author: R. Kelly Garrett; e-mail: garrett.258@osu.edu
676 Journal of Communication 59 (2009) 676699 2009 International Communication Association
R. K. Garrett Reframing the Selective Exposure Debate
the only viewpoints they encounter are their own. According to Sunstein, this
behavior will not be limited to small, deviant groups, but will come to dominate the
political landscape as more and more Americans go online for news.
Although reinforcement seeking and challenge avoidance both have important
consequences for attitude formation and opinion strength, avoiding opinion-
challenging information is particularly harmful. Exposure to political difference
is a dening element of effective deliberation and has a signicant inuence on
individuals ability to accept disagreement and seek political solutions. Exposure to
contrasting perspectives also increases familiarity with the rationales that motivate
opposing views, which can in turn foster political tolerance (Mutz, 2002; Price, Cap-
pella, &Nir, 2002). Contact with other political viewpoints stimulates more thorough
information searches and more careful scrutiny of alternatives (Delli Carpini, Cook,
& Jacobs, 2004; Mendelberg, 2002; Nemeth, 1986; Nemeth & Rogers, 1996). Con-
versely, if individuals effectively avoid opinion-challenging information, the society
to which they belong is likely to become more politically fragmented (Sunstein,
2001, 2002). In light of the important stakes for political deliberation and democratic
society, it is critical that we understand what choices people will make in the changing
political news environment.
The next section of this paper provides a more detailed overview of research
on the topic of politically motivated selective exposure. This review suggests that
an underlying pattern, namely, that people seek opinion-reinforcing information
without avoiding opinion-challenging information, can help to explain inconsisten-
cies between earlier ndings. A discussion of psychological mechanisms that help
to explain this behavior provides theoretical grounding for this claim. Results are
presented after describing the survey methodology and measurement techniques.
The data show that individuals who have additional control over their information
environment do experience an increase in exposure to opinion-reinforcing informa-
tion, but that their exposure to opinion-challenging information does not drop. The
paper concludes with a discussion of these results, and recommendations for future
research.
Selective exposure theory
Politically motivated selective exposure, the tendency to craft an information envi-
ronment that reects ones political beliefs, has been a topic of debate for several
decades (for reviews, see Frey, 1986; Sears & Freedman, 1967). Research on the
subject was largely dormant from the early 1970s until the late 1990s, but concern
over the consequences of the evolving information environment has breathed new
life into the topic (e.g., Mutz & Martin, 2001; Neuman, 1996, p. 15; Stroud, 2008;
Sunstein, 2001). In the beginning of the 21st century, selective exposure, including
politically motivated selectivity, has emerged as one of the most commonly used
theories in communication scholarship (Bryant & Miron, 2004).
Journal of Communication 59 (2009) 676699 2009 International Communication Association 677
Reframing the Selective Exposure Debate R. K. Garrett
According to selective exposure theory, individuals prefer exposure to arguments
supporting their position over those supporting other positions.
1
As a consequence,
individuals are more likely to read, listen to, or view a piece of information the
more it supports their opinion, and less likely to attend to it the more it challenges
their position (Fischer, Jonas, Frey, & Schulz-Hardt, 2005; Knobloch-Westerwick &
Meng, 2009; Lazarsfeld et al., 1944; Mutz, 2006; Sweeney & Gruber, 1984). Scholars
argue that this preference also leads individuals to prefer information sources that are
more supportive of their opinions over less supportive alternatives (e.g., Lowin, 1967;
Mutz & Martin, 2001). For example, Mutz and Martin (2001) assert that control
over opinion exposure is inversely correlated with exposure to opinion-challenging
information. Comparing peoples experiences with television, newspapers, news
magazines, and talk shows, the researchers found that individuals who use sources
offering more partisan information tend to encounter less opinion-challenging
information. For example, people are less likely to come across opinions that
differ from their own when listening to talk radio than when reading a newspaper
or watching television. They also found that individuals who can choose among
competing sources of partisan local news tend to have less contact with opinion-
challenging information than those living in areas served by a single (less partisan)
local news source. More recently, Mutz (2006) has argued that individuals social
networks are also unlikely to be a source of exposure to cross-cutting political
attitudes. This is because individuals are selective about who they associate with and
because they tend to shape their environments in ways that limit their exposure
to other viewpoints (e.g., through their choice of neighborhood, their associational
afliations, etc.).
Critics of the theory of politically motivated selective exposure question the
existence of anunderlyingpsychological tendencytoseeksupport andavoidchallenge.
According to these scholars, the data do not support the claim that citizens are
disproportionately exposed to opinion-supporting information. In their review,
Sears and Freedman (1967) contest the interpretation of the data reported in
Lazarsfeld et al. (1944) seminal work on selective exposure, and observe that many
early studies undermine selective exposure claims. Furthermore, they note that
choices that do yield exposure to predominantly opinion-reinforcing information
are not necessarily motivated by opinion-based selectivity per se, but may instead be
secondary consequences of decisions unrelated to ideology. They refer to this as de
facto selective exposure. For example, many nancial analysts may choose to read the
Wall Street Journal because of its coverage of nancial news. The analysts tendency
to agree with the papers political views is not motivated by an effort to nd support
or avoid challenge; instead, it is a reection of their political similarity with those
who write about nancial news.
In a second major reviewof literature on selective exposure, Frey (1986) sought to
re-establish the legitimacy of the phenomenon. He argued that the lack of support for
selectivity prior to 1967 could be explained by Festingers (1964) later articulations of
dissonance theory. In his revised version, Festinger specied that selective exposure
only occurs if an individuals position is a product of choice and if the individual
is personally committed to this position. Festinger also identied several conditions
under which opinion-challenging information would be desirable, such as when the
information might be useful for future decisions. Frey argued that earlier experiments
failedtoaccount for these important considerations. Reviewing researchthat attended
to these factors, he concluded that when individuals are selecting information about
decisions reached on their own accord and to which they are committed, they exhibit
a consistently strong preference for viewpoint-reinforcement.
Frey also argued that individuals engage in selective challenge avoidance, but
noted that the effect is much weaker than that of selective reinforcement seeking.
This conclusion is echoed in recent research looking at the inuence that new
information and communication technologies are having on peoples exposure to
political information. A study examining peoples use of an interactive CD-ROM
