Minerva (2011) 49:1–23

DOI 10.1007/s11024-011-9161-7

Public Value Mapping and Science Policy Evaluation

Barry Bozeman • Daniel Sarewitz

Published online: 9 February 2011

Ó Springer Science+Business Media B.V. 2011

Abstract Here we present the framework of a new approach to assessing the

capacity of research programs to achieve social goals. Research evaluation has
made great strides in addressing questions of scientific and economic impacts. It has
largely avoided, however, a more important challenge: assessing (prospectively or
retrospectively) the impacts of a given research endeavor on the non-scientific, non-
economic goals—what we here term ‘‘public values’’—that often are the core public
rationale for the endeavor. Research programs are typically justified in terms of their
capacity to achieve public values, and that articulation of public values is pervasive
in science policy-making. We outline the elements of a case-based approach to
‘‘public value mapping’’ of science policy, with a particular focus on developing
useful criteria and methods for assessing ‘‘public value failure,’’ with an intent to
provide an alternative to ‘‘market failure’’ thinking that has been so powerful in
science policy-making. So long as research evaluation avoids the problem of public
values, science policy decision makers will have little help from social science in
making choices among competing paths to desired social outcomes.

Keywords Public values
Research choice
Research evaluation

Science policy
Market failure

B. Bozeman (&)
Department of Public Administration and Policy, University of Georgia, 201 Baldwin Hall, Athens,
GA 30602-1615, USA
e-mail: bbozeman@uga.edu

D. Sarewitz
Consortium for Science, Policy and Outcomes, Arizona State University, PO Box 875603, Tempe,
AZ 85287, USA
e-mail: dsarewitz@asu.edu

123
2 B. Bozeman, D. Sarewitz

Introduction

Science policy did not become the focus of serious intellectual inquiry until the
early 1960s. Previously, debates raged in both England and the United States about
the extent to which science could be directed toward societal aims, but such
disagreements (exemplified in England by the 1930s debates between J.D. Bernal
and Michael Polanyi [e.g. see Freeman 1992], and, in the U.S. by the policy debates
that pitted Vannevar Bush against Harley Kilgore shortly after World War II [e.g.
Kevles 1995]) were little informed by robust theory and even less by data. In large
part, Minerva, first published in 1962, was created to help fill this intellectual and
research vacuum. ‘‘Improved understanding of the relations between government
and systematic and disciplined inquiry in science and scholarship was taken as the
subject matter of Minerva,’’ wrote its first editor, Edward Shils (1968, p. xiv). Shils’
explicit hope was that such understanding would ‘‘make scientific and academic
policy more reasonable and realistic’’ (p. xiv).
From the beginning, the central problem in science policy was recognized by
workers in the field as the problem of choice. In a world of finite resources, how
should policy-makers choose among the many competing scientific disciplines,
projects, and programs in making public investments? Minerva published three
early, seminal papers on this problem (Polanyi 1962; Weinberg 1963, and Toulmin
1964), and many others that contributed to setting the terms of the problem. As
Toulmin (1964) noted, the choice problem was both ‘‘difficult and inescapable.’’
Difficult not only because the problem itself was poorly specified, but also ‘‘because
we are sheerly ignorant about many of the relevant factors and relationships’’
between scientific advance and societal ‘‘repercussion’’ (p. 343).
Toulmin’s discussion of the choice problem remains particularly apt for two
reasons. First, he identified the two important poles that still represent the
organizing dichotomy of work in the subsequent four-plus decades: the ‘‘econo-
mist’s view, according to which science is basically deserving of support because it
is the handmaid of industrial growth; and a scientists view, representing technology
as a kind of scientific roulette in which those who plunge deepest tend to win the
biggest prize’’ (p. 348). Second, he recognized that the problem of choice was
significantly a ‘‘chalk-and-cheese’’ problem, where diverse activities categorized as
science (much as chalk and cheese might both be categorized as ‘‘crumbly white-ish
materials’’) were in fact constituted by a multitude of activities that were in many
ways incommensurable, so that, for example, ‘‘the choice between particle physics
and cancer research becomes a decision whether to allocate more funds (a) to the
patronage of the intellect or (b) to improving the nation’s health. This is not a
technical choice, but a political one’’ (p. 357).
Toulmin’s identification of science policy’s economic and scientific poles, and
his recognition of the chalk-and-cheese problem, help to explain why, despite three
decades’ progress in the ability to conceptualize, measure, and evaluate research
impacts, a gaping hole remains in research evaluation methods and technique: the
ability to evaluate the social and public value impacts of research. Indeed, such
impacts have been defined out of the problem as at once irrelevant (they are not

123
Public Value Mapping and Science Policy Evaluation 3

encompassed in the science-economy dimension) and inaccessible (they simply add

to the already intractable problem of incommensurability of choices).
Thus, professional researchers have developed powerful economic tools to
measure economic impacts of research, sophisticated bibliometric tools to measure
the impacts of research outputs on scientific fields and the course of science and
technology, and improved peer review techniques for assessing projects, programs
and proposals. But there has been remarkably little progress in the ability to measure
directly, systematically, and validly the impacts of research on social change. Many
scientists (e.g. Ziman 1968) have extolled the communal and cultural value of
scientific knowledge. However, without rejecting compelling arguments for the
intrinsic value of research in intellectual, cultural and aesthetic terms (Fischer
1997), most policy-makers and citizens seem to agree that the chief purpose for
public funding of research is to improve the quality of life (Johnson 1965). And
most scientists justify it as such.
As Toulmin (1964) and his colleagues well understood, the critical problem of
choice is not that the chalk-and-cheese problem is at heart a political one, but that
we have no satisfactory analytical tools for characterizing the social impacts of
chalk or of cheese—for understanding, that is, causal impact and magnitude of
effects of research activities on social change. This gap is not surprising when one
considers the difficulty of the task and the adolescent stage in the development of
research evaluation. Yet, part of the problem is self-imposed: if science policy
research in the past 40 years had focused as energetically on the problem of social
values and social impacts as it did on assessing scientific and economic impacts, we
might have made considerably more progress on resolving the problem of choice.
This counterfactual suggestion is not mere philosophizing, as we hope to
demonstrate by the case studies that follow this paper. Not only are the diverse
public values that are invariably deployed to justify scientific choices ascertainable,
so are the relations among such values, and, to an extent, the capacity to advance
them. With this in mind, and to provide a theoretical and methodological framework
for the subsequent case studies, in this paper we will consider: (1) why new
approaches and alternatives to research evaluation are needed and how they relate to
extant approaches; (2) special difficulties or challenges of developing such
approaches; and (3) a specific methodological framework that can be employed,
which we here term ‘‘public value mapping of science outcomes.’’ Importantly, the
work described in this paper and the cases that follow are supported by the U.S.
National Science Foundation program on the ‘Science of Science and Innovation
Policy,’ (http://www.nsf.gov/funding/pgm_summ.jsp?pims_id=501084) a program
that in turn responds to the observation by a former U.S. presidential science advisor
that ‘‘the nascent field of the social science of science policy needs to grow up, and
quickly’’ (Marburger, 2005).

