Full Text

Where You Stand Depends on What You See:
Connections among Values, Perceptions of Fact, and Political Prescriptions
Jennifer L. Hochschild
Princeton University
Prepared for Citizens and Politics: Perspectives from Political Psychology, edited by James
Kuklinski (Cambridge University Press, 1999).

We must note particularly... the insertion between man and his environment of a pseudoenvironment.
To that pseudo-environment his behavior is a response.... The analyst of
public opinion must begin then, by recognizing the triangular relationship between the
scene of action, the human picture of that scene, and the human response to that picture
working itself out upon the scene of action.
-- Walter Lippmann1

One good solid murder of a baby or a rape-murder of a 7-year-old girl will outweigh a ton
of statistics.
-- Lawrence Friedman, explaining popular
perceptions of a rising crime rate.2

Including values as a central component of the study of public opinion was long overdue,
and therefore their appearance in this volume is a cause for celebration. Of course, if values are
defined broadly enough, public opinion researchers have been studying values for decades if not
centuries. Alexis de Tocqueville "sampled" the American public in the 1830s, and analyzed
Americans' distinctive values both as "independent" and as "dependent" variables. So did James
Bryce a few decades later, and Hector St. John de Crévecoeur a few decades earlier. Scholars
using qualitative interviews and ethnographic techniques to explore values succeeded Europeans
with notepads around the turn of the century (Lynd and Lynd 1929; Warner et al. 1949; Lane
1962; Hochschild 1981), and they in turn were succeeded by survey researchers measuring the
origins of authoritarianism or the importance of family ties in partisan identification (Niemi and
Weisberg 1993: sections II and III; Adorno et al. 1950).
Acknowledging venerable ancestry, however, does nothing to diminish the claim that the
study of citizens' values has recently moved into maturity. The array of values under serious
investigation has expanded beyond Rokeach's grid or Maslow's hierarchy into a manageablysized
set of core concepts. Those concepts -- egalitarianism, individualism, humanitarianism,
religious fundamentalism, post-materialism, ethnocentrism -- are themselves being defined more
rigorously and compared more systematically than in earlier eras (for example, see Sears,
Sidanius and Bobo forthcoming). Values are being asked to do more explanatory work on more
issues than ever before, and they are responding in provocative ways.
The study of the role of values in public opinion is becoming broader as well as deeper.
Surveyors are now able to compare values held by citizens of different nations, and they are
using those comparisons to defend claims ranging from universal cultural transformation to
sharp ethnic differences within putative nation-states (Abramson and Inglehart 1994; Gibson and
Duch 1993; McIntosh and MacIver 1992; Hofrichter 1993). Cross-national comparisons of
values, and of their causes and effects, allow us to get a handle on the old question of American
exceptionalism as well as on new questions of cultural capital and the consequences of collective
identification.
The new studies of values make a methodological contribution almost as great as their
substantive ones. One must be as adept at interpreting political philosophy as regression
coefficients to understand the content and effect of values in citizens' belief systems. That fact
destroys, because it demonstrates the evident silliness of, methodological disputes between
science and philosophy, interpretation and explanation, qualitative and quantitative methods.
One must get one's philosophy right to do interesting science about values; one must get one's
science right to make empirical contributions to democratic theory and practice. That is probably
true for most interesting questions of politics, but it is a fortiori true for the study of values in
public opinion, so students of this topic can show the way for other social scientists.