containing campaigninformationabout the twomajor candidates inthe 2000 election
found only limited evidence that subjects political attitudes inuenced their use of
political information (Iyengar, Hahn, Krosnick, & Walker, 2008). Similarly, analyses
of 2000 General Social Survey data provide little evidence that people are using the
Internet to avoid political difference (DiMaggio & Sato, 2003). Respondents reported
using sites that are neutral or that challenge their opinion as often as they use those
that reinforce it. Furthermore, individuals frequently report that their use of these
sites leads them to revise or rene their opinions. Another survey conducted in 2000
examining knowledge of the presidential campaign suggests that the Internet could
ultimately reduce the effects of partisan selectivity. In her study, Stroud (2004) found
that online political information seeking was associated with comparable increases
in knowledge about both presidential candidates, especially among individuals with
the lowest levels of political knowledge.
The debate over selective exposure continues in the face of this contradictory
evidence, with scholars on both sides of the debate treating preference for support
and aversion to challenge as linked aspects of a single psychological preference.
I argue that by conceiving of these preferences as separate phenomena we can
reconcile these otherwise incompatible ndings regarding citizens political infor-
mation acquisition practices. Building on the work of Frey (1986) and others, I
suggest that most individuals are drawn to opinion-reinforcing information, but
they do not exhibit a systematic bias against opinion-challenging information. In
some circumstances, individuals may even seek out novel arguments with which they
disagree.
Prior research results on both sides of the selective exposure debate can be
explained in terms of a systematic preference for opinion reinforcement paired
with a weaker and less consistent attitude toward opinion challenge. For example,
Mutz and Martin (2001) take their ndings as evidence that people prefer opinion-
reinforcing partisan sources to those that include other opinions, but there is another
interpretation. Their data describe individuals preferences when selecting among
partisan sources, not between partisan sources and those that present a more diverse
range of views. That is, in the evidence presented, the presence of opinion-reinforcing
and the absence of opinion-challenging information are conated. This is a useful
starting point, but it is not possible touse these data todetermine whether respondents
were drawn to the sources with which they agreed, avoided those with which they
disagreed with, or both. I argue that these data reect a form of de facto selective
avoidance, motivated by an attraction to opinion-reinforcing information, not an
aversion to opinion-challenging information. On this view, I suggest that many of the
individuals who choose an opinion-reinforcing partisan source would have preferred
a source representing multiple opinions if one was available.
Mechanisms motivating distinct treatment of two information types
Before turning to methods, a brief discussion of the psychological mechanisms
underlying the hypothesized preferences is in order. Selective exposure is histori-
cally premised on Festingers (1957) theory of cognitive dissonance, which posits
that (a) attitude-consistent postdecisional information elicits positive feelings in
the decision-maker about the decision and him or herself, and that (b) attitude-
inconsistent information produces dissonance, a state of mental discomfort and
unease. Individuals are, for obvious reasons, motivated to reduce dissonance and
one way to do this is to seek out opinion-reinforcing information although avoiding
opinion-challenging information. There are, however, several mechanisms that help
to explain why individuals might engage in one form of selectivity (reinforcement
seeking) but not the other (challenge avoidance).
The rst mechanism is Festingers (1957) assertion that people prefer to min-
imize dissonance in the easiest way possible. Although avoidance is one strategy,
individuals can also strive to reduce dissonance after an information encounter.
For example, it may be that counter arguing opinion-challenging information or
seeking additional opinion-reinforcing information can reduce dissonance to a level
that is comparable to avoiding exposure to the challenge in the rst place, and that
this is easier than carefully avoiding all opinion-challenging political information
exposure.
Festingers (1964) revised model of cognitive dissonance offers a second mech-
anism for explaining why challenge avoidance is less common than reinforcement
seeking. As Frey (1986, p. 59) notes, there are a variety of circumstances under
which both opinion-reinforcing and opinion-challenging information is desirable.
Opinion-reinforcing information produces positive feelings in the decision-maker
and it can help to justify ones position. Although opinion-challenging information
may have a negative emotional valence, it can still be useful. For example, knowing
what your critics think, what arguments they are likely to make for their position,
and how they are likely to criticize your position, can be valuable if you are preparing
to engage in debate.
Finally, individuals who perceive of themselves as well informed and fair minded
have a strong disincentive against selective avoidance. Cognitive dissonance can
occur when ones positive self-image is challenged by evidence that ones actions
are inconsistent with that favorable assessment. In this case, intentionally avoiding
opinion-challenging information would conict with the perception that one fairly
weighs alternative arguments when forming opinions, and so the behavior would
itself become a source dissonance. On this view, the reduction in dissonance resulting
from fewer encounters with other viewpoints would be offset by the dissonance
produced by actively screening out political disagreement. The net benets would be
ambiguous at best.
In sum, although most selective exposure research has assumed that avoiding
opinion-challenging information and seeking opinion-reinforcing information are
equally likely, there are compelling theoretical reasons to expect that one will be
more common than the other. Furthermore, there is empirical evidence that support
seeking and challenge avoidance are distinct. The following hypotheses were formed
to test this claim.
H1: Individuals desire exposure to opinion-reinforcing information.
H1a: The stronger an individuals opinion, the more likely he or she is to use a source that
supports that opinion.
H1b: Individuals who have more control over their information exposure will have greater
familiarity with information supporting their opinion than those with less control.
H2: Individuals aversion to opinion-challenging information is weaker than their desire for
opinion-reinforcing information.
H2a: The strength of an individuals opinion is less likely to deter the use of an
opinion-challenging source than to promote the use of an opinion-reinforcing source.
H2b: Control over the information environment is less likely to reduce familiarity with
opinion-challenging information than it is to increase familiarity with opinion-reinforcing
information.
Methods
The online political information environment provides a unique opportunity to
explore exposure preferences. There are many types of online sources, including
major news organizations websites, web-based political discussion boards, political
blogs, and political e-mails (including those from issue-oriented groups, those
forwarded by family and friends, etc.). Two characteristics of these online media
are particularly important. First, the range of opinions accessible online is wider
than with older news media, such as television or newspapers. Politically extreme
groups have a signicant presence online (e.g., Zook, 1996), and partisan news sites
are commonplace (e.g., Atton, 2003). Second, the mechanisms for controlling which
opinions one encounters are powerful and increasingly effective (Bimber & Davis,
2003, p. 152; DiMaggio, Hargittai, Neuman, &Robinson, 2001; Neuman, 1996, p. 15).
Information seeking in this environment is a more active endeavor than it was in an
earlier era characterized by fewer outlets and more rigid publication schedules, and
is much more easily inuenced by individual preferences and attitudes (Valentino,
Hutchings, Banks, & Davis, 2009). Search engines, news aggregations services (such
as Google News), and partisan news sites afford opportunities for searching and
ltering information that are unparalleled in traditional news media. Given the
signicant opportunities for shaping information exposure online, the choices that
individuals make should more closely reect their underlying preferences. Thus, use
of online political news affords individuals increased control over their information
environment.
2
Survey data are used to assess the relationship between control over the infor-
mation environment and exposure to political information. The author collaborated
in the design of a national telephone survey that was sponsored by the Pew Internet
& American Life Project and administered by Princeton Survey Research Associates
betweenJune 14, 2004 andJuly 3, 2004, shortly before the Democratic andRepublican
Party conventions. Respondents were contacted via a random-digit sample of tele-
phone numbers, and interviewers asked to speak with either the youngest adult male
at home at the time of the call or, if no males were present, the oldest adult female.
This sampling strategy helps to ensure that young men and older women, who are
less likely to participate in telephone surveys, are well represented in the sample. The
technique has been shown to produce results similar to a random sample of all adults
living in the household (Keeter, Miller, Kohut, Groves, & Presser, 2000). The overall
response rate was 31.2% (based on a 77% contact rate, 43% cooperation rate, and
94% completion rate), yielding a representative sample of 1,510 English-speaking
adult Americans. Comparing the demographic characteristics of the unweighted
sample to those of the 2004 Annual Social and Economic Supplement Survey (ASES)
(U.S. Census Bureau, 2006) reveals that respondents tended to be slightly older
and more educated than the national average, were more likely to be female, and
were less likely to be of Hispanic descent. These differences were generally small.
Weighting parameters were generated based on a comparison of the demographic
characteristics of survey respondents to those of the Census Bureaus March 2003
ASES. These weights, which help to correct for known biases in telephone surveys,
were used in all analyses. Nevertheless, there may also be a sampling bias not evident
in the demographic characteristics. The survey included several factors potentially
related to individuals willingness to participate, and these are used as controls in the
analyses.
The survey included several measures that were relevant to this research. (See
Appendix for question wording.) In order to assess individuals ability to shape
their information environment, the survey included a battery of questions about
respondents use of the Internet to obtain political information, including their use
of partisan websites in the past year and the frequency with which they use online
news sources. Online news use is an aggregate measure that includes both web- and
e-mail-acquiredinformation. Web-basednews affords signicantlymore information
control than e-mail, as web pages must be sought out, while incoming e-mail
messages often arrive unsolicited. The survey does not distinguish between the two
technologies, however, because both offer more control on average than is offered by
ofine sources. These measures were usedtodiscriminate amongrespondents interms
of their opportunity to utilize the selectivity enhancing capabilities of the Internet.
The survey included a variety of controls, including campaign interest, measured
by how closely the respondent followed the campaign; the strength of candidate
support; and the number of ofine news sources used on a typical day. Finally, the
survey included several demographic measures including age, education, gender, and
race/ethnicity.
3
The assessment of political information exposure was based on respondents
exposure to information about the 2004 presidential election. Respondents were rst
asked a series of questions to determine which of the two leading candidatesGeorge
W. Bush, the Republican, or John Kerry, the Democraticthey preferred. Overall,
46% preferred Bush, 45% preferred Kerry. The remaining 9% refused to choose
between Bush and Kerry, either because they were undecided or because they would
only vote for a third-party candidate.
Respondents were subsequently asked about their familiarity with a series of
opinion statements about the two candidates. The series included two arguments
supporting and two challenging each candidate, for a total of eight arguments in
all (see the Appendix for statement wording). Generating the list of statements was
an iterative process. First, the survey design team assembled a list of statements
used in other recent pre-election surveys. Next, the author consulted Googles
directory service to identify four high-prole campaign websites. Content of these
sites, including talking points and advertisements, was then coded in order to identify
common themes. Finally, the original list of statements was modied to better reect
the concepts identied through the coding process. The nal list included a mix of
high- and low-prominence statements to help ensure that some arguments would be
unfamiliar even to those who regularly followed the campaign.
Two exposure scores were computed based on familiarity with these statements.
The opinion-reinforcement score is a summative measure based on individuals
familiarity with the statements favoring their preferred candidate or criticizing the
opponent, with respondents receiving one point for each argument they heard at
least once in a while. The opinion-challenge score is computed using the other four
items. The resultant scores are summarized in Table 1.
As evidenced by the section of the table that examines Bush and Kerry supporters
separately, the distributions of reinforcing and challenging statement exposure are
different for the two groups. This reects the fact that the opinion statements were
interpreted differently for the two types of votersan argument that was treated as
reinforcing for Bush supporters was said to be challenging for Kerry supporters and
vice versaand that Bush arguments (pro and con) were better known than Kerry
arguments overall (Bush = 3.0, Kerry = 2.3, t = 24.79, df = 1,369, p < .001).
Table 1 Summary of Argument Familiarity Scores
n Min Max M SD
Overall (Bush and Kerry supporters)
Pro-Bush statements 1,370 0 2 1.49 .72
Anti-Bush statements 1,370 0 2 1.55 .58
Total statements about Bush 1,370 0 4 3.04 1.04
Pro-Kerry statements 1,370 0 2 1.06 .80
Anti-Kerry statements 1,370 0 2 1.21 .78
Total statements about Kerry 1,370 0 4 2.27 1.28
Bush supporters
Reinforcing (pro-Bush +anti-Kerry) 694 0 4 3.04 1.07
Challenging (pro-Kerry +anti-Bush) 694 0 4 2.41 1.19
Kerry supporters
Reinforcing (pro-Kerry +anti-Bush) 676 0 4 2.83 1.06
Challenging (pro-Bush +anti-Kerry) 676 0 4 2.33 1.36
Undecided/Other candidate
a
Bush-favorable (pro-Bush +anti-Kerry) 140 0 4 2.09 1.44
Kerry-favorable (pro-Kerry +anti-Bush) 140 0 4 2.25 1.24