Growing Up Quickly: Bringing Public Values to Science Policy Processes

U.S. science policy since World War II has to a large extent centered on three inter-
related clusters of values. The value cluster we label ‘‘Scientific Productivity’’

123
4 B. Bozeman, D. Sarewitz

includes concerns about the quality and quantity of U.S. research output, perceived
or measured world leadership in science, human resources issues, including not only
the ‘‘pipeline’’ but also the capacity of scientific fields, and, of course, funding
issues, usually framed as ‘‘why we need more money for science.’’ These values
occupy Toulmin’s ‘‘science’’ pole. The second value cluster, corresponding to
Toulmin’s economic pole, is a continual and pervasive concern with economic
productivity, which includes concerns about innovation, technological advance,
economic growth and, in some instances, an implication that economic benefits,
widely shared, will advance social goals and quality of life. A third value cluster,
‘‘Defense and National Security,’’ includes concerns with weapons superiority,
developing non-weapons technology to support the military, and generally, using
science and technology to enhance military strength expressed regionally and
globally.
Obviously there are other core values associated with publicly funded science,
and it is certainly the case that U.S. science policy continues to add new values and
attendant missions. The 1970s witnessed the emergence of values clusters pertaining
to energy and environment. The vast expansion in the 1980s of the National
Institutes of Health was rationalized largely on the basis of improved health and
well-being. Yet, for all four broad areas of research—military, energy, environment,
and health—the public values served by such priorities have been significantly
subsumed by the demand for scientific excellence in the pursuit of enhanced
economic productivity (e.g. Sarewitz 1996). (For example, the military rationale has
been largely subsumed by the core values of science [i.e. via the crucial role of the
U.S. Department of Defense in building academic science programs, e.g., see Leslie
1993] and economic growth [e.g. ‘‘dual use technology’’ and ‘‘spin-offs,’’ e.g. see
Ruttan 2006].) The result is that the breadth of values expressed in U.S. science
policy is significantly wider than the breadth of values directly pursued or assessed.
In particular, and as documented in each of our case studies, even where broader
public and social values are expressed in science policy development, they are often
subverted, reinterpreted, and subjugated to the science-economy axis. The result is a
winnowing of values brought to science policy and, overall, a decrement in public
values. Thus, our goal in this paper is both to discuss why public values are so easily
deflected in science policy, and also to suggest an approach to tracking public values
and monitoring and evaluating their influence on science policy.
Before discussing the reasons why we feel it is especially difficult to infuse and
maintain public values in science policy, we consider the meaning of ‘‘public
values.’’ We use this definition:
‘‘A society’s ‘‘public values’’ are those providing normative consensus about
(1) the rights, benefits, and prerogatives to which citizens should (and should
not) be entitled; (2) the obligations of citizens to society, the state and one
another; (3) and the principles on which governments and policies should be
based’’ (Bozeman 2007: 37).
Our focus on public values requires at present no more precise definition, but we
note these implications of the above definition: (1) public values are not static and
immutable, there is no ‘‘natural law’’ or ‘‘natural rights’’ meaning to our concept of

123
Public Value Mapping and Science Policy Evaluation 5

public value; (2) economic values may in some instances qualify as public values;
(3) in some instances and for some policy controversies, there may be insufficient
consensus to identify public values (however, this is rare); (4) public values may
conflict (e.g. liberty and security, privacy and transparency); (5) public values may
or may not be interdependent.
We aim to increase the public values component of science policy, essentially
expanding science policies’ dominant value sets, by making it possible to consider
diverse values using methods and criteria comparable to those already widely
accepted and used for scientific and economic values. This goal immediately raises
a practical concern: Where can public values be found? A nation’s more
fundamental laws and, if there is one, its constitution, provide good starting points
for identifying public values, though public laws and public policies are best viewed
as reflecting and expressing public values rather than establishing them. If policies
do not necessarily tap the roots of public values, they often can be taken as
surrogates. For example, laws or policies may be justified (e.g. in legislative
language or agency strategic plans) on the basis of a public value such as
‘‘improvements in public health and longevity,’’ or ‘‘decreased infant deaths,’’ or
‘‘cleaner air.’’ One can expect disagreements on the need for and desirability of
additional increments in any of these values, but few would find the public values
themselves to be objectionable.
Another obvious and time-tested approach to tapping citizens’ values is by public
opinion survey.1 There is abundant information about such issues as trust in
government, division of responsibility between federal and state governments,
political ideology, and responsiveness of government (see Nye 1997). Careful study
of general views and values of citizens—studies performed apart from any specific
decision—may be useful for making specific choices, such as whether government
or the private sector should manage the state prison. This suggestion is not unlike
the widespread practice, used especially in local governance, and, recently, in
conjunction with new internet and telecommunications technology (e.g. Budd and
Connaway 1997), of citizen polling (e.g. van Houten and Hatry 1987). It is simply
polling citizens about their most fundamental values.
Third, some recent studies (Bozeman and Sarewitz 2005; Feeney and Bozeman
2007) seeking to apply public values criteria have employed a posited model
(Bozeman 2001), one that has the value of being explicit and of providing criteria
not dissimilar to those from ubiquitously applied market failure models. We discuss
the posited model below. Finally, we note that public policy statements can in many
instances be taken as de facto public values and that a valuable activity of policy
analysts is to track the evolution of those values as policies evolve from ideas
premised on diverse and deep values to practical, front-line policies that may be far
removed from the values summoned initially to articulate or defend policy ideas. A
recent study has applied public value mapping in just this manner to the field of
nanoscale science and engineering (Fisher et al. 2010).
1
A distinction should be made between public opinion and public values: Whereas public opinion is
highly volatile, both in its concerns and its directions, public values are much more stable. New public
values may enter and old ones may exit but generally only after great social change and the passing of
generations.

123
6 B. Bozeman, D. Sarewitz

In short, the answer to the question ‘‘Where can public values be found?’’ is that
they are located in a great many places: formal scholarly literature, cultural artifacts
and traditions, government documents, even some opinion polls (ones receiving
valid and representative responses to questions about core values). Jorgensen and
Bozeman (2007) sought to develop an inventory of public values from such sources,
as well as from the public documents and literature on public values, public interest
and governance. Civil societies are necessarily permeated by public values since it
is these that provide much of the structure of civil societies to begin with. And,
crucially, specific public values are selected to justify science policy and other
government actions. A greater problem than identifying public values is
understanding them in some analytically useful form.

The Need for Public Values in Science Policy

Many major science policy initiatives are premised on values one might take to be
‘‘public values.’’ It is not difficult to find values statements supporting research for
environmental protection or health, for example. Why do we feel there is an under-
emphasis in public values in science policy? In the first place, public values are
more likely to contain as their content the end state outcomes ultimately important
to most people. For example, few care about economic growth per se. Economic
growth often is taken as a surrogate for well being or even happiness, but in fact
economic growth is by most accounts an instrumental value, a way of achieving
broader public values such as family health, leisure, safety, clean air, education and
job attainment and career satisfaction. But why begin with surrogate values? Is it not
more sensible to premise policies on the outcomes they should achieve rather than
the instruments presumed (perhaps erroneously) to enable those outcomes?
Second, public science is supported by tax dollars, under a tax system that is
designed (however inadequately) to be progressive and promote equity. One reason
to infuse public values in science policy is that they are by definition broader values
and by implication ones more likely to affect all or most citizens. Yet, it is by no
means clear that the dominant values of scientific excellence and economic
productivity are sufficient to account for the broad range of values that the public
hopes to gain from science. Most obviously, even the idea of the linear model of
science policy, science leading to technology leading to goods and services leading
to economic growth, has been thoroughly repudiated by economists of innovation
(e.g. Rosenberg 1982). But even if it were true, the idea that all will benefit from the
economic growth ends of science and technology, even though widely asserted, has
little plausibility (e.g. Woodhouse and Sarewitz 2007).
Third, one must be vigilant about public values in science policy because they are
so easily subverted. This point is subtler than the previous two. We can say that
science policy values, and indeed all values expressed in all major policies, are
dynamic, and that they evolve in stages, albeit not always in a straightforward
fashion and not always sequentially. In most instances, stages include (1) agenda-
setting, (2) policy design(s), (3) policy choice, (4) policy implementation and
(usually but not always), (5) policy assessment or even systematic evaluation.

123
Public Value Mapping and Science Policy Evaluation 7

Values are important at every stage, but they are volatile. In some cases values
change as a result of learning, in some cases they atrophy for lack of advocacy, and
in still others they fall under the weight of new values infused by self-interested
parties (i.e. politics).
Different types of values are privileged at different stages (Baumgartner and
Jones 1991). In particular, broad public and social values fare well at the very outset
when policy rhetoric and ideals are articulated and advocates seek support for
policies. At that stage, before policy-makers and other parties have settled to the
business of making difficult choices and having to mount rationales or even
evidence for those choices, it is easy enough to speak broadly about public values.
Once the dust settles, and policy options are winnowed, public values are often
shunted to the background as advocates and disputants begin to negotiate, usually
on a narrower basis. The public values remain as justifications for policies,
sometimes even tacked on as rhetorical cover to sub-optimal or patched together
policies that accommodate a great many conflicting values, including many private
ones or ones advanced by narrow coalitions.
But public values usually are not advanced during choice processes, for three
related reasons. First, science values and economic values are available as accepted
and dominant surrogates for all other values. Second, public values are supported by
no coherent set of conceptual tools to aid in choice. Third, and in contrast, many
such tools are available for science and economic values. In particular, the market
failure model is easily available, widely known, anchored in theory assumptions
consonant with much of U.S. policy-making, and, thus, it often plays a role in
framing choices and in the policy choices themselves. The market failure model also
directly links the core science value of knowledge creation to the economic
rationale via the discredited yet ever-present linear model. Availability is everything
and there is no corresponding model of public values or public interest to compete
with market failure and similar decision models based loosely on microeconomics.
After their initial use as rationales or rhetorical devices, public values tend to stay
at the rear throughout the remainder of the R&D policy process until such a time as
public officials or other interested parties began to question whether the policy has
had desired results. In some such instances public values make a return appearance,
but usually not for long. The reason they are again set to the side is that one quickly
finds that both the analytical tools at evaluators’ disposal usually have little to do
with public values—but mesh quite well with Economic Productivity or Scientific
Productivity values sets. One sees this process at work in each of the case studies
that follow this introductory paper. For example, the application of nanotechnology
to cancer has been justified for its potential to contribute to equitable health
outcomes, but in the end is assessed in terms of the broader economic goals of
nanotechnology innovation (Slade, this issue); research on hurricane tracking is
supported for its capacity to improve preparedness but continues to displace other,
perhaps more vital, lines of research on the basis of claims of scientific excellence
and opportunity (Maricle, this issue).
To reiterate our most fundamental point, while there are many reasons to expect
that public values will often be displaced in science policy, there are two key
problems that can be addressed and remediated. First, the lack of adequate

123
8 B. Bozeman, D. Sarewitz

conceptual apparatuses to compete against market failure and other economics-

based models means that advocates for public values have limited analytical
support. We have begun to address this issue elsewhere (Bozeman 2001; Bozeman
and Sarewitz 2005) by providing a ‘‘public values failure’’ model that is a rough
equivalent of the market failure model, and which is adopted in all the following
case studies. Second, approaches to evaluating science and technology outcomes
have been dominated by techniques and methods anchored either in microeconom-
ics (e.g. cost benefit analysis), supporting the economic productivity value set, or
bibliometrics (e.g. citation analysis, co-citation networks) supporting the academic
productivity value set. As we will show, however, a competing approach rooted in
public value assessment is not merely practicable, but revealing.
Bearing in mind this relationship between public values infusion in science
policy and the difficulty of bringing public values to bear in evaluation, we consider
in the next section the evolution of research evaluation and, particularly, reasons
why public values have been conspicuously absent in both formal evaluations and
indicators-based assessments. We focus particularly on the economics origins of
research evaluation because most research evaluation remains centered on
economics and because bibliometric approaches tend to be less general in their
purview and less often in competition with public values concerns.

Economics Bases of Research Evaluation

For present purposes, we mean by ‘‘research evaluation’’ any systematic, data based
(including qualitative data) analysis that seeks as its objective to determine or to
forecast the social or economic impacts of research and attendant technical activity.2
(For a more detailed but similar definition, see Luukkonen-Gronow 2007). In post
hoc research evaluation, the focus is generally on some set of discrete scientific or
technological outputs such as publications, patents or some other expression of
intellectual property. Importantly, formal research evaluation always involves some
such discrete commodity, either singly or aggregated. Rarely does it begin explicitly
with the goals, objectives, or values of the program and then trace back to various
outputs and impacts.
Formal research evaluation is a recent invention and its origins tell us much
about why it remains dominated by economic analysis and economic values, with
public values having made little headway. As late as the 1980s, research
evaluation was a field with few practitioners, mostly focused on economic
evaluation of industrial firms’ internal rate of return.3 Whereas the Canadian

2
By ‘‘research assessment,’’ not our focus in this paper, we mean an investigation with similar objectives
but not necessarily including data and perhaps premised on indicators but with no formal analysis.
3
During the history of modern science and technology policy and research evaluation, the most
prominent approach to assessment has been peer review. While recognizing that peer review is crucially
important, the present study focuses on systematic and potentially quantitative or mixed-method
approaches and, thus, does not discuss peer review approaches to research evaluation. Similarly, this
paper does not deal with the many and increasingly useful bibliometic approaches to research evaluation.

123
Public Value Mapping and Science Policy Evaluation 9

government4 and some European nations5 had begun systematic evaluation of

publicly funded research, in the United States and many other nations, evaluation
of public research impacts was not a field at all, but rather an agglomeration of
fragmented, largely isolated works, many unpublished.
To understand the roots of research evaluation one can consider the state-of-the-
art as reported in one the earliest reviews focusing specifically on studies of the
evaluation of publicly funded research. Salasin, Hattery and Ramsay’s The
Evaluation of Federal Research Programs (1980) stated intention was to ‘‘identify
useful approaches for evaluating R&D programs conducted and sponsored by the
federal government’’ (p. 1) and in pursuit of that objective they interviewed more
than two hundred experts in evaluation generally or research evaluation specifically,
most of them based in industry. The resulting monograph cited 49 papers, including
exactly one journal article (Rubenstein 1976) and one book (Andrews 1979)
focusing explicitly on systematic evaluation of government-sponsored research. The
monograph identified four problems endemic to evaluating government research
impacts, including (1) lack of a straightforward definition of effectiveness; (2)
multiple and competing objectives; (3) problems in aggregating products and
results, especially across programs, and (4) reconciling political and scientific
measures of success—a list that would work just as well today.
Since then, studies and methods of R&D evaluation have greatly proliferated (for
an overview of approaches and methods for research evaluation, see Bozeman and
Melkers 1993; OECD 1997 and OECD, in press). But most of the problems
identified nearly three decades ago in the Salasin, Hattery and Ramsay’s pioneering
monograph still exist, particularly the problems associated with a focus on discrete
R&D outputs. This is especially inimical to public-values-based research evalua-
tions inasmuch as discrete outputs in most cases can hardly begin to provide an
adequate gauge for the social change sought from research programs.

Economics Bases for Research Impacts Evaluation

Economic assessments of research and technology generally fall into two basic
categories, one of which is most relevant to practical research evaluation. Less
relevant to practical evaluation, but influential to broad science policy decision
making and rationalization—not to mention the core rhetoric of national politics—
are aggregate-level production function analyses (e.g. Solow 1957), typically
focusing on the contribution of technology to national or regional economic growth.
More useful for research evaluation are those economic studies seeking social rates
of return (e.g. Jones and Williams 1998), that is, approaches that use indicators of
marginal economic benefit as a surrogate for estimating the social utility of research
and technology. The implication is that wealth can be used to obtain socially

4
For a history of government mandated research evaluation in Canada, including research evaluation,
see Auditor General (1993). For a history of research evaluation activities in Canada, see Barbarie (1993).
5
Several publications provide synoptic reviews of the history and methods of research evaluation in
European nations; see, for example, Luukkonen (2002); Callon, Laredo and Mustar (1997).

123
10 B. Bozeman, D. Sarewitz

desirable outcomes and, thus, increments in wealth can be taken as indicators of

social benefit. Among social rate of return approaches, benefit-cost analysis has
been most common and most prominent in project and program-level evaluations of
research (see, for example, Link 1996a; 1996b; Ruegg 1996; Audretsch et al. 2002).
Aside from the possibility that one may not wish to assume that economic
benefits, measured in monetary terms, fully express or stand in for public values,
there is also the issue that very few such studies even begin to consider equity issues
in the distribution of benefits. This is not because the evaluators do not recognize the
distributional issues in play in science and technology outcomes, but rather because
the methods and techniques employed cannot in most instances accommodate
distributional variables (Martens 2009).
Because economics approaches to research evaluation focus on discrete techno-
logical outputs such as patents and articles, they are useful for those who wish to
aggregate outputs and consider them in connection with, for example, the
performance of technology transfer programs and regional commercialization efforts.
The utility of these approaches should be obvious even to skeptics. While the benefits
of economics-based approaches to evaluation are explored in more detail elsewhere
(Link 1996b), we can for present purposes summarize them as follows:
1. Evaluation rooted in neoclassical economics seems to hold forth promise of
‘‘harder’’ more rigorous analysis and, thus, matches well the policy-maker’s
need for metrics to justify expenditures. Typically, these approaches yield
numerical assessments of such factors as increments in patents or job creation
or firm partnerships (Kostoff 2001).
2. Whereas most approaches to research evaluation are either atheoretical and
exclusively tool-oriented, or based on poorly developed theory, economics
approaches can draw from decades of development of relatively strong (for the
social sciences) theories of the firm, rational choice and economic growth.
3. While economists recognize that there are values that cannot be well accounted
for by monetized units, many have been quite creative in developing quasi-
economic techniques based on preference functions and units that mimic rational
economic choice (e.g. contingent value analysis [Cummings and Taylor 1999]).
4. Economic development and growth is a driving impetus for policy and politics
throughout the world, and as we have emphasized, science and technology
policies are strongly rationalized in terms of pursuit of economic growth. It is
not surprising that economics-based approaches to research evaluation underpin
economics-rationalized science and technology policy.
Despite their many advantages, economics-based approaches to research
evaluation have many limitations, especially if one is interested in gauging the
impact of research on public values and social change.
1. As already mentioned, most economic approaches to research evaluation focus
on the discrete products of research. While this is methodologically sensible, in
that it promotes measurement, it also promotes a narrowness of view. For
example, if one is interested in the long-range capability to produce innovation,
then simply counting the results of discrete products may not provide a good

123
Public Value Mapping and Science Policy Evaluation 11

insight into the health and viability of scientific fields or a nation’s innovation
systems. If one is interested in the capacity to produce innovation, rather than
just the innovation products themselves, then a focus on ‘‘scientific and
technical human capital’’—the integrated social networks and aggregate skills
of scientists (Bozeman et al. 2001)—and other, non-economic, approaches to
evaluation are required. As well, a focus on particular products and projects
works best when there are crisp boundaries (e.g. a single R&D project), while
most social objectives do not have easily discernible boundaries and are
influenced by myriad causal factors. An approach focused on assessing the
capacity to achieve non-economic public values requires methods permitting
soft boundaries.
2. Despite efforts to consider implications of future streams of benefit, economics-
based evaluations tend to be static. They rarely take into consideration the
mutability of the ‘‘products’’ evaluated, much less the changes in the persons
and institutions producing them. Thus, an economic analysis of the impacts of a
mechanical heart valve innovation would have great difficulty taking into
account broad secondary effects such as the implications for a longer-lived
population or for equity of health care access, and would also have difficulty
tracing the differential impacts of successive generations of the technology.
3. Product-oriented and output-focused evaluations tend to give short shrift to the
generation of capacity in science and technology, and to the ability to produce
sustained knowledge and innovations.
4. Most important and obvious for present purposes, there are just some things that
money can’t buy: many social benefits and costs of research are not well or even
validly accounted for in monetary units. For example, while economics does a
good job of precisely measuring the value of a human life, the question of whether
such measures as life-time earnings capabilities are also accurate indicators is
utterly laden with values that are non-economic. Indeed, as the subsequent case
studies document, research is generally justified on non-monetary values and,
thus, the evaluation of research in purely monetary terms amounts to a sort of bait-
and-switch, where public policy intent becomes transformed by subjecting it to
the available theories and evaluation methods, as if one went to a doctor for a
health examination and ended up with an assessment of one’s earning potential.
It is this latter limitation, the inadequacy of economics-based approaches for
measuring and providing understanding about the social impacts of research, that is
our chief concern here. To be sure, economics approaches are not unique in their
inadequacy for this task. Currently, no satisfactory method (except, perhaps, case
studies that are very context specific and rarely generalizable) has been developed to
validly assess the impacts of research on social change.

Social Impacts of Research: Challenges to Theory and Method

A methodological problem in all approaches to research evaluation is that research

is often only one factor in determining social outcomes and is rarely the most
important one. The science advisor to President Obama has identified a series of

123
12 B. Bozeman, D. Sarewitz

‘‘challenges’’ for U.S. science and technology that include ‘‘better [health care]
outcomes for all at lower cost,’’ ‘‘poverty eradication,’’ ‘‘transforming the global
energy system,’’ and ‘‘reducing risks from biological and nuclear weapons,’’
(Holdren 2009) but of course when research plays any significant role in achieving
such desirable social outcomes it is in concert with a great many other social,
economic and natural determinants. The outcomes, that is, are highly-overdeter-
mined. In such circumstances, it is virtually impossible to parse out the contribution
of research; this is what Toulmin (1964 p. 343) meant when he observed that ‘‘we
are sheerly ignorant about many of the relevant factors and relationships’’
connecting science to outcomes. Our ability to trace these links is not much better
now than it was in 1964. Whether one employs standard economics-based
approaches such as cost-benefit analysis, social indicators monitoring and social
accounting, or even in-depth case studies, causal attribution for complex social
impacts is always fraught with great difficulty.
A related problem pertains to the ‘‘dependent variables.’’ Determining causation is
difficult enough, but often the effects are themselves interwoven in ways that are
difficult to understand or unravel. Social outcomes occur in clusters. For example, in
the case of automobile safety, research has shown that safety innovations such as disk
brakes or even seat belt laws can actually result in more accidents as drivers’ behavior
becomes more risky as a result of technologies providing an increased sense of
security (e.g. Adams 2006). Similarly, many of the social and public health gains that
have been realized by smoking cessation programs are offset by the fact that
reductions of smoking have contributed in some degree to the increase in obesity rates.
Our case studies demonstrate these complex and contradictory effects: technology
transfer programs may lead both to increased wealth and to greater inequities
(Valdivia, this issue); advances in nanotechnology-based cancer treatments appear
likely to increase health inequities that are already significant in the U.S. (Slade, this
issue). In short, in modeling social outcomes from research one has difficulty not only
tracing cause to effect, but also setting boundaries on effects. This is one of the reasons
why we have adopted an open, ‘‘mapping’’ approach to evaluating public values. We
are not seeking a deterministic model, but rather an approach that can enhance insight,
debate, and expectations—and thus improve decision outcomes.
A related complication to developing public values theory in general and public
values in science policy in particular is that not all values are public values, and
means of demarcating values are hardly clear-cut. Consider this general definition of
‘‘value’’: ‘‘A value is a complex and broad-based assessment of an object or set of
objects (where the objects may be concrete, psychological, socially constructed, or
a combination of all three) characterized by both cognitive and emotive elements,
arrived at after some deliberation, and, because a value is part of the individual’s
definition of self, it is not easily changed and it has the potential to elicit action’’
(adapted from Gaus 1990). Given this not unfamiliar description of value and the
previous definition of ‘‘public value,’’ it is perhaps apparent that the distance
traveled from one to the next is considerable. From the standpoint of empirical
social science, the fact that values held by individuals are not agent-neutral provides
limits in values analysis. However, if the role of social science is limited with
respect to such private values, it is virtually unbounded (though poorly developed)

123
Public Value Mapping and Science Policy Evaluation 13

with respect to public values, because public values are typically instrumental, or
employed instrumentally.
In seeking public values and their application, public value theory (Bozeman
2002; 2007) embraces empirical social science. We begin with the assumption that
all instrumental values, public, economic or private, can be viewed as causal
hypotheses that are, in principle, subject to empirical tests. From here it becomes
possible to seek and even test public values statements found in broadly held
articulations of desirable states toward which progress can be assessed (‘‘decreased
infant deaths’’; ‘‘cleaner air’’). We are not after prediction or proof, we are after
plausibility, which seems to us a desirable, reasonable and achievable expectation
for science policy-making (or any policy-making, for that matter). Meyer (this
issue) shows, for example, that the internal logic of the climate science policy
process in the U.S. is completely incoherent, and for that reason alone can have little
capacity to achieve the goals that justify and motivate the program. What makes this
analysis possible is the recognition of the public values embedded in the process,
and the logical relations (or lack thereof) among these public values.

Getting on With It: A Sober and Humble Rationale for Evaluation of Social
Impacts of Research

The foregoing section identifies formidable obstacles to assessing the social impacts
of research and, unfortunately, the list above is not exhaustive (for a more detailed
discussion of problems in tracing social and public value impacts, see Bozeman
2007). Yet, whether or not fully adequate analytical tools are available, policy-
makers will continue to make choices about research funding. These choices will
continue to be premised on a causal logic. As we have discussed, and as the cases
document in greater detail, in making decisions about investments in research,
policy-makers make assumptions about the effects of those investments on such
social outcomes as public health, transportation systems, education, and wealth
creation. In most instances those choices will, perforce, be based on limited
information provided by interested parties. Any evidence that can be brought to bear
on those choices, even when fraught with known methodological limitations, is
likely an improvement over intuition, habit, rough-hewn ideology, political self-
interest, powerful myths about how the world works, and other such biases that so
typically guide investments in research aimed at solving social problems. If nothing
else, new approaches can contribute to: a) disciplined discussion, healthy skepticism
and reflection and b) openness to other, clearer, non-scientific options. It is in that
spirit that we began to fashion the approach we refer to as ‘‘public value mapping.’’

Public Value Mapping and Its Lineage

Put simply, public value mapping is an approach to identifying the public value
premises of public policy and then tracking their evolution and impacts on policies
and, ultimately, social outcomes. The primary rationales for the public value

123
14 B. Bozeman, D. Sarewitz

mapping of science (PVM) are that (1) the focus of science policy should be on
social goals and public values, and (2) current research evaluation and science
policy analysis methods and techniques are not sufficient for analysis of the impacts
of research on public values and fundamental social goals.
From the outset, public value mapping, an approach aspiring to practical
application, has been rooted in public value theory. While we do not have space or
great need to go into the details of public value theory, it is perhaps useful to
provide a modest introduction.
A PVM taproot is new theoretical thinking about the value of knowledge and its
assessment (see Bozeman 2007 for a summary). To a large extent, recent work on
the value of knowledge is a response to limitations of economic theory in
understanding knowledge value (e.g. Bozeman and Rogers 2002). Economists have
never made much headway valuing scientific knowledge (see Machlup 1962).
Scientific knowledge, in economic terms, is generally considered a pure public
good, and thus, an example of pricing inefficiency. In the world of public finance
economics, theory loses its power in instances where markets do not work in
straightforward fashion and where efficient pricing is impossible. However, it is
generally in these realms that public values and, for that matter, governments and
policies operate. Thus, stretching economic theory to the breaking point, rather than
developing theories of public and social value, seems a poor route forward.
Philosopher Elizabeth Anderson (1993) presents an especially interesting analysis
of economic value and value theory as it pertains to economics. Anderson’s position,
one that would perhaps seem radical to many social scientists, is that economic values
are inherently monistic. Because of the fundamental structure of assumptions built
into economic values, they cannot accommodate pluralistic approaches to values. To
put it another way, an analysis valuing exchanges, commodities, and services on the
basis of market standards pre-empts simultaneous, comparable reference to other
standards (see Marmolo 1999; Anderson 1993). These assertions have direct
implications for models of innovation and the impacts of scientific and technical
knowledge.
In addition to philosophy of values, public value theory draws from the field of
public administration (e.g. Van Wart 1998; Jorgensen, 1996; Van Deth and
Scarbrough 1995; Kirlin 1996; Bozeman 2007). In public administration, a useful
theory is a theory in practice (Adams, 1992). Perhaps for that reason, much theory that
underpins analysis of public values is anchored in various aspects of pragmatism (e.g.
Shields 1996; Garrison 2000; Bozeman 2007) and especially developing communi-
tarian and procedural approaches suitable for the identification of and support of
public values. Public administration literature (e.g. Van Wart 1998; Jorgensen 1996;
Van Deth and Scarbrough 1995; Kirlin 1996) has begun to move from philosophical
discussion of the public interest to a concern with identifying aspects of publicness or
public values. Case studies (e.g. Jorgensen and Bozeman 2002) focus on how public
values are infused (or not) in public decisions. To a large extent, the cases following
this paper have that intent: to demonstrate and assess the extent of public values in
public policies and to trace their roles and impacts.
Public value mapping can best be thought of as an analytical confederation. It is
not a unified method nor does it aspire to closure. Indeed, it is not a method, per se.

123
Public Value Mapping and Science Policy Evaluation 15

It is better viewed as a loose set of heuristics for developing analyses of public

values. Public value mapping begins with a set of core assumptions but these are not
inviolable. Table 1 (Bozeman 2003) provides these core assumptions. The cases
following this paper, diverse as they are, strive to conform to these assumptions.

Table 1 Core Assumptions of Public Value Mapping

1. PVM is either prospective (analyzing planned or projected research activities), ‘‘formative’’
(analyzing such activities as they are occurring), or ‘‘summative’’ (evaluating activities and their
impacts after they have occurred).
2. It seeks to take into account the highest order impacts of activities (i.e. broad social aggregates) and,
thus, focuses on social indices and social indicators.
3. It is multi-level in its analysis, seeking to show linkages among particular program activities of an
agency or institution, activities of other agencies or institutions, relationships- either intended or
not- among various institutional actors and their activities. Related,
4. PVM is concerned with understanding the environmental context for research and related
programmatic activities, locating the activities and their institutional actors in terms of other actors
in the environment, the constraints, opportunities and resources presented in the environment.
5. Research in any field by any method is embedded in a social context; in PVM analysis of the social
context of the research (i.e. characteristics of research performers, their attributes and social
relations) is a part of the analysis.
6. PVM is guided by a ‘‘public value model of science outcomes’’ rather than a market-based or market
failure model. PVM explicitly rejects evaluation and assessment based on commodification of
research values and outcomes. Market prices are viewed as weak partial indicators of the social
value of research and research outcomes. Even as a partial indicator, market value is considered in
terms of not only magnitude but also distribution and equity criteria.
7. Since market value is eschewed in PVM and since social values are not interpersonally transmissible,
PVM anchors its outcomes values in a wide range of criteria derived from diverse sources
including: [1] official, legitimated statements of policy goals; [2] goals implicit in poorly
articulated policy statements; [3] government agencies’ goal statements in strategic plans; [4]
aggregated statements of value represented in opinion polls; [5] official policy statements by
government actors; [6] official policy statements by relevant NGOs.
8. PVM analyzes (maps) the causal logic relating goals statements (any of the above) to science and
research activities, impacts and outcomes, both measured and hypothesized. When possible, this
analysis begins with the causal logic articulated by responsible officials. The causal logics, explicit
or implicit, that are the basis of science and research activities are then considered in relation to
various plausible alternative hypotheses and alternative causal logics invented by the analyst.
9. PVM is not an analytical technique or even a set of analytical techniques, but a model that includes a
guiding theoretical framework (public value theory), a set of assumptions and procedures.
Research techniques employed in PVM depend upon the needs and possibilities afforded by the
context of its application. The only technical approach used in all applications of PVM is the case
study method.
10. After gathering data to test hypotheses about causal logics and outcomes, appropriate analysis
(selected depending upon specific analytical techniques used), is employed to test hypotheses and,
at the same time, measure impacts and outcomes. Results of analysis focus on interrelationships
among the causal logic, the environmental context and measured impacts and outcomes.
11. PVM concludes with a linkage of impact and outcome measures back to aggregate social indicators
or other appropriately broad-based, trans-institutional, trans-research program measures of social
well being.
12. PVM concludes with analysis and recommendations focusing on possible changes (in research or
program activity, causal logic, implementation) that seem likely to lead to improved social
outcomes.

123
16 B. Bozeman, D. Sarewitz

Public Value Mapping Criteria Model

Just as important as the core assumptions of PVM are the analytical heuristics it
brings to bear. Among these, we have found the public value mapping criteria model
is useful for structuring analysis and assessment. These PVM criteria begin with a
set of general criteria developed for judging public values failure (Bozeman 2002;
Bozeman and Sarewitz 2005), but there is no claim that the criteria are canonical or
exhaustive. Indeed, as presented in several of the cases that follow, the PVM
approach encourages the further articulation of the criteria developed previously as
well as the stipulation and justification of new criteria.
Public value criteria serve as heuristics for deliberation. Discussion and
argumentation about public values and their measurement proves less troubling in
those instances when there is a clear starting point, when one has at his or her
disposal public value criteria. Even when debates rage about choices of public
value, concepts of public value, and the relevance of public values to particular
states-of-affairs, one has hope of making headway if there are recognized public
value criteria structuring arguments. Perhaps more important, the lack of public
values criteria explains in part why economic frameworks such as market failure
have often held sway, even in cases where they seem poorly adapted to the problem
at hand.
Initial criteria for judging public values failure emerged in large measure as a
conceptual parallel to traditional market failure criteria. While these public value
criteria were set as companions to market failure criteria, initially they were posited
rather than derived empirically. However, the criteria were subsequently submitted
to test in various case studies, including cases pertaining to genetic modification
(Bozeman 2002; 2007), public health issues in influenza vaccine (Feeney and
Bozeman 2007), and nano-scale science (Fisher et al. 2010), among others.
Public value failure is not a conceptual alternative to market failure. Rather,
public values failure occurs when neither the market nor public sector provides
goods and services required to achieve public values. This implies that public
values can be realized (or can fail) under any set of market conditions. The chief
point of PVM criteria is to expand the discussion of public policy and management
by assuming that government (and market organizations as well) need be more than
a means of ensuring market successes and technical efficiency in pricing structures.
A fundamental assumption of the PVM model is that, contrary to current political
dogma and academic thinking, market failure actually tells us little about whether
government should ‘‘intervene.’’ With PVM, the key policy question becomes:
‘‘Whether or not the market is efficient is there nonetheless a failure to provide an
essential public value?’’ The PVM criteria model provides multiple lenses for
viewing this question. It is not a precise decision making tool (a la benefit-cost
analysis), but a framework to (1) promote deliberation about public value (and its
relation to economic value) and (2) provide guideposts for analysis and evaluation,
within the context of public value mapping.
The PVM criteria themselves (Table 2, adapted from Bozeman 2007 and Bozeman
and Sarewitz 2005) are not, then, actual public values but, rather, a set of diagnos-
tics applicable to questions of science policy (see Bozeman and Sarewitz 2005)

123
 Table 2 Public failure and public policy: a general diagnostic model
Public Failure Criterion Failure Definition Science Policy Example

Mechanisms for Political processes and social cohesion insufficient to ensure Peer review, the favored means of making decisions of
values articulation and effective communication and processing of public values individual-level projects, is appropriated for decisions about
aggregation huge scientific programs, resulting in the displacement of
social goals for more easily resolved technical goals
Imperfect monopolies Private provision of goods and services permitted even though When public authorities abrogate their responsibility for
Government monopoly deemed in the public interest overseeing public safety in clinical trials for medical
research, there is potential for violation of public trust and
public value
Scarcity of providers Despite the recognition of a public value and agreement on the The premature privatization of the Landsat program shows that
public provision of goods and services, they are not provided a scarcity of providers can create a public failure potentially
because of the unavailability of providers remediable by government action
Public Value Mapping and Science Policy Evaluation

Short time horizon A short-term time horizon is employed when a longer term Policy for energy R&D, by considering the short term, fails to
view shows that a set of actions is counter to public value fully capture the costs of global climate change on future
generations
Substitutability vs. Policies focus on either substitutability or indemnification even No-net-loss’ policies fail to take into account the
conservation of in cases when there is no satisfactory substitute nonsubstitutability of many natural organisms ranging from
wetlands protection to prohibiting the sale of human organs
resources on the open market
Benefit hoarding Public commodities and services have been captured by A prime technical success of genetic engineering, the
individuals or groups, limiting distribution to the population ‘terminator gene,’ proves an excellent means of enhancing
the efficiency of agricultural markets, potentially to the
detriment of millions of subsistence farmers throughout the
world
17

123
18 B. Bozeman, D. Sarewitz

and research evaluation. Since science policy, as is the case with nearly all legitimate
public policies, seeks ultimately to produce positive social change, it is subject
to many of the same public and social values as other policies and, thus, the same
values criteria prove useful. The case studies following this paper each employ the
PVM criteria model in their analysis and different aspects are relevant in different
cases.

Developing and Applying PVM

While considerable conceptual work has already been undertaken to provide

building blocks for public value mapping, efforts thus far have been incremental.
The goal of this paper and accompanying cases is to begin to create a viable
approach that can generate practical analytical tools and insights about public values
in science policy. It is also to inspire others working in the field of science policy
research to critique, adopt, transform, and apply public value mapping to what
remains the central unresolved challenge of science policy practice: the develop-
ment of evaluative methods that can help us make progress—theoretical,
methodological, and most important, practical—on the problem of choice.
All PVM approaches, included those presented here, begin as case study analyses.
Each of the cases strives to implement the following four steps (though they vary in the
extent to which each is emphasized). At this point in the development of the model we
consider these steps to be the procedural core of PVM:
1. A search for ‘‘public values’’ pertaining to the case: We have discussed several
approaches to identifying public values, including (a) surrogate public values
(government mission statements, strategic plans, and broad policies, statutes);
(b) distillation of public values from relevant academic literatures; (c) public
values as expressed in public opinion polls and public statements.
2. Application of the PVM criteria: Each case study examines the course of public
values in light of the criteria presented in the PVM criteria model, using these
criteria as a means of assessing possible failures in achieving public values. Not
only do analysts apply useful and appropriate PVM criteria, but they are
encouraged to identify criteria that do not fit their case and to begin to develop
new PVM criteria that may be useful for their case and may be generalizable to
others. There is an explicit expectation that the PVM criteria model will expand
and refine its criteria, not to such an extent that highly idiosyncratic criteria will
be included but to include criteria that have some potential for use in multiple
cases and analyses.
3. Developing value analysis chains: Among the many reasons why public value
analysis of science policy has made little headway is that values analysis itself
is remarkably underdeveloped. One of the difficulties of values analysis (Gaus
1990) is that analysts sometimes fail to consider interrelationships among
values, including such features as values hierarchies, conditional relations
among values, logical structures of multiple and related values, and ends-means
relations (Braybrooke and Lindblom 1963). One of the key objectives of the