What Not To Do in Studying Values
I want first to note four dangers facing the study of values, before exploring the edges of
yet another new frontier that the study of values opens up. First, we must take care not to allow
the study of values to be subsumed by the study of ideology, especially the traditional liberalconservative
dimension. That dimension is itself deeply problematic conceptually and often has
little leverage politically. Social conservatives, who want the government to enforce traditional
norms of morality through legislation and exhortation, have little in common with libertarians,
who are not much concerned about traditional norms of morality and who want as little
governmental intervention in people's lives as possible. Yet both are called "conservatives" in
common parlance. Similarly, "liberals" may all share egalitarian convictions and a belief that
government should act to promote equality, but they vehemently disagree with one another over,
for example, whether to promote universal policies of family support or to target assistance to
the poorest or most discriminated against. When Robert Kennedy was assassinated in 1968,
many of his supporters turned next to George Wallace; H. Ross Perot's supporters in 1992 were
about evenly divided between George Bush and Bill Clinton as their second choice for President.
As Greg Markus demonstrates in this volume, liberals are individualists when individualism is
defined as autonomy, and conservatives are individualists when individualism is defined as self4
reliance.
In short, one can distinguish between liberals and conservatives philosophically only by
adding many qualifiers, and one can make few predictions about how individuals' attitudes will
be arrayed or how political disputes will be resolved by labeling one set of views or contenders
"liberal" and the other "conservative." Students of the role of values in public opinion know all
of this; frustration with the liberal-conservative dimension is largely why they turned to the
study of values to begin with. But the dimension has such a powerful hold on American
linguistic norms and such deep roots in the academic study of public opinion, that it will be
equally hard to relinquish it or to incorporate it into work on values without letting it take over.
A second danger in studying values is that the researcher will be tempted to move directly
from determining that individuals share a given value to presuming that they share a given
attitude or policy prescription. That is, one might presume that individualists will oppose
affirmative action and egalitarians will support it. But an egalitarian like Randall Kennedy
(1997) or Orlando Patterson (1973) may oppose affirmative action because it perpetuates an
employer's tendency to think in hierarchical racial categories rather than to see people as
potentially equal separate persons. Or an individualist may support affirmative action programs
because they are the only way to ensure that people with certain ascriptive traits get the chance
to show what they can really do (Rosenfeld 1992).
There are actually two dangers here. One is that researchers will unwarrantedly infer
respondents' attitudes or policy prescriptions from their values. A second danger is that
researchers will actually do the work to determine how prescriptions relate to values, but then
assume that if the respondent's policy preferences do not "fit" his or her value in a way that
makes sense to the researcher, the respondent's attitudes are "unconstrained" (in a new usage of
this term). After climbing out of the hole caused by the assumption that views should be arrayed
along a single ideological dimension (such as liberalism/conservatism), one may fall into the
nearby trap of assuming that specific policy positions should map neatly onto general values.
This danger arises at the level of societies as well as of individuals. It would be a mistake,
for example, to assume that because Americans are more individualistic than the French, they
have a less robust social welfare state. That conclusion depends on a narrow definition of social
welfare. For example, if one includes a free system of lightly-tracked public education through
age 18 (and an almost-free system of somewhat more tracked higher education) in one's
definition of a social welfare state -- a system that is well-suited to a highly individualistic and
success-oriented society -- the U.S. is not a welfare laggard at all. Thus societies with different
value emphases may have similar, or at least comparable, policy outcomes. Conversely,
societies with similar value emphases may have highly disparate policy outcomes; Great Britain
and the U.S. are almost equally individualistic, but out of that shared normative base they have
developed very different health care systems (Jacobs 1993).
A third potential pitfall for the study of values lies in defining the relationship between
values and self-interest. Some define the relationship as antithetical: altruism is contrasted with
self-protection, or adversary democracy is posed against unitary (or communitarian) democracy
(Monroe 1996; Mansbridge 1980, 1990). Others define self-interest as one among other values;
in this volume Stoker defines partial judgments as more self-interested than impartial judgments,
but they are no less values. Others such as Feldman and Steenbergen in this volume define the
degree to which one is self-interested or empathetic as a feeling rather than a value, so self6
interest is neither posed against nor included within the category of "values."
All of these choices are conceptually defensible and all can do real work in an analysis of
public opinion. The dangers come when analysts are imprecise about how they relate values to
interests, when they are unwarrantedly imperialistic or timid, or when they choose a definition
tendentiously. The dangers of imprecision are sufficiently obvious as to need no further
discussion. The dangers of imperialism would occur, for example, if a researcher redefined the
value of altruism into (perhaps genetically-driven) self-interest over a long period, thereby losing
an interesting distinction in the service of an ideological claim.3 That problem would be
compounded if a researcher who responds that altruism or love really does exist is pushed into
seeking for the "pure" Kantian altruist (or the perfect parent!), who cannot admit any selfinterested
motives and still be considered altruistic.
The other problem that occurs frequently in the context of relating interests to values lies
in submerged valuations masquerading as definitions. The category of "self-interested" is elided
with "selfish;" alternatively, respondents exhibiting empathy are made to seem naive or
hypocritical. The clearest example of this phenomenon that I know of, with perhaps the greatest
political justification, is the claim that African Americans cannot be racist. By definition in this
view, racism is necessarily associated with dominance; thus in the United States only whites, or
the mainstream society, can be racist. If proponents of this claim were truly concerned with
definitional exactness, they could easily distinguish between institutional racism, which can only
be enacted by the dominant race, and individual racism, which may be expressed by anyone. But
their concern is not definitional precision; it is rather to link the pejorative connotation of
“racist” with whites and to proclaim blacks’ innocence. One may sympathize with the anger
behind this elision, but from the perspective of the study of values in public opinion, building a
sharp value judgment into a definition seems mistaken.
A final danger in the study of values in public opinion is the temptation of product
differentiation. The best analysts of public attitudes have always examined values under one or
another rubric. That fact does not undermine the importance of the new attentiveness to values,
so it should be built on rather than downplayed. Conversely, students of values should not
dismiss the old focuses on party identification, childhood socialization, or demographic
characteristics in their eagerness to broach new frontiers. This point is more than an exhortation
to be comprehensive; as I argue below with regard to demographics, one may sometimes be
unable to analyze values properly without embedding them in other characteristics of the
population being studied.