Note: Weighted data (unweighted n = 1,510).
a
Includes individuals who declined to state a preference when asked to choose between Bush
and Kerry.
Given the difference in the interpretation and distribution of the dependent variables
for Bush and Kerry supporters, the two groups are treated separately in all analyses.
The method of assessing exposure employed in this study is unique, and offers
some advantages over previously usedalternatives. Avariety of self-reportedmeasures
of exposure have been used in previous surveys. Some studies ask respondents to
provide an estimate of aggregate agreement with the news media they use. For
example, howoften does the respondent disagree with the political views encountered
when reading the newspaper (Mutz & Martin, 2001). Other studies focus on recalling
specic media usage behaviors. For example, did the respondent pay attention to a
TV news story about a specic candidate (Chaffee, Saphir, Graf, Sandvig, & Hahn,
2001). One limitation of measures such as these is that arriving at a response entails
complex mental processes. In the rst case, the assumption is that the response will
accurately account for all news encounters in the specied outlet. A comprehensive
review of these encounters seems unlikely, but even a more limited review would
be taxing. The second example is equally demanding. In order for this technique to
be effective, the respondent must recall with a high degree of specicity his media
consumption practices. He must survey his memory for every instance in which he
encountered political information, he must recall what the overall position taken by
the news item was (e.g., did it support the preferred candidate or someone else),
and he must recall how he reacted (e.g., did he read the item or not?). These are
burdensome tasks, and there are numerous opportunities for bias to inuence the
data. Given that individuals responses to survey questions are strongly inuenced by
heuristics and top-of-the-head considerations (Zaller, 1992), the measures will tend
to reect recent news encounters or, more concerning, a generalized belief about
how the respondent thinks he approaches political news. For example, an individual
who believes that he usually considers all sides equally may use this rule of thumb to
help him remember which viewpoints he attended to.
The alternative to relying on recalled media exposure employed in this study is
to measure recalled argument exposure. Argument exposure is a direct consequence
of media use, so that differences in argument exposure reect differences in media
exposure behavior. The memory task, however, is less difcult. Rather than reviewing
all media exposure events, the individual focuses on whether or not he has previously
encountered a series of arguments. By comparing argument exposure, the product
of media use, to an individuals attitude, we can arrive at a reliable and a relatively
straightforward measure of contact with opinion-reinforcing and opinion-contrary
information.
One concern about this measure is that it still depends on respondents memory,
and is therefore susceptible to error. That is, respondents may have forgotten some of
the arguments they heard. The measurement technique, however, should not produce
systemic biases associated with political attitude. Although people are more likely to
remember encountering information that runs counter to their expectations (Koriat,
Goldsmith, & Pansky, 2000; Stangor & McMillan, 1992), there is little evidence that
attitude inuences memory (Eagly, Chen, Chaiken, & Shaw-Barnes, 1999; Holbrook,
Berent, Krosnick, Visser, & Boninger, 2005, p. 750). Individuals do not exhibit a
general bias toward either attitude-consistent or attitude-discrepant information.
Furthermore, even if individuals were to more accurately recall information that
supports their viewpoint than information that does not, the method of assessing
the different forms of selective exposure used in this study would still be effective.
There is no reason to expect selective retention effects to be more pronounced among
online news users than among nonusers, so any correlation between argument
familiarity and exposure control is the product of selective exposure. If selective
retention did occur, it would be expected to lead to an underestimate of exposure to
opinion-challenging information, exaggerating the evidence of selective avoidance.
Thus, selective memory would make results that contradict the hypotheses advanced
here more likely.
Results
The rst stage of analysis examines the most direct measure of exposure: respondents
self-reported use of web-based sources of political information. Focusing on indi-
viduals use of partisan media sources provides a useful starting point for examining
selectivity, although its value is somewhat limited by the fact that the presence
of opinion reinforcement and the absence of opinion challenge are linked. (This
Table 2 Political Site Use by Candidate Preferences (Internet users only)
n Liberal Site Democratic Site Conservative Site Republican Site
Overall 1,017 9.6% 8.1% 10.4% 8.3%
Bush supporters 469 5.6 3.9 12.7 11.6
Strong 306 4.9 4.3 14.9 15.1
Not strong 162 6.8 3.1 9.3 4.9
Kerry supporters 459 14.8 12.9 9.0 5.7
Strong 139 20.9 20.9 13.0 5.8
Not strong 320 12.5 9.4 7.2 5.3
Undecided/other 90 3.4 5.6 5.6 4.5
limitation is addressed in subsequent analyses.) Table 2 compares the use of partisan
websites by supporters of each of the two leading candidates, and those who were
undecided or who only expressed a preference for a third-party candidate. Together,
these data shed some light on how political opinions inuence information source
use. We begin by considering the distribution of responses in the sample before
turning to statistical analyses of these results.
Two implications of the data are immediately evident. First, partisan websites
use is relatively uncommon: Only about one in 10 Internet users visited each of
the sites identied. Second and more interesting, selective exposure appears to
occur in some form. A larger proportion of Bush supporters used conservative and
Republican websites than used liberal or Democratic websites, and the pattern was
reversed among Kerry supporters. As a consequence, site usage appears polarized,
with individuals preferring sites that are aligned with their candidate preferences.
Respondents who did not support either candidate are included for comparison.
Unsurprisingly, this group exhibited the lowest levels of partisan website use, and
usage levels were comparable across the sites. At this level of analysis, however, we are
unable to distinguish between a preference for opinion-reinforcing and an aversion
to opinion-challenging information.
Strong candidate support not associated challenge avoidance
By splitting supporters of the two candidates into strong supporters and those whose
support for their preferred candidate was not strong we can distinguish between
preferences regarding opinion-reinforcing and opinion-challenging information. As
historically interpreted, selective exposure theory predicts that strong supporters
would be more likely than weaker supporters to use opinion-reinforcing sites at the
expense of opinion-challenging sites (Brannon, Tagler, & Eagly, 2007; Stroud, 2008).
In contrast, this paper asserts that although strong support will be associated with
more exposure to reinforcing sources, it will not be linked to a comparable drop in
exposure to challenging sources.
4
In other words, strong support will produce a net
Table 3 Effect of Candidate Preference (Bush vs. Kerry) on Political Site Use (Internet users
only)
Liberal Site Democratic Site Conservative Site Republican Site
Stage 1 Stage 2 Stage 1 Stage 2 Stage 1 Stage 2 Stage 1 Stage 2
Supports
Bush
1.22
.74
1.53
1.07
.21 .26 .63
.07
Strong
support
.10 .33 .52
.65
.45
.49 .43 .12

Bush
strong
.82 .65 .09 .98
n 908 907 902 906

Cox and
Snell R
2
.04 .05 .07 .07 .03 .03 .04 .05
Notes: Weighted data (unweighted n between 919 and 924). Cell entries are coefcients
of logistic regression predicting use of specied site. Controls for education, attention to
campaign, age, and gender were included in the analysis, but are omitted here for clarity.
p < .1.
p < .05.
p < .001.
increase in exposure that is biased toward the favored candidate without substantially
reducing contact with other viewpoints.
Returning to Table 2, the data appear to be consistent with these expectations.
Candidate support levels do seem to inuence use of sites favoring the preferred
candidate. Nearly one in ve strong Kerry supporters visited each of the Kerry-
favorable sites, almost double the proportion of weaker supporters who visited those
same sites. Bush supporters use of Bush-favorable sites exhibited a similar pattern.
In contrast, strength of candidate support does not seem to have much inuence
of the use of sites favoring the candidate opposed. In fact in this survey, a larger
proportion of strong Kerry supporters had visited a conservative site than weaker
supporter.
There are a number of potential confounds that are not accounted for in the
table above. For example, strength of support may be associated with overall political
interest. On this view, any tendency to avoid other perspectives might be offset by
the greater tendency to use all types of sources. A series of logistic regressions are
used to more rigorously test the inuence of candidate preference on political site
use. In the models, site use is predicted based on candidate preference, strength of
preference, and the interaction between these two terms. The models are computed
in two stages, beginning with candidate preference and strength and then introducing
the interaction term. In order to control for confounds, education, attention to the
campaign (a proxy for campaign interest), age, and gender were also included as
covariates in the model. The results are shown in Table 3.
These statistical tests demonstrate that the trends noted above hold up under
conditions of multivariate control, providing evidence of the revised conceptual-
ization of selective exposure proposed in this article. From the stage 1 models, we
see that Kerry supporters are more likely than Bush supporters to use either liberal
or Democratic websites, and Bush supporters are marginally more likely to use a
conservative site. In other words, respondents are more likely to use the site associated
with their preferred candidate than with the candidates opponent. This could be a
consequence of seeking reinforcement or avoiding challenge. The stage 2 models are
more informative. In the presences of the interaction term, the coefcient on support
strength when predicting Democratic site use is positive, indicating that strong Kerry
supporters are more likely than weak supporters to use a Democratic site. Similarly,
the coefcient on the interaction between support strength and Bush support is
positive, though only marginally signicant, when predicting Republican site use,
suggesting that strong Bush support is associated with Republican site use. Thus, as
predicted in H1a, strong supporters are more likely to use opinion-reinforcing sites
than weaker supporters. The inuence of support strength on use of sites favoring the
opposed candidate, however, is not signicantly different than zero. In other words,
strong support is less likely to deter use of opinion-challenging sources (the effect
was not signicant) than to promote use of opinion-reinforcing sources (the effect
was signicant), as predicted by H2a.
A complementary source of evidence of selective exposure via news media choice
is respondents use of mainstream and alternative news outlets. If individuals desire
opinion-reinforcing information they can turn to the more partisan alternative news
media readily available online. Conversely, if individuals want to avoid opinion-
challenging information, one strategy is to avoid newspapers and television news,
which are explicitly marketed as giving all sides of a debate comparable consideration
and fair treatment. Table 4 compares the use of these two types of news media by
respondents candidate preferences. Consider rst the use of partisan sources, which
are likely to offer extensive opinion reinforcement. The results suggest that more
strong supporters than weak supporters get news fromthese outlets. This is consistent
with the idea that people seek out opinion-reinforcing sourcing of information(H1a).
The results, however, provide no evidence that strong partisans are abandoning the
mainstream news media. To the contrary, among Kerry supporters in this sample,
a strong candidate preference is associated with increased use of newspapers and
television news. Thus, it does not appear that these Internet users are trying to
avoid sources that include other viewpoints (H2a). Testing the inuence of candidate
preference on media use in the context of multivariate controls (as partisan site use is
tested above) there are no signicant effects associated with candidate preference or
strength of support. This suggests that attitude strength plays a relatively trivial role
in shaping whether or not individuals use online alternatives to the mainstream news
media.
In sum, the usage patterns evident in the data are consistent with the hypotheses
concerning source use (H1a and H2a), although the strength of the effects is only
Table 4 News Outlet Use by Candidate Preferences (Internet users only)
(n) Mainstream News Media Alternative News Media
Overall 1,017 83.9% 10.8%
Bush supporters 469 85.7 10.9
Strong 306 86.3 11.5
Not strong 162 85.2 9.4
Kerry supporters 459 83.4 11.6
Strong 139 89.1 13.9
Not strong 320 81.2 10.6
Undecided/other 90 76.7 5.6
modest. Individuals do prefer sources offering opinion-reinforcing information to
those offering opinion-challenging information, but for highly committed supporters
this effect is driven by a desire for opinion reinforcement more than by a desire
to avoid opinion challenges. These analyses have focused on exposure differences
that are based on individuals commitment to their preferred candidate and are
limited by their emphasis on one-sided sources. Another approach to understanding
selective exposure preferences is to examine how individuals exposure to political
information changes when they have more opportunity to be selective.
Online news use not associated challenge avoidance
Now we turn to individuals familiarity with the eight opinion statements about the
leading candidates in the 2004 presidential election. Focusing on opinion statement
exposure (as opposed to source exposure) more effectively decouples opinion-
reinforcing and opinion-challenging information exposure, enabling us to assess
the distinct inuence of these factors on individuals exposure decisions. Examining
Table 2 again, notice that supporters of both candidates knew more on average about
their preferred candidate than about the challenger (t = 14.171, df = 693, p < .001
for Bush supporters, and t = 10.537, df = 675, p < .001 for Kerry supporters). This
again suggests that there is a relationship between individuals political beliefs and
their exposure to campaign informationevidence that some form of selective
exposure is occurringbut it does not allow us to assess the hypotheses advanced in
this article.
Knowing more about a preferred candidate tells us nothing about the respective
contributions of opinion-reinforcement seeking and opinion-challenge avoidance.
The hypotheses assert that although individuals do seek reinforcement, efforts to
avoid opinion-challenging information will be comparatively weak. On this view,
we should expect that individuals who have more control over their information
exposure will be familiar with more of the arguments supporting their candidate, but
the inuence on exposure to opinion-challenging arguments will be slight. When
Table 5 Effect of Online News Use on Exposure to Opinion-Relevant Statements
Opinion-Reinforcing Opinion-Challenging
Statements Statements
Bush
Supporters
Kerry
Supporters
Bush
Supporters
Kerry
Supporters
Online news frequency .07
.11
.03 .10
No. ofine sources used .09
.09
.09
.08
Campaign interest .40
.24
.35
.22
Education .09
.13
.15
.13
Age .18
.11
.13
.23
Strong candidate support .05 .04 .02 .03

n 676 663 676 663
R
2
.30 .18 .22 .19
Note: Weighteddata (unweightedn = 681[Bushsupporters] andn = 656[Kerry supporters]).
Cell entries are standardized coefcients of OLS regression.
p < .05.
p < .01.
p < .001.
testing these predictions, we must control for numerous other factors that are known
to inuence familiarity with political arguments.
A series of OLS regression models are used to examine differences in voters
exposure topolitical opinionstatements. As notedabove, the distributionof argument
familiarity varied depending on which candidate a voter supported, so Bush and
Kerry supporters are treated separately. For each group of supporters, there are
two models, one predicting familiarity with opinion-reinforcing arguments, and the
other predicting familiarity with opinion-challenging arguments. The results of these
models are summarized in Table 5.
The rst pair of models, which predict familiarity with opinion-reinforcing
statements, explain around a quarter of the variation (Bush support R
2
= .30, Kerry
support R
2
= .18) in respondents argument familiarity. Several of the control factors
are associated with increased exposure to opinion reinforcement. For both groups of
supporters, campaign interest, education, and age are each positively correlated with
exposure. Strength of candidate support also merits brief comment. The coefcient
on this variable is nonsignicant for both groups of supporters, but its sign is
consistent with evidence presented above that strong supporters are more likely than
weaker supporters to engage in selective reinforcement seeking.
The key variable given the theoretical concern of this paper is the frequency
with which respondents got campaign news online, as this practice affords users a
unique opportunity to shape their political information environment. The sign of the
coefcient was positive and signicant for both groups, although the association was
stronger among Kerry supporters. The magnitude of inuence of this factor among
Kerry supporters is particularly noteworthy. The standardized coefcient rivals that
of education or age, and although it is not as large as the coefcients associated
with campaign interest, it is of the same order of magnitude as that important
political variable. Overall, these results support the hypothesis that familiarity with
opinion-reinforcing information will growas exposure control increases (H1b). Both
Kerry and Bush supporters use the additional control afforded by the Internet to
augment their exposure to opinion-favorable information.
The second pair of models predicts respondents familiarity with opinion-
challenging information. These models explain about a fth of the variance (Bush
support R
2
= .22, Kerry support R
2
= .19), and most factors have a similar inuence
on Bush and Kerry supporters. Importantly, campaign interest and age are again
among the most inuential items, providing additional evidence that many factors
promote exposure to both types of information (Chaffee et al., 2001). It is also
appropriate to mention the results concerning candidate support strength again. As
before, the coefcient is nonsignicant for both group, but this time sign is negative.
If candidate commitment is promoting selective avoidance of opinion challenges, its
inuence is very small.
Returning to the key theoretical variables, we nd that Bush and Kerry supporters
differ, and the relationship is only signicant for Kerry supports. Surprisingly,
the frequency of online news use among this group is correlated with increasing
exposure to opinion-challenging statements. (The sign on the coefcient is also
positive for Bush supporters.) Traditional conceptualizations of selective exposure,
which predict that individual try to avoid other viewpoints, would lead us to
expect the coefcient to be negative. Instead, we nd no evidence that individuals
with more control over their information environment limit contact with opinion
challenges. This is a more serious challenge to the notion that individuals avoid
other viewpoints than anticipated at the outset of this article. The hypothesis (H2b)
predicted that exposure control would be associated with a decrease in exposure
to opinion-challenging information that was small compared to the increase in
exposure to opinion-reinforcing information. Instead, we nd that Kerry supporters
use their expanding inuence to increase their exposure to other points of view, and
that Bush supporters exposure is not signicantly reduced. In sum, these analyses
provide strong evidence that control over the information environment promotes
exposure to opinion-reinforcing information, and that the tendency to avoid opinion
challenges in comparatively weak.
Discussion
Politically motivated selective exposure has been a topic of debate for more than
half a century, and the explosive growth of specialized news media and high-control
news environments, such as those available online, have brought renewed interest
in this question. Americans preferences regarding exposure to political information
will profoundly shape the political environment in the years to come. Understanding
whether people have a preference for opinion reinforcement and/or an aversion
to opinion challenges is important because the implications of these preferences
are strikingly different. One form of selective exposure poses a greater threat to
democratic deliberation than the other. Although seeking out reinforcement may
produce deeper convictions and more passionate beliefs, such behavior does not
necessarily reduce exposure to other perspectives. Avoiding other perspectives,
on the other hand, presents a number of threats. Exposure to other opinions is
important because it fosters political tolerance and can improve group deliberation
processes. As this exposure drops, the evidence suggests that our society will become
more polarized and politically fragmented, that political tolerance will drop, and
that citizens will be less able to effectively deliberate over important political
problems.
The research described here suggests that individuals political attitudes do
inuence their attention to relevant news and information. However, contrary
to prior interpretations of selective exposure theory, the data demonstrate that
seeking opinion-reinforcing and avoiding opinion-challenging information are not
equivalent. The results support the hypotheses that individuals are using control
over their political information environment to increase their exposure to opinion-
reinforcing information, but that they are not using this control to systematically
screen out other opinions. On one hand, individuals seeking political information in
anonline environment that facilitates politically motivatedselectivity seek out sources
that support their political opinions, and have a larger repertoire of arguments with
which to justify their opinions. On the other hand, individuals who get their news
online are not avoiding sources with which they disagree, and they are no less familiar
with arguments justifying other perspectives. To the contrary, Kerry-supporting
online news users were more aware of the rationales for supporting Bush than Kerry
supporters who tend not to use online outlets.
One interpretation of the argument exposure results among Kerry supporters
would be to echo the claim made by (Chaffee et al. 2001) that selective exposure
does not occur, and that it is more important to focus on the factors that promote
exposure to both opinion-reinforcing and opinion-challenging information. In this
case, it appears that given more control, many people are simply exposed to more
information about the election. To arrive at such a conclusion, however, would
be to disregard the strong evidence that people know more about the candidate
they prefer than the other candidate, and they are more likely to use a website
promoting this preference. I suggest that to explain these contrasting results, we
should view opinion-reinforcing information as a more powerful predictor of
exposure than opinion-challenging information. This explanation simultaneously
accounts for the fact that there is no evidence that people systematically avoid
exposure to other viewpoints, and that they still tend to know more supporting their
own.
The context of this study, a highly contentious U.S. presidential election, affords
a robust test of the prediction that people do not lter out opinion challenges.
Situations that evoke a sense of threat tend to promote selective exposure among
authoritarians (Lavine, Lodge, & Freitas, 2005). In 2004, supporters on both sides
regularly argued that electing the other candidate would put U.S. interests at greater
risk both at home and abroad, claims which suggest that many people viewed the
outcome of the election as potentially threatening. In the face of these fears, people
are particularly prone to avoiding opinion-challenging information, which makes
the lack of evidence for this behavior all the more noteworthy.
The observed difference between the two groups of supporters, both in terms of
their political website use and opinion statement exposure is provocative. Table 2
suggests that strong Kerry supporters may be slightly more likely to use a website
representing the opposing political ideology than strong Bush supporters (although
Table 3shows that the inuence of support strengthinthis sample was not signicant).
More importantly, Table 5 indicates that Kerry supporters see an increase in exposure
to opinion challenges when getting news online, but Bush supporters do not. One
interpretationof these results is that Kerry supporters, who tend to leanto the political
left, are more attentive to opinion-challenging information than Bush supporters,
who lean to the right.
The idea that conservatives are less likely than liberals to seek out other opinions
has precedent in the research literature. A number of previous studies have revealed
similar patterns, dating as far back the Erie County study data. Sears and Freedman
(1967, p. 199) note, for example, that when broken down in terms of political afl-
iation, Republicans tended to encounter more information favoring the Republican
candidate than favoring the Democratic candidate, though Democrats encountered
a balanced mix of political information. Mutz (2006, p. 33) notes a similar trend in
her analyses of interpersonal discussion, nding that conservatives and Republicans
tend to engage in less cross-cutting discussion than liberals or Democrats. And
Iyengar et al. (2008) only nd evidence of selective exposure among Bush supporters
in their study of the 2000 U.S. presidential election. These results can be situated
within the broader debate over the rigidity-of-the-Right hypothesis, which asserts
that conservatives (especially right-wing authoritarians) tend to be less receptive to
attitude challenges and more resistant to attitude change (Amodio, Jost, Master, &
Yee, 2007; Jost, Glaser, Kruglanski, & Sulloway, 2003; Tetlock, 1989). A thorough test
of these claims is beyond the scope of this paper, but the results reported here suggest
that further examination of the role of political ideology on the selective exposure
phenomenon is merited.
These results paint a moderately encouraging portrait of individual exposure
preferences. They suggest that peoples use of the news media, although selective, is
not as strongly driven by a desire to avoid political difference as some scholars have
argued (e.g., Mutz & Martin, 2001; Sunstein, 2001). Instead, these results suggest that
individuals are primarily motivated by a desire to encounter information that justies
their opinion when using the news media. It should be noted, however, that this
preference could be unique to mediated communication. The debate over whether
people systematically avoid political disagreement in their interpersonal relationships
is an important complement to the issues raised here (see Huckfeldt, Johnson, &
Sprague, 2004; Mutz, 2006).
The more limited form of selectivity described in this article also raises important
questions about Sunsteins assertion that the Internet fundamentally promotes
isolation from political disagreement. People have not abandoned the mainstream
media for more partisanalternatives, as Sunsteinsuggests they might, and a continued
awareness of competing ideas, attitudes, and opinions would appear to run counter
to the notion of echo chambers. Concerns about polarization are not entirely
inconsistent with the nding reported here: When people must choose among one-
sided partisan sources, seeking opinion reinforcement will be equivalent to avoiding
challenges. Nevertheless, Sunsteins prediction that people will take any opportunity
to screen out other perspectives seems unnecessarily grave.
Several unanswered questions remain. One of these pertains to the individual
exposure decisions that, in aggregate, result in these overall exposure patterns.
The nding that people who get their news online exhibit no less familiarity with
opinions that differ from their own can be explained in at least two ways. On one
hand, it might simply be that individuals do not tend to avoid opinion-challenging
information. In this case, individual decisions about which content (e.g., news
stories) to examine will be uninuenced by the presence of other opinions. On the
other hand, individuals might seek ways to minimize the frequency of exposure to
opinion-challenging information without sacricing their breadth of exposure. On
this view, news consumers might avoid some content as long as they could preserve
contact with a range of other perspectives via other sources.
A limitation of this study is that it does not address the possibility that preferences
are inuenced by individual characteristics and social contexts. For example, there is
signicant evidence that attention to attitude-discrepant information is greater when
such exposure is anticipated to be useful (Frey, 1986; Valentino et al., 2009). To
the extent that politically active individuals are invested in being able to respond to
criticismof their position, we might expect political engagement to be associated with
cross-cutting exposure. This would suggest that individuals with limited political
interest and weak preferences avoid opinion challenges, though those with stronger
beliefs or greater interest levels seek challenge out. Future work should address this
shortcoming.
The emphasis on opinion statements is another potential limitation of this study.
These statements may disproportionately represent informationfeaturedincampaign
ads, and thus may be less prone to selective exposure effects. Other exposure measures
might produce stronger effects. Nevertheless, the relationship between the forms of
selective exposure should remain the same.
Finally, it should be noted that the data used here say nothing about individuals
comprehension or evaluation of political information. It seems reasonable to think
that many people encounter ideas with which they disagree in an opinion-reinforcing
setting. For example, both liberal viewers of the Daily Show and conservative listeners
of the Rush Limbaugh Show may hear a bit about how the other side defends its
position without ever seriously considering the merit of those arguments. In other
words, recalling exposure to opinionstatements does not meanthat the individual has
critically engaged the content. Exposure is necessary but not sufcient to ensure that
anindividual integrates the arguments intohis or her broader political understanding.
Still, other research suggests that recognizing that other viewpoints exist, even when
such views are readily dismissed, can have a positive inuence on decision making
(e.g., Nemeth, 1986).
The lessons learned here are important despite the questions that remain. This
article has argued that a desire for exposure to opinion-reinforcing information is
not synonymous with an aversion to other opinions. Although individuals are likely
to use the information resources available to them to expand their familiarity with
the arguments that support their position, there is no reason to expect that they
will systematically screen out exposure to other opinions. In a media landscape
characterized by the presence of highly partisan outlets alongside those that are more
balanced, most people will not abandon contact with other perspectives. Although
strongly motivated to seek out informationconsistent with their opinions, individuals
may simultaneously act to maintain familiarity with a range of counterarguments.
Thus, a news environment that facilitates exposure choices will not necessarily lead
to political intolerance or a breakdown of deliberative democracy.
Acknowledgments
The author is grateful to Paul Resnick for his advice and encouragement throughout
the project, and to Bruce Bimber, Paul Edwards, and Russ Neuman for their
many helpful insights. Thanks also to the Pew Internet & American Life Project
for sponsoring the survey research and to John Horrigan and Lee Rainie for their
valuable contribution to its design and analysis. Finally, thanks to the anonymous
reviewers for their feedback on prior versions of this manuscript.
Appendix
Candidate preferences, opinion statements, and campaign interest
Suppose the election for president were being held today and the candidates were
George W. Bush, the Republican; John Kerry, the Democrat; and Ralph Nader, an
Independent candidate. Who would you vote for? (40% Bush, 37% Kerry, 6%
Nader, 17% undecided/other). As of today, do you lean more toward Bush, the
Republican; Kerry, the Democrat; or Nader, the Independent? (44% Bush, 39%
Kerry, 7% Nader, 10% undecided/other). Suppose there were only two presidential
candidates on the ballot and you had to choose between George W. Bush, the
Republican; and John Kerry, the Democrat? If the election were held today, who
would you vote for? (46% Bush, 45% Kerry, 9% undecided/other).
Im going to read different arguments people make about the Presidential
candidates and their policies. Please tell me how often you have heard or read each
argumentfrequently, just once in a while, or never. The Bush administrations
policies have helped the economy begin to recover (76% heard at least once in a
while); George Bush is a stronger leader than John Kerry in the war on terrorism
(70%); John Kerry changes his positions on the issues when he thinks it will help
him win the election (70%); John Kerry has a history of accepting money from
special interest groups (51%); The Bushadministrationmisledthe Americanpublic
about the reasons for going to war about Iraq (93%); Some Bush administration
policies are a threat to basic civil rights and civil liberties (61%); John Kerry
has a better strategy than George Bush for creating peace in Iraq (53%); John
Kerry will end special treatment for corporations and wealthy Americans (51%).
Statements about the preferred candidate were read rst, and were rotated within
blocks.
How closely have you been following news about the upcoming Presidential
election? Very closely (1), somewhat closely, not too closely, or not at all closely (4)?
(M = 2.0, SD = .9).
Sources of information
Please tell me if you ever get news or information from each of the following
sources. If yes, then ask Did you happen to gets news or information from
[source] yesterday, or not? before proceeding to next source. Sources: newspapers
(51%), television (74%), magazines (21%), radio (54%), friends and family (41%).
Respondents who used either a newspaper or television news are said to have used
a mainstream news outlet (82%). The number ofine sources used was calculated
by counting the number of afrmative answers to the second question. (M = 2.4,
SD = 1.3).
Do you ever go online to access the Internet or World Wide Web or to send and
receive e-mail? (67% Yes) If so, Do you ever get news or information about the
candidates and the campaign on the Internet or through e-mail? (28% Yes) If so,
How often do you get news or information about the candidates and the campaign
on the Internet or through e-mail everyday or almost everyday (1), several times a
week, several times a month, or less often (4)? (M = 2.2, SD = 1.0).
Internet users were also asked, In the past 12 months, did you happen to visit any
of the following websites? The website of an international news organization, such
as the BBC or Aljazeera? (18%); The website of an alternative news organization,
such as Alternet.org or NewsMax.com? (11%); The website of a politically liberal
organization, such as People for the American Way or Moveon.org (10%); The
website of a politically conservative organization, such as the Christian Coalition or
the American Enterprise Institute (10%). For questions about the candidates sites,
the sample was evenly split between two forms. Form A: GeorgeWBush.com, the
Presidents ofcial reelection website (9%); JohnKerry.com, the ofcial website
of the Kerry campaign (10%). Form B: RNC.com, the ofcial website of the
Republican National Committee (7%); DNC.com, the ofcial website of the
Democratic National Committee (6%).
Notes
1 This study of attitude-basedselectivity is situatedinthe political domain, whichis reected
in the use of the theory and concepts throughout the article. This decision reects the
focus of this work, and in no way diminishes the importance of other types of selective
exposure research (e.g., Zillmann & Bryant, 1985).
2 One concern regarding this approach is that Internet use has historically been correlated
with other characteristics might also inuence political information seeking behavior.
For example, the earliest Internet adopters tended to be male and well educated, though
older Americans have been comparatively slow to adopt the new communication
technology. It is, therefore, necessary to control for these potential confounds in the
analyses.
3 This study does not include a control for political knowledge. Although correlated with
education and political interest, political knowledge is a distinct characteristic and could
have a unique inuence on information exposure. This measure should be included in
future studies of this topic.
4 This prediction is tempered somewhat by the fact that the sites under consideration tend
to be one-sided. As a consequence, the absence of opinion-reinforcing information (and
not the presence of opinion challenges) on opinion-challenging sites could motivate
people to go elsewhere for political information. Nevertheless, we still expect a larger
change in the use of reinforcing sources than in challenging sources if individuals are not
averse to opinion challenges.
References
Amodio, D. M., Jost, J. T., Master, S. L., & Yee, C. M. (2007). Neurocognitive correlates of
liberalism and conservatism. Nature Neuroscience, 10(10), 12461247.
Atton, C. (2003). Reshaping social movement media for a new millennium. Social Movement
Studies, 2(1), 415.
Bimber, B., & Davis, R. (2003). Campaigning online: The Internet in U.S. elections. New York:
Oxford University Press.
Brannon, L. A., Tagler, M. J., & Eagly, A. H. (2007). The moderating role of attitude strength
in selective exposure to information. Journal of Experimental Social Psychology, 43(4),
611617.
Bryant, J., & Miron, D. (2004). Theory and research in mass communication. Journal of
Communication, 54(4), 662704.
Chaffee, S. H., Saphir, M. N., Graf, J., Sandvig, C., & Hahn, K. S. (2001). Attention to
counter-attitudinal messages in a state election campaign. Political Communication, 18
247272.
Delli Carpini, M. X., Cook, F. L., & Jacobs, L. R. (2004). Public deliberation, discursive
participation, and citizen engagement: a review of the empirical literature. Annual Review
of Political Science, 7, 315344.
DiMaggio, P., & Sato, K. (2003). Does the internet balkanize political attention?: A test of the
Sunstein theory. Paper presented at the Annual meeting of the American Sociological
Association, Atlanta.
DiMaggio, P., Hargittai, E., Neuman, W. R., & Robinson, J. P. (2001). Social implications of
the Internet. Annual Review of Sociology, 27, 307336.
Eagly, A. H., Chen, S., Chaiken, S., & Shaw-Barnes, K. (1999). The impact of attitudes on
memory: An affair to remember. Psychological Bulletin, 125(1), 6489.
Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford University Press.
Festinger, L. (1964). Conict, decision, and dissonance. Stanford, CA: Stanford University
Press.
Fischer, P., Jonas, E., Frey, D., & Schulz-Hardt, S. (2005). Selective exposure to information:
The impact of information limits. European Journal of Social Psychology, 35(4), 469492.
Frey, D. (1986). Recent research on selective exposure to information. Advances in
Experimental Social Psychology, 19, 4180.
Holbrook, A. L., Berent, M. K., Krosnick, J. A., Visser, P. S., & Boninger, D. S. (2005).
Attitude importance and the accumulation of attitude-relevant knowledge in memory.
Journal of Personality and Social Psychology, 88(5), 749769.
Huckfeldt, R. R., Johnson, P. E., & Sprague, J. (2004). Political disagreement: The survival of
diverse opinions within communication networks. Cambridge: Cambridge University Press.
Iyengar, S., Hahn, K. S., Krosnick, J. A., & Walker, J. (2008). Selective exposure to campaign
communication: The role of anticipated agreement and issue public membership. The
Journal of Politics, 70(1), 186200.
Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003). Political conservatism as
motivated social cognition. Psychological Bulletin, 129(3), 339375.
Keeter, S., Miller, C., Kohut, A., Groves, R. M., & Presser, S. (2000). Consequences of
reducing nonresponse in a national telephone survey. Public Opinion Quarterly, 64(2),
125148.
Knobloch-Westerwick, S., & Meng, J. (2009). Looking the other way: Selective exposure to
attitude-consistent and counter-attitudinal political information. Communication
Research, 36(3), 426448.
Koriat, A., Goldsmith, M., & Pansky, A. (2000). Toward a psychology of memory accuracy.
Annual Review of Psychology, 51, 481537.
Lavine, H., Lodge, M., & Freitas, K. (2005). Threat, authoritarianism, and selective exposure
to information. Political Psychology, 26(2), 219244.
Lazarsfeld, P. F., Berelson, B., & Gaudet, H. (1944). The peoples choice. New York: Columbia
University Press.
Lowin, A. (1967). Approach and avoidance: Alternative modes of selective exposure to
information. Journal of Personality and Social Psychology, 6(1), 19.
Mendelberg, T. (2002). The deliberative citizen: Theory and evidence. Political decision
making, deliberation and participation, 6, 151193.
Mutz, D. C. (2002). The consequences of cross-cutting networks for political participation.
American Journal of Political Science, 46(4), 838855.
Mutz, D. C. (2006). Hearing the other side: Deliberative versus participatory democracy. New
York: Cambridge University Press.
Mutz, D. C., & Martin, P. S. (2001). Facilitating communication across lines of political
difference: The role of mass media. American Political Science Review, 95(1), 97114.
Nemeth, C. J. (1986). Differential contributions of majority and minority inuence.
Psychological Review, 93(1), 2332.
Nemeth, C. J., & Rogers, J. (1996). Dissent and the search for information. British Journal of
Social Psychology, 35, 6776.
Neuman, W. R. (1991). The future of the mass audience. Cambridge, U.K.: Cambridge
University Press.
Neuman, W. R. (1996). Political communications infrastructure. Annals of the American
Academy of Political and Social Science, 546, 921.
Price, V., Cappella, J. N., & Nir, L. (2002). Does disagreement contribute to more
deliberative opinion? Political Communication, 19, 95112.
Sears, D. O., & Freedman, J. L. (1967). Selective exposure to information: A critical review.
Public Opinion Quarterly, 31(2), 194213.
Stangor, C., & McMillan, D. (1992). Memory for expectancy-congruent and
expectancy-incongruent information: a review of the social and social developmental
literatures. Psychological Bulletin, 111(1), 4261.
Stroud, N. J. (2004). Wheres the party? An investigation of the Internet and partisan selectivity.
Paper presented at the annual meeting of the National Communication Association,
Chicago, IL.
Stroud, N. J. (2008). Media use and political predispositions: Revisiting the concept of
selective exposure. Political Behavior, 30, 341366.
Sunstein, C. R. (2001). Republic.com. Princeton, NJ: Princeton University Press.
Sunstein, C. R. (2002). The law of group polarization. The Journal of Political Philosophy,
10(2), 175195.
Sweeney, P. D., & Gruber, K. L. (1984). Selective exposure: Voter information preferences
and the Watergate affair. Journal of Personality and Social Psychology, 46(6), 12081221.
Tetlock, P. E. (1989). Structure and function in political belief systems In A. R. Pratkanis,
S. J. Breckler, & A. G. Greenwald (Eds.), Attitude structure and function (pp. 129152).
Hillsdale, NJ: Erlbaum.
U.S. Census Bureau. (2006, August 21). CPS table creator. Retrieved December 19, 2006,
from http://www.census.gov/hhes/www/cpstc/cps table creator.html
Valentino, N. A., Hutchings, V. L., Banks, A. J., & Davis, A. K. (2009). Selective exposure in
the Internet Age: The Interaction between Anxiety and Information Utility. Political
Psychology, 30(4), 591613.
Zaller, J. R. (1992). The nature and origins of mass opinion. New York: Cambridge University
Press.
Zillmann, D., &Bryant, J. (1985). Selective exposure to communication. Hillsdale, NJ: Erlbaum.
Zook, M. (1996). The unorganized militia networks: Conspiracies, computers, and
community. Berkeley Planning Journal, 11(1), 2648.
Copyright of Journal of Communication is the property of Blackwell Publishing Limited and its content may
not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.

Garret 2009

Uploaded by

Copyright:

Available Formats

Garret 2009

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

Garret 2009

Uploaded by

Copyright:

Available Formats

Journal of Communication ISSN 0021-9916

ORI GI NAL ARTI CLE

.21 .26 .63

.49 .43 .12

n 908 907 902 906

No. ofine sources used .09

Campaign interest .40

Strong candidate support .05 .04 .02 .03

You might also like