123
Public Value Mapping and Science Policy Evaluation 19

public value mapping is thus to develop the ability to clarify relationships

among values, and assess how those relations influence the links between the
conduct of research and the pursuit of particular desired outcomes.
4. Graphically display the relations between market failure/success and public
values failure/success, using the public values failure grid (Bozeman 2002;
Bozeman and Sarewitz 2005). The grid provides a qualitative, synoptic view of
the results of the public value mapping, and thus helps with both communi-
cation and comparison. (For reasons of space, we do not illustrate the grid here;
it can be seen in each of the case studies.)
In addition to adopting these four steps to ensure some intercomparability of both
method and result, the cases also (1) perform the traditional case study role of ‘‘thick
description’’; (2) provide a context for the application of a variety of analytical
approaches, including logic models and value chain analysis; (3) help determine the
extent to which it is possible to distill public values in a satisfactory manner; (4)
extend the theories upon which PVM is premised; (5) point the way for further
development of analytical tools.
Finally, each case implicitly adopts one or more analytical lenses to guide its
development. The analytical lenses can be thought of as essentially master
hypotheses about possible determinants of the public value outcomes for the cases.
We articulate the lenses in terms of the following contextual factors that affect the
social impacts of research and science and technology policy.
a. Characteristics of the knowledge that the research produces. In some instances
knowledge creation processes, innovation, and, ultimately, social impacts are
very much governed by inherent characteristics of the science or technology
(e.g. ‘‘technology push’’), for example, conventional chemistry often uses
stoichiometric reagents as a basis for creating desired products because they are
convenient, economical, and traditional. But they are also wasteful and
polluting. In contrast, ‘‘green’’ chemistry, which includes as an explicit public
value the reduction of chemical pollution and toxicity, focuses on using
catalysts to achieve the same economic and public values as conventional
chemistry, but with much less waste (Logar, this issue).
b. Institutional arrangements and management affecting knowledge production
and use. This lens pertains to the configuration of producers and users of
scientific and technical knowledge, the ways in which they interact, their
internal and network management. In the case of climate change research, for
example, institutional networks developed largely in support of values
associated with ensuring high quality science, a situation that has proven very
hard to alter despite a growing awareness that the current arrangements are not
serving a range of public values that justify the investment in climate science
(Meyer, this issue). Slade (this issue) shows that, in research on nanotechnol-
ogy-based cancer treatments, insufficient attention to questions of diversity and
inclusiveness in both basic and clinical research potentially undermines the
capacity of the research to reduce health disparities.
c. Policy and political domains of knowledge production and use. This analytical
lens examines the political, legal, public policy and normative factors that

123
20 B. Bozeman, D. Sarewitz

determine research choices, utilization and impact (e.g. characteristics of

intellectual property policy or structures of budgets for research). Valdivia (this
issue) shows, for example, that although Congressional debates over technology
transfer policy in the early 1980s did indeed show attention to a broad range of
public values, the legislative and regulatory processes ended up neglecting such
values for a much narrower focus on economic efficiency and wealth creation.

Conclusion

We began this paper with Toulmin’s chalk-and-cheese metaphor pertaining to

incommensurate activities grouped together under the category ‘‘science.’’ His
metaphor speaks to specious aggregation, a difficulty that in turn conceals a still
greater problem: omission. We may recognize that biochemistry cannot easily be
played off against subatomic particle physics, but in its focus on the outputs of
science qua science, research evaluation elides the deeper policy question: on what
basis can we decide the extent to which either is worth doing? In publicly funded
science, public values are often displaced, minimized, misrepresented or altogether
missing. A common scenario (we hope not the most common) is ‘‘bait and switch.’’
During the agenda-setting phase of public policy, tax dollars spent on science and
research are rationalized in terms of explicit and invariably lofty public values
(everything is worth doing). After securing a place in the policy agenda, science
policies fall prey to the same interest group politics and forces of institutional inertia
that characterize most policy domains, often leading to incremental changes in the
status quo. Scientists find ways to rationalize just how the research they have always
been doing fits perfectly with the new goals and public value rhetoric. We could
refer to this as the ‘‘phrenology problem’’: were the government to provide a billion
dollars for phrenology research, many scientists would find ways to call themselves
phrenologists. Institutions are no less indefatigable in their adaptation. Bureaucratic
sinkholes serve as burial ground for public values. By the time research policy plays
itself out at the level of individual choice (that is to say, the same scientists
conducting science in the same institutional settings in which they have always
worked), we can only hope for some remnants of public value. At this point, any
public values that may have been prominent at the outset of research policies are
likely only to be resurrected to justify the next increment of funding.
This is, of course, a pessimistic view, perhaps even a bit extreme, but generally
descriptive of most instances of large-scale policy choice in science, where high
minded beginnings translate into claims of progress measured by publications,
citations, or patents and justified in terms of ‘‘scientific excellence’’ or contributions
to ‘‘innovation’’ or economic growth. One role of evaluation, including research
policy evaluation, is to hold feet to the fire. In the cases where the pessimistic
scenario described above is also a most accurate one, should we not expect research
evaluators to play a role as truthsayers, or if that is too ambitious a term, an
alternative voice? By this point, research evaluation has developed serious methods
and techniques and even if the term ‘‘science of science policy’’ remains more
aspiration than reality, evaluation has made great strides. However, those strides

123
Public Value Mapping and Science Policy Evaluation 21

have been in three directions, none leading directly to the amelioration of what we
may refer to, with apologies to Toulmin, as the chalk-cheese-choice problem. The
oldest form of systematic evaluation, peer review, while changed, remains
recognizable to peer reviewers of decades ago. Would we expect peer reviewers,
when ‘‘peer’’ is defined in terms of technical expertise, to provide insight into bait
and switch or the displacement of public values? Greater strides have been made in
the application of economics to research evaluation. While there is some potential
for adapting microeconomics and cost-benefit analysis to research evaluation, most
any economist will readily declare that the discipline does very well with questions
of efficiency and is generally unsuited to questions of equity. To the extent that
public values in science questions are efficiency questions, then economics has
much to contribute. Perhaps the greatest strides in research evaluation have been in
the area of bibliometrics. In less than thirty years, the field has gone from citation
counts to all manner of sophisticated analyses of academic productivity and
collaboration networks. That is to say, bibliometrics can tell us much about the input
side of science and quite a bit about the output side, but, as we have said, the impact
focus remains narrow: productivity and collaboration.
The research presented in these next papers is not as venerable as peer review,
not as precise as economics-based evaluation, and not as ruthlessly objective as
bibliometrics. However, the papers do seek, through detailed exploration of cases
using the PVM lens we have outlined here, to directly address the science policy
elephant-in-the-room: the public value of publicly funded science. We recognize
that these are tentative steps, but at least they are looking in the right direction,
rather than under the proverbial lamp post that shines its bright light on what we can
already do—assess scientific excellence and economic productivity—while leaving
the rest in abject obscurity. We believe the papers represent progress toward
addressing the chalk-cheese-choice problem but, even if they do not, we hope they
will stimulate others to give more thought to the formidable challenge of expanding
the domain of exploration to address and assess what is surely the core claim of
science policy: that science outcomes should serve and advance public values.
Acknowledgments The authors acknowledge the support of the U.S. National Science Foundation’s
‘‘Science of Science Policy’’ program (award number 0738203, Arizona State University, ‘‘Public Value
Mapping: Developing a Non-Economic Model of the Social Value of Science and Innovation Policy’’).
We are grateful for the assistance and ideas of the members of the ‘‘Public Value Mapping’’ project,
including: Catherine Slade, Ryan Meyer, Erik Fisher, Genevieve Maricle, Walter Valdivia, Nathaniel
Logar, Stephanie Moulton, Cynthia Schwartz, and David Guston.

References

Adams, Guy B. 1992. Enthralled with Modernity: The Historical Context of Knowledge and Theory
Development in Public Administration. Public Administration Review 52(4): 363–373.
Adams, John. 2006. The Failure of Seat-Belt Legislation. In Clumsy Solutions for a Complex World, eds.
M. Verweij, and M. Thompson, 132–154. Houndmills, UK: Palgrave Macmillan.
Anderson, Elizabeth. 1993. Value in ethics and Economics. Cambridge, MA: Harvard University Press.
Andrews, Frank M. 1979. Scientific productivity, the effectiveness of research groups in six countries.
Ann Arbor, MI.: University of Michigan Press.
Auditor General. 1993. Program Evaluation in the Federal Government. Treasury Board of Canada: The
Case for Program Evaluation.

123
22 B. Bozeman, D. Sarewitz

Audretsch, David B., Barry Bozeman, Kathryn Combs, Maryanne Feldman, Albert Link, Donald Siegel,
Paula Stephan, Gregory Tassey, and Charles Wessner. 2002. The Economics of Science and
Technology. Journal of Technology Transfer 27(2): 155–203.
Barbarie, Alain. 1993. Evaluating Federal R&D in Canada. In Evaluating R&D Impacts: Methods and
Practice, eds. Barry Bozeman, and Julia Melkers, 155–162. Norwell, MA: Kluwer.
Baumgartner, Frank R., and Bryan D. Jones. 1991. Agenda Dynamics and Policy Subsystems. Journal of
Politics 53(4): 1044–1074.
Bozeman, Barry. 2007. Public Values and Public Interest: Counter-balancing Economic Individualism.
Washington, D.C.: Georgetown University Press.
Bozeman, Barry. 2002. Public Value Failure and Market Failure. Lead Article, Public Administration
Review 62(2): 145–161.
Bozeman, Barry 2003. Public Value Mapping of Science Outcomes: Theory and Method. In D. Sarewitz,
et. al. Knowledge Flows & Knowledge Collectives: Understanding the Role of Science &
Technology Policies in Development. 2 (1).
Bozeman, Barry, James Dietz, and Monica Gaughan. 2001. Scientific and technical human capital: an
alternative model for research evaluation. International Journal of Technology Management 22(7/
8): 716–740.
Bozeman, Barry, and Julia Melkers (eds.). 1993. Evaluating R&D Impacts: Methods and Practice.
Boston: Kluwer.
Bozeman, Barry, and Juan R. Rogers. 2002. A Churn Model of Scientific Knowledge Value: Internet
Researchers as a Knowledge Value Collective. Research Policy 31(5): 769–794.
Bozeman, Barry, and Daniel Sarewitz. 2005. Public Values and Public Failure in U.S. Science Policy.
Science and Public Policy 32(2): 119–136.
Braybrooke, David, and Charles E. Lindblom. 1963. A Strategy of Decision: Policy Evaluation as a
Social Process. New York: Free Press.
Budd, John, and Lynn Connaway. 1997. University Faculty and Networked Information: Results of a
Survey. Journal of the American Society for Information Science 48(9): 843–852.
Cummings, Ronald, and Laura Taylor. 1999. Unbiased Value Estimates for Environmental Goods: A
Cheap Talk Design for the Contingent Valuation Method. American Economic Review 89(3):
649–665.
Feeney, Mary, and Barry Bozeman. 2007. The 2004–2005 Influenza Episode as a Case of Public Failure.
Journal of Public Integrity 9(2): 179–195.
Fischer, Ernest Peter. 1997. Beauty and the Beast: The Aesthetic Moment in Science. trans. Elizabeth
Oehlkers. New York: Plenum Trade.
Fisher, Erik, Catherine Slade, Derrick Anderson and Barry Bozeman. 2010. The Public Value of
Nanotechnology? Scientometrics 85(1):29–39.
Freeman, Christopher. 1992. The Economics of Hope: Essays on Technical Change, Economic Growth
and the Environment, London: Pinter Publishers, 1992. London: Thompson Learning.
Garrison, Jim. 2000. Pragmatism and Public Administration. Administration and Society 32(4): 458–478.
Gaus, Gerald F. 1990. Value and Justification: The Foundations of Liberal Theory. New York:
Cambridge University Press.
Johnson, Harry G. 1965. Federal Support of Basic Research: Some Economic Issues. Minerva 3(4):
500–514.
Jones, Charles I., and John C. Williams. 1998. Measuring the Social Return to R&D. Quarterly Journal of
Economics 113(4): 1119–1135.
Kevles, Daniel. 1995. The Physicists: The History of a Scientific Community in Modern America.
Cambridge, MA: Harvard University Press.
Kostoff, Ronald. 2001. The Metrics of Science and Technology. Scientometrics 50(2): 353–361.
Holdren, John P. 2009. Science and Technology Policy in the Obama Administration, Remarks for the
Business Higher Education Forum, Washington, D.C., 16 June (Powerpoint presentation).
Kirlin, John. 1996. What Government Must Do Well: Creating Value for Society. Journal of Public
Administration Theory and Research 6(1): 161–185.
Leslie, Stuart W. 1993. The Cold War and American Science. New York: Columbia University Press.
Link, Albert N. 1996a. Economic Performance Measures for Evaluating Government Sponsored
Research. Scientometrics 36(3): 325–342.
Link, Albert N. 1996b. Evaluating Public Sector Research & Development. New York: Greenwood.
Luukkonen, Terttu. 2002. Research evaluation in Europe: state of the art, 11 (2): 81-84.

123
Public Value Mapping and Science Policy Evaluation 23

Luukkonen-Gronow, Terttu. 2007. Scientific Research Evaluation: A Review of Methods and Various
Contexts of their Application. R&D Management 17(3): 207–221.
Machlup, Fritz. 1962. The Production and Distribution of Knowledge in the United States. Princeton, NJ:
Princeton University Press.
Marburger, John. 2005. Speech at the 30th Annual AAAS Forum on Science and Technology Policy in
Washington, D.C. (April 21), available at: http://www.aaas.org/news/releases/2005/0421marburger
Text.shtml.
Marmolo, Elisabetta. 1999. A Constitutional Theory of Public Goods. Journal of Economic Behavior &
Organization 38(1): 27–42.
Martens, Karel. 2009. Equity Concerns and Cost-Benefit Analysis: Opening the Black Box. Washington,
DC.: Transportation Research Board, Paper #09-0586.
Nye, Joseph. 1997. In Government We Don’t Trust. Foreign Policy 108(2): 99–111.
OECD. 1997. The Evaluation of Scientific Research: selected experiences. Paris: OECD, Committee for
Scientific and Technological Policy. Document OECD/GD(97)194. http://www.oecd.org/dsti/sti/s_t/
scs/prod/e_97-194.htm.
OECD. In press. Enhancing Public Research Performance through Evaluation, Impact Assessment and
Priority Setting. Paris: OECD, Directorate for Science, Technology and Industry.
Polanyi, Michael. 1962. The Republic of Science: It’s Political and Economic Theory. Minerva 1(1):
54–73.
Rubenstein, Albert. 1976. Effectiveness of Federal Civilian-Oriented R&D Programs. Policy Studies
Journal 5(2): 217–227.
Rosenberg, Nathan. 1982. How Exogenous is Science? In Inside the Black Box (NY: Cambridge
University Press), p. 141–159.
Ruegg, Rosalie. 1996. ‘‘Guidelines for Economic Evaluation of the Advanced Technology Program,’’
NIST Internal Report 5896.
Ruttan, Vernon. 2006. Is War Necessary for Economic Growth? New York: Oxford University Press.
Salasin, John, Lowell Hattery, and Ramsey Thomas. 1980. The Evaluation of Federal Research Programs,
MITRE Technical Report MTR-80W123, June 1980.
Sarewitz, Daniel. 1996. Frontiers of Illusion: Science, Technology, and the Politics of Progress.
Philadelphia, PA: Temple University Press.
Shields, Patricia M. 1996. Pragmatism: Exploring Public Administration’s Policy Imprint. Administration
and Society 28(3): 390–411.
Shils, Edward. 1968. Introduction. In Criteria for Scientific Development: Public Policy and National
Goals, ed. E. Shils. Cambridge, MA: MIT Press, p. iv–v.
Solow, Robert M. 1957. Technical change and the aggregate production function. Review of Economics
and Statistics 39(3): 312–320.
Toulmin, Stephen. 1964. The Complexity of Scientific Choice: A Stocktaking. Minerva 2(3): 343–359.
Van Deth, Jan W., and Elinor Scarbrough. 1995. The Impact of Values. Oxford: Oxford University.
Van Houten, Therese, and Harry Hatry. 1987. How to Conduct a Citizen Survey. Washington, D.C.:
American Planning Association.
Weinberg, Alvin. 1963. Criteria for Scientific Choice. Minerva 1(2): 159–171.
Woodhouse, Edward, and Daniel Sarewitz. 2007. Science Policies for Reducing Societal Inequities.
Science and Public Policy 34(2): 139–150.
Ziman, John. 1968. Public Knowledge: The Social Dimensions of Science. Cambridge: Cambridge
University Press.

123