What to Add to the Study of Values
Having just argued for methodological expertise ranging from philosophy to statistics,
and for adding the study of values to the plethora of old concerns rather than allowing it to
substitute for them, I will now compound my unreasonableness by seeking to persuade students
of values in public opinion to add an additional element to their analyses. That element is
respondents' perceptions of facts.
The rest of this chapter focuses on what we know about citizens’ perceptions of facts (not
very much), and what we need to learn (a lot). But I want first to set the stage with a few
observations about why we should care.
Various people should care about citizens’ perceptions of facts for varied reasons.
Psychologists have a long tradition of distinguishing beliefs from values, attitudes, opinions,
preferences, emotions, causal attributions and so on, and seeking to tease out the relations among
those forms of mentation. That is a worthy enterprise, but not mine. Behavioral students of
public opinion have an equally long tradition of seeking to explain why some citizens hold
certain values, attitudes, causal attributions etc., whereas other citizens hold other values,
attitudes, etc. That is the chief enterprise of most contributors to this volume, but in the end it
also is not mine. Political actors seek to change people’s values, attitudes etc., or to galvanize
people who already hold the “right” values and attitudes into political action -- that too is not my
agenda.
I wish to focus on the political, policy, and normative implications of the fact that people’s
perceptions of fact are frequently wrong, and wrong in particularly patterned ways. Let us start
with a simple but not trivial example: most Anglo Americans estimate the number of black and
Latino Americans to be at least twice the actual number in the American population. It makes
good sense to oppose a strong program of affirmative action in that case, because if half of the
population are potential beneficiaries of “special preferences,” you are quite likely to lose your
job or promotion to them. Now suppose that whites suddenly learned that only 12 percent of the
population is black and 10 percent is Latino (and a disproportionate number of them are children,
and thus not competitors for coveted jobs). Might those newly-educated whites now be more
willing to say, “I still don’t like affirmative action but it is not the threat that I thought it was, so
I guess I can live with it”? Probably not -- none of us are that logical, and opposition to
affirmative action has many more causes than an arithmetic calculation of its threat to one’s own
position. But that hypothetical example suggests the reason for my interest in this topic.
To what degree are citizens’ political attitudes and policy preferences based on, or at least
supported by, mistaken perceptions of fact? How much credence should we grant to such
political and policy views? How firmly can we assert that people act in their own interests if
they do not know facts that might be crucial to their deciding what is in their interest? How much
effort should be devoted, and by whom, to helping people get the facts right before they make
judgments about what should be done in the political arena? How malleable are people’s values
on the one hand, and policy preferences or political attitudes on the other, if somehow they did
learn to correct their perceptual mistakes? May we legitimately evaluate the quality of the
democratic process partly in terms of how well informed citizens were when they elected
political leaders or responded to referenda offerings? Should factually-oriented decision makers
be granted more authority than others?
These are ultimately unanswerable questions, because they involve deep and inherently
contestable normative positions, because we cannot conduct experiments with individual voters
or the democratic process, and because they focus on only one piece of an enormously
complicated structure of political decision-making. Nevertheless, they are important questions,
and I want to draw our attention to their importance by focusing on an aspect of public opinion
that is poorly developed compared with the study of attitudes or even of values.

Misperceptions of Social Facts
Social scientists have pointed out for a long time that many Americans do not know basic
political facts such as the name of the vice president or the meaning of the Bill of Rights.4 That
phenomenon is important politically and normatively. But it does not in itself speak to the role
of values in shaping public opinion; one can be a strong individualist and support freedom of
speech even for expounders of disliked opinions without being able to identify the First
Amendment.
Some perceptions of fact do, however, matter for our interpretation of values and for our
understanding of how general values connect with particular political attitudes and policy
preferences: