Paul Goren - On Voter Competence
Paul Goren - On Voter Competence
Paul Goren - On Voter Competence
Series Editor
John T. Jost
Editorial Board
Mahzarin Banaji, Gian Vittorio Caprara, Christopher Federico, Don Green,
John Hibbing, Jon Krosnick, Arie Kruglanski, Kathleen McGraw, David Sears,
Jim Sidanius, Phil Tetlock, Tom Tyler
Representing Red and Blue: How the Culture Wars Change the Way Citizens
Speak and Politicians Listen
David C. Barker and Christopher Jan Carman
On Voter Competence
Paul Goren
On Voter Competence
Paul Goren
1
3
Oxford University Press is a department of the University of Oxford. It furthers the
University’s objective of excellence in research, scholarship, and education by publishing
worldwide.
Oxford New York
Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto
With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam
Oxford is a registered trade mark of Oxford University Press in the UK and
certain other countries.
Published in the United States of America by
Oxford University Press
198 Madison Avenue, New York, NY 10016
© Oxford University Press 2013
All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system, or transmitted, in any form or by any means, without the prior permission in writing
of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed
with the appropriate reproduction rights organization. Inquiries concerning reproduction
outside the scope of the above should be sent to the Rights Department, Oxford University
Press, at the address above.
You must not circulate this work in any other form
and you must impose this same condition on any acquirer.
9 8 7 6 5 4 3 2 1
Printed in the United States of America on acid-free paper
Contents
Acknowledgments vii
References 249
Index 263
v
This page intentionally left blank
Acknowledgments
vii
viii ACKNOWLEDGMENTS
generous help. I do know it would have been much worse. Howie Lavine
read every chapter, gave excellent commentary across the board, and supplied
plenty of help along the way. Chris Federico read most of the book, made
many astute comments about psychological theory and American politics, and
offered sage statistical advice. I thank him too for scoring tickets for the Boris ~
Kurihara and GBV shows. Chris Chapp somehow managed to read my man-
uscript while completing his own book. His commentary was always expert
and his suggestions spot on (and his spots in the rack greatly appreciated). Ed
Schiappa read the entire manuscript with remarkable dispatch, offered cogent
advice, and pushed me to keep my eye on the big picture, all the while serving
as chair of the Department of Communication Studies at Minnesota. Where he
found the time I will never know. Tim Johnson and David Kimball read the
complete manuscript as well and made a number of very useful suggestions.
They too pushed me to emphasize broad themes and clarify my prose. David
Samuels, Markus Prior, and Aaron Hoffman deserve credit on this score as
well. Joanne Miller offered much needed criticism on early drafts of the con-
ceptual/theoretical chapters, forcing me to simplify matters in my own mind.
Two historians, James Patterson at Brown University and Robert Collins at
the University of Missouri, answered an out-of-the-blue cry for help from
a political scientist desperately trying to condense a great deal of historical
material into a coherent narrative. Both graciously read drafts of Chapter 2
with a careful and critical eye, saved me from several embarrassing factual
errors, and taught me a thing or two about the historian’s craft. Finally, I thank
Paul Sniderman. Though I did not realize it at the time, the idea for the book
originated in a conference paper I presented at the American Political Science
Association’s annual meeting in 2001. After the presentation, Paul introduced
himself, shared some kind words about my efforts, and encouraged me to
think big. Later on in another venue, he furnished critical advice about the
project, all of this despite the fact that some of what I was doing challenged
his work.
Many other people provided helpful comments or assistance along the way,
including Scott Abernathy, Liz Beaumont, John Bullock, Logan Dancey, Mat-
thew DeBell, John Freeman, Kim Fridkin, Sarah Allen Gershon, Jim Gimpel,
Jessica Goren, Rick Herrera, Jon Hurwitz, Bill Jacoby, Stephen Jessee, Andy
Karch, Pat Kenney, Miki Caul Kittilson, Pay Luevano, Makoto Kawabata, Ron
Krebs, Jay McCann, Bill McCready, Takashi Mizutani, Jeff Mondak, Christopher
Muste, Kathryn Pearson, Brian Rathbun, Mark Ramirez, Shalom Schwartz, Phil
Shively, Dara Strolovitch, John Sullivan, Shawn Treier, and Joan Tronto. Finan-
cial support for the three surveys described in Chapter 7 was provided by the
Graduate School and the Center for the Study of Political Psychology at the
Acknowledgments ix
1
2 ON VOTER COMPETENCE
those who are preparing to divide us, the spin masters and the negative ad ped-
dlers who embrace the politics of anything goes. Well, I say to them tonight,
there’s not a liberal America and a conservative America; there’s the United
States of America” (Washington Post, 2004).1
These examples illustrate how political leaders invoke liberal and conser-
vative labels to communicate with voters. They do so because these labels
summarize information about candidate and party positions on the proper
economic, social, and political order; the role of government; and positions
on diverse issues. Those comfortable with ideological frames of reference can
follow the dialogue with ease. When a politician is described as conservative,
exceptionally sophisticated citizens—those who are deeply informed about
government and politics—can infer the candidate favors lower taxes, smaller
government, the right to life, higher military spending, and so on. And when
hearing the opponent is liberal, they recognize this candidate holds oppos-
ing positions. Knowing where the candidates fall on the liberal–conservative
continuum, highly sophisticated voters utilize a simple and effective decision
rule: choose the candidate who lies closer to them on the liberal–conservative
spectrum. Policy voting is thereby assured.
But what of those who quickly scan the front page of the local paper before
turning to the sports pages? What of those who pause briefly, if at all, on the
evening news before switching to reruns of their favorite sitcom? How about
people who surf the web for everything but political news? What about those
who have never listened to Keith Olbermann or Bill O’Reilly, who have never
read Paul Krugman or David Brooks, who know more about the celebrity
scandal du jour than the federal budget? Do these citizens know what liber-
alism and conservatism mean? Can they use ideological labels to summarize
positions on dozens of issues? Will they ground their votes in abstract liberal
or conservative principles?
Scholars’ answers to each of these questions is “no.” There is no doubt
that locations along the liberal–conservative continuum represent the ultimate
political shorthand, the political heuristic par excellence, in American political
discourse (Downs, 1957). But since most members of the public do not under-
stand what liberalism and conservatism mean, widespread ideological voting
does not occur. Citizens perform no better when it comes to concrete issues.
Notwithstanding the occasional controversy that captures the public’s atten-
tion, large swatches of the electorate lack meaningful attitudes on the issues of
the day. Hence, the prospects for pervasive issue voting appear equally dim.
1
Unless otherwise noted, all convention and state of the union quotes are from
Woolley and Peters (2011).
The Indictment of the American Voter 3
This pessimistic view has dominated the study of voting behavior for over
half a century. In 1960, Campbell, Converse, Miller, and Stokes concluded their
landmark work The American Voter on this dour note.
Nearly half a century later, Kinder (2006) concurred: “when it comes to poli-
tics, most citizens are ideologically innocent: indifferent to standard ideolog-
ical concepts, lacking a consistent outlook on public policy, in possession of
genuine opinions on only a few issues, and knowing damn little” (p. 199).
The normative corollaries attending this empirical portrait are bleak. In
some recent reflections, Converse (2006) opined:
I cannot say that voters have to be users of the ideology heuristic to vote
‘sensibly.’ On the other hand, since it offers unusual powers of economy in
both understanding arriving flows of political information and in retain-
ing it, and since the world of politics is objectively complex and the voter
is a cognitive miser, its use is to be recommended, and the fact that it is not
widely understood is, to me, a central indicator of the problem surround-
ing voter competence. (p. 310)
On Voter Competence takes a far more optimistic view of the American voter.
In this book, I argue everyday citizens choose based on core policy principles,
but that professional students of electoral behavior have missed this because
they have searched for evidence of policy voting in the wrong places. Once we
turn away from liberal–conservative predispositions and preferences on dis-
crete issues, we discover that nearly everyone in the mass public holds policy
principles and uses these to guide candidate selection in U.S. presidential elec-
tions. Contrary to the indictment leveled by much of the scholarly commu-
nity, citizens who are not deeply informed about public affairs prove as adept
as their more sophisticated counterparts at grounding presidential votes in
abstract views about public policy.
Three principles corresponding directly to the major policy cleavages that
have divided the Democratic and Republican parties for the past several
decades are paramount: limited government, traditional morality, and military
strength. My most important claims are that attitudes toward these principles
are (1) available in the minds of nearly all citizens, (2) function as central heu-
ristics in the belief systems of the politically sophisticated and unsophisticated
4 ON VOTER COMPETENCE
alike, (3) are rooted deeply in basic human values, and (4) guide presidential
choice to a comparable degree for voters across the sophistication spectrum.
Put simply, the critical point to take away from my book is that most citizens
have genuine policy principles and rely heavily on these when casting pres-
idential ballots. Insofar as we can equate the development and use of policy
principles with citizen competence, the American voter performs far better
than has been recognized.
2
Policy principles also stand apart from basic human or personal values, a con-
cept with which they are sometimes confused. Policy principles center on what
people think should be done in a given issue area in the political sphere, whereas
personal values revolve around the importance individuals attach to abstract
goals that transcend politics (Rokeach, 1973; Schwartz, 1992). I expand on these
differences in Chapter 7.
The Indictment of the American Voter 7
most voters these attitudes are imbued with little policy content, revolving
instead around gut-level feelings about disparate social groups and symbols.
Even among the most sophisticated third or so of the public, ideological vot-
ing is probably best viewed as an expression of symbolic affinity for a given
candidate rather than an attempt to send policy signals about what the can-
didate should do if elected. In a similar vein, accumulated research finds that
those who are knowledgeable about politics in general or some particular con-
troversy issue vote more than those lacking knowledge (Anand & Krosnick,
2003; Delli Carpini & Keeter, 1996). Other research suggests a salient issue may
weigh heavily on voter choice in a given election, but fail to matter in subse-
quent contests (Abramowitz, 1995).
On Voter Competence accepts that most Americans are innocent of ideology
and fuzzy on most issues, but it rejects the corollary that their presidential
votes are untouched by policy considerations. In the pages to follow, I argue a
set of bedrock principles shapes voter choice to a comparable degree for politi-
cally aware and unaware citizens. It matters little whether people know a
great deal about government and politics, whether they prefer The News Hour
or Smackdown, whether they read Thomas Sowell and Frank Rich or Stephen
King and Nora Roberts. Nearly everyone acquires and subsequently uses pol-
icy principles to guide candidate choice.
At this point, it should prove helpful to preview the conceptual and theoret-
ical arguments I make throughout the book. First, I draw upon the Eagly and
Chaiken definition of the attitude construct to argue that liberal–conservative
orientations, issue preferences, and policy principles are best conceptualized
as different types of policy attitudes. Eagly and Chaiken (1993) define an atti-
tude as “a psychological tendency that is expressed by evaluating a particular
entity with some degree of favor or disfavor” (p. 1). This inclusive definition
runs the gamut from dispositional psychological states that endure over time
to temporary evaluations constructed on the spot as demanded by the situa-
tion. It applies readily to the aforementioned policy attitudes.
Next, I argue that most individuals develop and maintain genuine atti-
tudes toward limited government, traditional morality, and military strength.
By develop and maintain, I mean that domain-specific principles are available
and accessible in mass belief systems. To understand how ordinary men and
women acquire these, it must be recognized that the idea embodied by each
principle is sufficiently clear that it can be evaluated without difficulty (cf.
Carmines & Stimson, 1980). In contrast to ideological labels, whose meaning
eludes the uninitiated, there is no need to learn what “government help for
average people” or “traditional morals” or “force versus diplomacy” signify.
Large stores of political knowledge are not required for deciding how we feel
8 ON VOTER COMPETENCE
about these ideas. Because the meaning of each is self-evident, evaluation fol-
lows automatically upon exposure.
The question remains as to whether citizens receive sufficient exposure to
these ideas to actually evaluate them. There are compelling grounds for believ-
ing that they do. Political debate in this country has revolved around the role
the federal government should play in the economic welfare domain since
Franklin Roosevelt established the New Deal in the 1930s, around what moral
vision should prevail in American society since the 1960s, and about the proper
use of military force since the onset of the Cold War in the late 1940s. Because
these divisions have been so prominent for so long members of the voting
public need not monitor political debate in real time to learn what the major
policy dimensions are. Political and campaign discourse are rich in overlap-
ping cues that tell the same story about political conflict election after election
and year after year (Feldman, 1988; Rahn, Aldrich, Borgida, & Sullivan, 1990).
Although most people do not know what liberalism and conservatism mean
and remain unclear about, if not oblivious to, the latest controversy preoccu-
pying the chattering classes in Washington, DC and the blogsphere, they see
where the major fault lines lie in the political system. As such, they can form
durable impressions about the core conflicts that drive American politics. In
this manner, attitudes toward limited government, moral traditionalism, and
American military power become lodged in mass belief systems.
I then integrate research on political psychology and public opinion to
argue these orientations operate as central political heuristics in the minds of
most citizens. Political attitudes do not exist in isolation, but rather are linked
together in broader associative networks (Hurwitz & Peffley, 1987). The key
distinction in these attitude structures is between their central and periph-
eral elements. Central attitudes systematically affect peripheral evaluations
without, in turn, being shaped by them. Policy principles, I argue, influence
short-term political evaluations along these lines. Because almost all voters
have some experience using principles to guide the construction of issue pref-
erences and other transitory evaluations, they can be applied heuristically to
the problem of candidate choice. Doing so allows voters to decide quickly
and effectively without taxing their finite cognitive resources. The question
of origins is then taken up. I posit that basic human values and party identi-
fication are candidates to shape policy principles. Moreover, personal values
should guide the positions people take on policy principles both directly and
indirectly via their influence over partisanship.
Lastly, I posit that policy principles drive the presidential vote to an anal-
ogous degree for sophisticated and unsophisticated voters. This should hold
when factors such as partisanship, liberal and conservative orientations,
The Indictment of the American Voter 9
retrospective judgments, and particular issues are taken into account. Because
views on government power, moral conventions, and military might are
widely held and readily invoked to inform short-term preferences, voters
should encounter little difficulty linking principles to candidate evaluations as
long as they recognize where the candidates stand on each dimension. Because
presidential contenders and the national parties have a long history of taking
clear and contrasting stands on these principles, reminders of which appear
regularly in ongoing political debate and over the course of every presidential
campaign, most voters develop a sense that Democrats stand to the left of the
GOP on each cleavage. In this way, cue consistency and redundancy facili-
tate principle-based choice. In contrast to ideological and issue voting, both of
which are conditional on prior knowledge and interest, principle-based vot-
ing is not limited to a thin slice of the electorate. Instead, policy principles
shape candidate choice for nearly everyone.
To sum up, we must broaden our theoretical understanding of policy vot-
ing beyond liberal–conservative orientations and issue preferences. Once we
do we find that regardless of how much or little Americans know about poli-
tics, abstract policy principles are available in their political psyches, serve as
central heuristics in their belief systems, and guide their presidential votes.
The combination of political ignorance and ideological naiveté bemoaned so
often and so vociferously by so many does not preclude widespread policy
voting in the American electorate.
CONTRIBUTIONS
Numerous variants of democratic theory hold that if the demos are to rule in any
meaningful sense of the term, they must play a leading role in the determina-
tion of public policy (Pennock, 1979; Thompson, 1970). For this to happen, indi-
viduals must hold policy attitudes and use these to guide their choices come
Election Day. The question of whether citizens meet these conditions has gen-
erated a tremendous amount of empirical research over the years. Despite the
occasional dissent (Aldrich, Sullivan, & Borgida, 1989; Ansolabehere, Rodden,
& Snyder, 2008), the dominant view holds that many if not most voting age
adults lack genuine ideological and issue attitudes, implying that policy vot-
ing lies beyond the reach of the typical American voter (Berelson, Lazarsfeld,
& McPhee, 1954; Campbell et al., 1960; Converse, 1964; Delli Carpini & Keeter,
1996; Lau & Redlawsk, 2006; Lewis-Beck, Jacoby, Norpoth, & Weisberg, 2008;
Luskin, 2002; Neuman, 1986). The normative corollary is clear: the demos fail
to meet a key criterion of political competence. This is perhaps the most dam-
aging charge in the indictment against the American voter.
10 ON VOTER COMPETENCE
indeed facilitate some desirable political behaviors, but its absence need not
preclude them.
This is not to say that sophistication is irrelevant. As I show in Chapter 5,
sophistication promotes attitude stability for both policy principles and party
identification. Similarly, evidence in Chapter 6 suggests the sophisticated rely
a bit more heavily on general principles than the unsophisticated when fig-
uring out issue positions. Findings presented in Chapter 8 reveal that sophis-
tication enhances ideological voting. The sophistication interaction model
possesses some validity. But, as I will show, these are marginal rather than
fundamental differences. Said otherwise, the sophistication gradient is mild
rather than steep. My book affirms that sophistication matters in some ways,
but less so than typically imagined.
Fourth, On Voter Competence has positive implications regarding the polit-
ical intelligence of the American voter. Normative and empirical theories of
democracy insist that policy views must impact voter choice (Fiorina, 1981;
Key, 1966; Pennock, 1979). Policy views signal what people want the national
government to do and allow them to hold elected officials accountable for
their actions in office. Indeed, these signals provide a means through which
nonelites exercise some measure of popular control over government. Because
my work shows that most citizens possess genuine attitudes about the funda-
mental divisions that define policy space in the United States and call upon
these to reward or punish the candidates and parties, as well as signal the
direction government should take in a given policy area, fears about voter
incompetence are overblown. Put simply, the typical American voter performs
reasonably well as judged by these tenets of democratic theory.
3
I designed two of these surveys and commissioned Knowledge Networks to field
them in 2007 and 2008. The third survey was designed by a team of researchers
led by William Chittick and Jason Reifler and administered in 2011 by YouGov/
Polimetrix.
This page intentionally left blank
CHAPTER 2
15
16 ON VOTER COMPETENCE
of each cleavage, trace its evolution over the long haul, and examine its con-
nections to presidential politics and the national parties. Doing so helps illu-
minate the psychological processes I elaborate in Chapters 3 and 4 by which
ordinary citizens acquire and subsequently apply policy principles to their
electoral choices. Readers should note that I delimit my historical analysis in
three ways. First, discussion centers on what I see as the major developments
within the aforementioned policy domains. A detailed account of every issue
controversy lies beyond the scope of my inquiry. Second, given my interest in
policy voting in U.S. presidential elections, I highlight the politics of presiden-
tial campaigning and governance. Far less attention is devoted to the congres-
sional side of the ledger. Third, issues in other domains that have experienced
short lifespans command little attention in what follows.
1
My discussion of each issue area has been informed by many outstanding works
of history and political science. For the economic welfare domain, I draw on Blum
(1991), Boyer (2001), Brewer and Stonecash (2007, 2009), Collins (2000, 2007),
Dallek (2003), Edsall and Edsall (1991), Hamby (1992), Kennedy (1999), Page
(1978), Patterson (1994, 1996, 2005), Sundquist (1968, 1983), and Wilentz (2008).
Policy Cleavages in Historical Context 17
Into the breach stepped Franklin Delano Roosevelt, a skilled politician and
inveterate reformer who believed the national government was obligated to
use its power to restrain private interests on behalf of the public good. He
planned to use the levers of federal power to alleviate economic suffering and
stabilize the devastated economy. At the Democratic Party convention that
July, Roosevelt previewed his New Deal philosophy.
2
Recall that all convention quotes are from Woolley and Peters (2011).
18 ON VOTER COMPETENCE
of mass joblessness, soup lines, and “Hoovervilles” did lasting damage to the
Republican brand (Campbell et al., 1960; Lubell, 1956).
In 1964, Barry Goldwater offered America “a choice, not an echo” by
roundly condemning the welfare state. The Senator from Arizona had long
railed against big government and bureaucratic centralization and praised
individual initiative and the free market. Although he backed away from
some earlier controversial comments, such as making Social Security a vol-
untary program, he left no doubt about what he believed. After Johnson
routed Goldwater, neither Nixon nor his successor Gerald Ford challenged the
legitimacy of core New Deal functions. Indeed, although Nixon derided the
Great Society and often clashed with congressional Democrats over specific
programs, his administration oversaw a sizable expansion of social services
(e.g., expanded Social Security benefits) and federal regulatory capacities (e.g.,
the Environmental Protection Agency and Occupational Safety and Health
Administration were launched during his administration). Of course, the driv-
ing forces here were strong Democratic majorities in the House and Senate.
Nixon cared more about foreign policy and was content to sign Democratic
sponsored bills backed by the public.
By the late 1970s a faltering economy, declining trust in government, and a
resurgent business community helped pave the way for a renewed assault on
the welfare state. Campaigning in 1980, Ronald Reagan proclaimed “govern-
ment is not the solution to our problems, government is the problem.” Reagan
argued that cutting taxes and slashing government spending would revital-
ize an economy ravaged by inflation, unemployment, and murderous inter-
est rates. After his victory, Reagan moved quickly to pass his program. He
delivered in the summer of 1981 by signing the Economic Recovery Tax Act
that phased in substantial personal income and business tax cuts over 3 years,
reduced a number of means tested programs for the poor, terminated public
employment programs, and removed hundreds of thousands of beneficiaries
from the Social Security disability program.
Although the administration had some success in scaling back safety net
programs, wholesale restructuring of the welfare state remained elusive.
Nevertheless, Reagan transformed the political landscape (Collins, 2007;
Wilentz, 2008). Before Reagan, the Republican Party had for the most part
accommodated itself to the existence of the welfare state. Republicans might
not have liked it and frequently sought to resist its expansion, but they knew
where public sentiment lay. After the triumph of “Reaganomics,” limited gov-
ernment and lower taxes became defining features of the Republican Party
brand. Efforts by GOP congresses in the mid-1990s to reduce funding for var-
ious safety net programs, the Social Security partial privatization plan offered
22 ON VOTER COMPETENCE
Before 1932 the federal government had been a remote authority with
limited range. It operated the postal system, improved rivers and har-
bors, maintained armed forces on a scale fearsome only to banana repub-
lics, and performed other functions of which the average citizen was
hardly aware. Within a brief time it became an institution that affected
intimately the lives and fortunes of most, if not all, citizens. Measures of
recovery and of reform . . . contributed to this fundamental alteration of
federal activities. (p. 31)
Since the New Deal era, philosophical battles over what the federal gov-
ernment should do have profoundly impacted American politics and soci-
ety. Debates over the Fair Deal, the Great Society, Reaganomics, health care
3
Related questions about the role of government have infused public debate at
different periods in American history. For instance, questions about federal sup-
port for agriculture were paramount during the New Deal era. Likewise, ques-
tions about government regulation of business have arisen at various moments
in time since the 1970s. My point is not that these are trivial controversies. They
clearly matter, especially to those most directly affected by the action in ques-
tion. Instead, my claim is that ordinary citizens are less likely to pay attention to
matters such as agricultural subsidies or deregulation than to the broader ques-
tion of whether the government in Washington has a responsibility to help those
struggling in the market economy. For evidence that the public thinks about gov-
ernment activism along these lines, see Campbell et al. (1960), Jacoby (1994), and
Lewis-Beck et al. (2008).
Policy Cleavages in Historical Context 23
reform, tax versus spending tradeoffs, and entitlement reforms have domi-
nated domestic policy like no other cluster of issues. Other conflicts have mat-
tered during various intervals, including issues such as civil rights for blacks
in the 1960s and busing and affirmative action in the 1970s (I have more to say
about race later in the chapter); however, in terms of salience and persistence
as a fundamental policy cleavage none can approach, let alone match, conflict
over the size and scope of government.
Second, presidential incumbents, those seeking the office, and the national
parties have, generally speaking, taken distinct positions on this dimension.
Sometimes the space between the candidates has been immense, such as 1936,
1964, 1980, and 2008. At other times, the distance has narrowed, such as 1952,
1956, 1960, and 1976. Moreover, some candidates have adopted positions on
discrete issues associated with the opposition party, such as Clinton’s 1992
welfare reform pledge or Bush’s vow in 2000 to seek a Medicare prescrip-
tion drug benefit. Although deviations like these suggest Clinton was less
committed to activist government than Johnson and Bush less conservative
than Goldwater, historical continuities surely matter less to voters than where
the current candidates stand. The bottom line is that although the personali-
ties and politics have changed a good deal from the 1930s to the present, the
Democratic Party’s commitment to marshalling federal power to help those at
the economic margins of society has held steady. And although the GOP has
had its share of centrists in the past, the party has always stood to the right of
Democrats on this continuum (cf. Gerring, 1998).
4
The following works have guided my analysis of cultural issues: Boyer (2001),
Blum (1991), Brewer and Stonecash (2007, 2009), Carter (1995), Collins (2007),
Halberstam (1993), Hunter (1992), Layman (2001), Liebman, Wuthnow, and Guth
(1983), Patterson (1996, 2005), Perlstein (2008), Scammon and Wattenberg (1970),
Sundquist (1983), Wilentz (2008), and Wuthnow (1988).
24 ON VOTER COMPETENCE
issues were inconsequential prior to the 1960s, but rather that their relevance
ascended to a degree hitherto unseen. Moreover, previous moral disputes had
usually been tied to single issues such as abolition, evolution, or temperance,
to name a few, rather than reflecting a broader cultural syndrome.
The Supreme Court fired some early salvos in the culture wars with two
controversial decisions passed down early in the decade. In Engel v. Vitale
(1962) the Court ruled voluntary and nondenominational prayers in public
schools violated the first amendment establishment clause and, hence, were
unconstitutional. The following year the Court barred schools from mandat-
ing Bible study (Abington School District v. Schempp, 1963). Ignoring nearly two
centuries of practice, within a year the Supreme Court had banished God and
religion from the nation’s classrooms. Although hailed by some, these deci-
sions angered the much larger share of the public that believed religion had a
role to play in public schools (Page & Shapiro, 1992).
Conflict in the schools spilled over to the curriculum. One set of skirmishes
centered on the Biblical view of creation. The Supreme Court undercut tra-
ditional viewpoints again by invalidating state laws prohibiting the teaching of
evolution (Epperson v. Arkansas, 1968). Since that time the Supreme Court and
other lower courts have repeatedly barred creationism from public schools.
Battles over sex education represented another front in the curriculum wars.
The topic had been taught in some areas of the country for decades, but the
emphasis had been on biology, chastity, and personal hygiene. In the 1960s
coverage expanded to more controversial topics and new parts of the country.
Previously taboo subjects, such as masturbation, birth control, and premari-
tal sex, now supplemented standard fare. Needless to say these changes pro-
voked resistance in many areas, dissension that continues to this day.
Next, we have the sexual revolution. The development and dissemination
of cheap and reliable oral contraceptives helped transform sexual practice by
giving women more control over their sexual choices and reducing the risks
of unplanned pregnancies. The 1965 Supreme Court decision in Griswold v.
Connecticut, which decreed married couples were entitled to contraceptives,
and the 1972 Eisenstadt v. Baird case, which extended this right to everyone
else, facilitated dissemination. Additionally, discussion of human sexuality
moved into public discourse in ways previously unimaginable. The pioneer-
ing studies of Alfred Kinsey in the late 1940s and early 1950s on male and
female sexual behavior, along with William Masters and Virginia Johnson’s
1963 study on the physiology of sex, catalyzed this process. Mass circulation
publications reflected evolving mores as well. To take a couple of examples,
Playboy magazine, with its photographs of nude women, debuted in 1953 and
grew in popularity over the next two decades. In 1962, Helen Gurley Brown’s
Policy Cleavages in Historical Context 25
book Sex and the Single Girl, which encouraged women to find sexual fulfill-
ment before—or in place of—marriage, was published and sold well. As a
result of these and other factors, public attitudes toward sex liberalized, par-
ticularly in matters of premarital relations and cohabitation. These attitudes,
in turn, reinforced changing practices, especially among the young, many of
whom openly flaunted the conventions of their parents and elders. This is
not to say that most joined the revolution or that traditional courtship rituals
such as “going steady” were abandoned. The difference was that once settled
norms were now competing with newer, more lenient standards.
Sex spread to popular culture. From everyday speech to receding hemlines
to advertising, sex was out in the open like never before. On television, 1950s
family sitcoms such as Leave it to Beaver and Ozzie and Harriet that featured
wholesome middle class families gave way to 1970s shows such as Charlie’s
Angels and Three’s Company that featured attractive women prancing about
and adopting provocative poses on the sets. By the 1990s, programs such as
Sex and the City celebrated casual sex while pornography was available widely
on cable television and the Internet. Crassness permeated other areas of popu-
lar culture as well: music became louder and more offensive, Hollywood fare
increasingly lurid, and standup comedy more profane.
Another departure from convention concerned the changing role of women
in American society. At the beginning of the 1960s, women were expected to
marry relatively young, raise children, and leave careers to their husbands,
in short, to place devotion to family above all else. “The two big steps that
women must take are to help their husbands decide where they are going and
use their pretty heads to help them get there” Mrs. Dale Carnegie intoned in
Better Homes and Gardens in 1955 (Halberstam, 1993, p. 592). Cultural mores
began to change rapidly over the next decade. Betty Friedan’s 1963 book The
Feminine Mystique attacked gender norms for condemning (educated, middle
class) women to lives of domestic tedium and unfulfilled potential. Millions
of women proved receptive to this message. Changing mindsets were accom-
panied by interest group mobilization. In 1966, activists founded the National
Organization for Women (NOW) to advocate for federal action across a range
of women’s issues. Although few women thought of themselves as feminists
or joined such organizations, many sympathized with movement aspirations
and goals. Of course, millions more women, and not a few men, viewed fem-
inism with utter disdain.
Abortion came next. Building on its earlier Griswold decision, which estab-
lished an inherent right to privacy, the Supreme Court in Roe v. Wade (1973)
invalidated state laws criminalizing abortion. This tremendous victory for the
women’s movement evoked disbelief and horror in the minds of those who
26 ON VOTER COMPETENCE
revered the “sanctity of life.” The decision had real consequences as abortion
rates accelerated thereafter before leveling off in the 1980s (Brewer & Stonecash,
2007, p. 97). Since then pro-life and pro-choice forces have clashed at abortion
clinics, at the ballot box, in statehouses, in Congress, and in state and fed-
eral courts. Two notable Supreme Court cases (Webster v. Reproductive Health
Services, 1989 and Planned Parenthood v. Casey, 1992) placed some restrictions
on abortion but preserved a woman’s right to choose. However, in Gonzales
v. Carhart (2007) the court upheld a ban on partial birth abortions, a decision
widely viewed as a setback for abortion rights.
Social change has affected the American family as well. Female labor partic-
ipation, including that of mothers with young children, climbed steadily after
1960 (Brewer & Stonecash, 2007, pp. 99, 102). Although many were untroubled
by this, others despaired at the falling number of stay at home moms and the
growing number of “latchkey” children home alone. The increasing economic
and sexual independence of women along with a rise in no fault divorce laws
contributed to a jump in the divorce rate between 1960 and the mid-1970s
(Brewer & Stonecash, 2007, p. 104). Lastly, society witnessed an ever mount-
ing rise in the number of out-of-wedlock births from the early 1960s onward
(Brewer & Stonecash, 2007, p. 105). These changes left many children in single
parent (almost always female headed) households. For those committed to
traditional family values, such developments were deeply unsettling.
Another threat to old fashioned ways emerged from an unanticipated
direction—that of the homosexual community. Throughout American history,
gays and lesbians had been ridiculed, marginalized, and persecuted. To take a
few examples, same sex relations were criminalized, the American Psychiatric
Association (APA) classified homosexuality as a mental illness, and gays were
barred from military service. A few homophile organizations founded in the
1950s sought to improve matters, but there existed no mass movement work-
ing on behalf of gays and lesbians. Things began to change in the summer of
1969. On June 27 police raided the Stonewall Inn in Greenwich Village, New
York City, a bar with many gay patrons that was subject to constant police
harassment. On this night, gays fought back for the first time and battled cops
for several days. The “Stonewall Riot” awakened group consciousness and
the gay liberation movement was born. Although the movement experienced
some early successes (e.g., gay pride parades were organized in many large
cities and the APA declassified homosexuality as a mental disease), struggles
persisted over antisodomy laws, AIDS, gays in the military, same sex marriage,
and other issues. The marriage issue in particular has been tremendously divi-
sive, especially at the state level where multiple ballot initiatives banning gay
marriage have appeared in recent years.
Policy Cleavages in Historical Context 27
I conclude with the “law and order” issue. On August 6, 1965, President
Johnson signed the Voting Rights Act into law. Many hoped this landmark
achievement, along with the 1964 Civil Rights Act, would mark the begin-
ning of the end of America’s race problem. Five days later, the predominantly
black Watts neighborhood in Los Angeles was engulfed in flames after white
policemen pulled over a black motorist for speeding. A routine moving viola-
tion quickly spiraled out of control, causing a riot that lasted 6 days, claimed
34 lives, and caused 35 million dollars worth of damage (Blum, 1991, p. 253).
Riots erupted in many other cities over the next three summers. Many blacks,
long frustrated by discrimination and segregation, a lack of opportunity, and
police harassment, felt trapped and hopeless. For many, formal legal equality
and the franchise did little to improve their daily lives.
Worries about law and order were not confined to central cities. Unrest
on college campuses fed the belief among some people that the social order
was collapsing. New Left organizations, such as Students for a Democratic
Society, began protesting Johnson’s Vietnam policies in 1964. There were other
grievances. For instance, the 1964 Free Speech Movement at the University of
California, Berkeley sought to expand the rights of political speech on campus.
It garnered much attention as thousands of student protesters clashed openly
with university police. Another notable rebellion occurred in 1968 when stu-
dent radicals at Columbia University occupied several campus buildings for
6 days to decry various university policies before being routed by police.
Although many young people were sympathetic to causes espoused by cam-
pus activists, others in the broader public were mortified as they watched
events unfold on their television screens.
Fear of crime has long outlived the 1960s. Though media coverage often
inflamed public anxiety, such fears were tied to real world events as rates of
murder, violent crime, juvenile delinquency, and drug usage climbed inex-
orably throughout the 1960s and 1970s (Brewer & Stonecash, 2007, p. 107;
Miringoff & Miringoff, 1999, p. 45). In the 1980s, new concerns arose about the
underclass, crack cocaine, gangs, and teenage drug use. Public outrage helped
pave the way for the war on drugs, “three strikes and you’re out” sentencing
guidelines, and an exploding prison population.
To sum up, the far reaching changes unleashed during the 1960s struck at
the heart of traditional morality. The fact that such changes touched on the
most basic and personal notions of right and wrong ensured cultural con-
flict would attract extraordinary attention. Though many who came of age
during these turbulent times embraced novelty or took it in stride, the large
share of the public wedded to conventional moral standards was appalled
by what they saw. Among the most distressed were evangelical Christians
28 ON VOTER COMPETENCE
and committed Roman Catholics. Many members of the middle and working
classes, though not especially pious, were also deeply troubled by the ravages
of modernity. For culturally conservative groups such as these, the disparate
issues were symptomatic of a broader problem—the collapse of moral author-
ity. In their view, America was becoming too liberal, too secular, too permissive.
As the sociologist Jerome Himmelstein (1983, p. 16) describes the syndrome:
“Abortion, the Equal Rights Amendment, busing, affirmative action, sexual
promiscuity, drugs, prohibitions on school prayer, the secular curriculum in
public schools, and many similar things are opposed on the grounds that they
contribute to this process of social breakdown and moral decay.”
Several developments would help pave the way for the eventual politiciza-
tion of these grievances. To begin, the post-World War II era witnessed a sharp
rise in the size of the Protestant evangelical community along with a corre-
sponding decline in mainline Protestant denominations, providing future
leaders with a wellspring of religious conservatives to back demands for
moral restoration. Moreover, differences in the educational and social bases of
Protestant and Catholic denominations waned, thereby enhancing the pros-
pects of interfaith cooperation among conservative Christians on the right or
religious liberals on the left. A 1978 Internal Revenue Service (IRS) proposal to
strip the tax exempt status of private Christian schools that failed to meet its
standard of racial integration ignited a firestorm of protest within the evangel-
ical community. Finally, skilled leadership channeled moral indignation into
a potent political force.
To build on this last point, charismatic preachers used local pulpits, tel-
evision ministries, and radio programs to rail against the heresy of liberal
Christian denominations and secular humanism more broadly. Jerry Falwell,
Pat Robertson, James Dobson, and others lamented America’s cultural decline
and began organizing in the mid-1970s. Political operatives played a key role
in the rise of the New Christian or Religious Right. Conservative activists such
as Paul Weyrich and Richard Viguerie, who had been involved in secular pol-
itics, wanted a means to push the Republican Party to the right. Recognizing
that previously apolitical evangelicals and fundamentalists could help serve
this goal, they persuaded religious leaders such as Falwell to get involved in
politics. By the end of the 1970s, the Religious Right had coalesced into a loose
collection of organizations capable of rallying evangelicals. Organizations such
as the Moral Majority, the Religious Roundtable, the Christian Voice, Focus on
the Family, the Family Research Council, and then, in the 1980s and 1990s,
the Christian Coalition, sought to harness devotees’ anger—and dollars—
to support public officials who shared their views. Other groups mobilized
around a particular cause, such as the Concerned Women for America, Phyllis
Policy Cleavages in Historical Context 29
Schlafly’s Stop ERA and Eagle Forum, Randall Terry’s Operation Rescue, and
various antigay and antipornography organizations. Though airing different
complaints, they were united by a transcendent concern with moral decline.
This burgeoning discontent provided strategic politicians with an oppor-
tunity to reap success at the ballot box by capitalizing on collective anxieties
about social change. George Wallace was the first to campaign on the “social
issue” during his long shot bid for the 1964 Democratic Party nomination.
During the general election campaign that fall Barry Goldwater also lamented
moral decline. For both men this strategy proved futile against the Johnson
juggernaut. Two years later, however, Ronald Reagan’s decisive victory over
incumbent Pat Brown in the 1966 California gubernatorial race suggested
cultural backlash could yield electoral success. Taking notice, Richard Nixon
aped Reagan in 1968 by campaigning on law and order, time honored values,
and moral decency in order to court support from what he would later call the
“silent majority.”
Throughout his term in office and during the 1972 campaign, Nixon posi-
tioned himself as the champion of “Middle America” and the traditional
values it subscribed to. That year Democratic challenger George McGovern
played the foil by taking left of center positions on new cultural issues such as
busing, amnesty for Vietnam draft evaders, and penalties for marijuana use.
The Democrats’ commitment to cultural liberalism was on full display at their
national convention that July. As presidential campaign chronicler Theodore
White described it (1973):
One could . . . watch the parade of women across the podium . . . forc-
ing into politics matters never before publicly discussed at a national con-
vention—for example, the laws of sex . . . Then came homosexuals to the
microphone and camera, men openly demanding before the nation that
the coupling of males be accepted not furtively, but as a natural and legal
right. (p. 180)
candidate atop the ticket might prove attractive to some Democrats (e.g.,
working class Roman Catholics, southern evangelicals). That candidate was
Ronald Reagan. Given the Religious Right’s organizational prowess and poten-
tial to mobilize votes, Reagan eagerly courted their support. At the Religious
Roundtable’s annual meeting in Dallas that summer, Reagan delighted his
audience by saying “You may not endorse me, but I endorse you” (Wilentz,
2008, p. 123). New Christian Right leaders let the faithful know Reagan was
their man in 1980 and again in 1984. Although he delivered little in the way of
tangible results, President Reagan offered symbolic and rhetorical support to
those who revered traditional and religious values. Conservative Christians
and other Middle Americans moved steadily into the GOP fold throughout
the 1980s (Greeley & Hout, 2006; Layman, 2001).
Cultural divisions have intensified since then. Recall Patrick Buchanan’s
clarion call for a relgious war at the 1992 Republican Party convention. Other
examples of partisan cultural conflict spring to mind, such as Republican
attacks on Clinton’s moral character that culminated in his 1998 impeachment;
the divisive battles over gay rights during the Clinton and Bush II years at
the national and state level; renewed struggles over the teaching of creation-
ism, rebranded as intelligent design, in public schools; and the intractable and
sometimes deadly fights over abortion.
Having reviewed social and political developments in the cultural issues
domain over the past 40 years, I now summarize the takeaway points. Moral
conflict moved into the public sphere in the 1960s and has remained deeply
entrenched ever since. Although the battles have often focused on concrete
issues such as school prayer, feminism, abortion, gay rights, and gun control,
the broader battle has always been about what set of moral standards should
govern the public and private life of the nation (Hunter, 1992). As the historian
Robert Collins (2007, p. 173) has put it, this “clash between the older ortho-
doxy of Christian values and the beleaguered but still dominant bourgeois
order on the one hand and the new radical cultural forces on the other consti-
tuted the central battle of the culture war.”
As the tides of change swept across American society and advocates of
cultural liberalism mobilized to press a disparate set of demands, backlash
emerged from the right as cultural and religious conservatives entered poli-
tics to protect traditional rules of moral behavior. GOP officials, on the lookout
for new issues to blunt the Democratic advantage in the economic welfare
domain, moved quickly to exploit the opportunities afforded by this backlash.
Throughout his first term Nixon spoke for those who feared or resented social
change by paying tribute to traditional family values, while the Democratic
Party moved left on moral standards, especially during the 1972 campaign.
Policy Cleavages in Historical Context 31
Although the parties did not polarize immediately on these matters, the seeds
of future conflict were planted by the mid-1970s. With the election of Reagan
in 1980 polarization began before accelerating once Clinton moved into the
Oval Office 12 years later. The process continued during the Bush presidency
of 2001–2009. In this manner, apparently disconnected battles over a plethora
of moral issues have evolved into a deep-seated cleavage dividing the parties
and their presidential standard bearers.
5
These works guided my review of foreign policy history: Addington (2000),
Aitken (1993), Berinsky (2009), Blum (1991), Collins (2007), Dallek (2003), Gaddis
(2005), Halberstam (2001), Herring (2002, 2008), Oberdofer (1998), Packer (2005),
Patterson (1996, 2005), Perlstein (2008), Ricks (2006), Sundquist (1983), Suskind
(2006), and Wilentz (2008).
32 ON VOTER COMPETENCE
of course. Disputes predictably arose over which party could better protect
the nation. For instance, after Chinese forces entered the Korean War in 1950,
some congressional Republicans backed Douglas MacArthur’s call to expand
the war to the Chinese mainland. Likewise, during the 1952 campaign the
GOP tarred Democrats as soft on communism. Eight years later John Kennedy
returned fire by promising to fix a “missile gap” that allegedly developed on
Eisenhower’s watch. Without gainsaying these differences, the important
point to remember is that politicians from both sides of the aisle were commit-
ted to maintaining American power and prestige via military strength.
The first strains in the consensus surfaced during the Vietnam War era. In
1964, U.S. intelligence indicated that absent sustained U.S. support the gov-
ernment of South Vietnam would fall to communist insurgents, known as the
National Liberation Front (NLF), who were backed by Ho Chi Minh’s commu-
nist government in North Vietnam. Recalling what “losing China” had cost
Truman, Johnson feared disengaging in Vietnam would render him vulner-
able to blistering criticism from the right. But he knew that an intensifying
military conflict during an election year would be politically toxic as well. So
he campaigned as a man of peace, promising the American public “we are not
about to send American boys nine or ten thousand miles away from home
to do what Asian boys ought to be doing for themselves” (Goldberg, 1995,
p. 231). Taking no chances, he further implied GOP opponent Barry Goldwater
was an erratic warmonger not to be trusted with his finger on the nuclear
button.
After the election the choice confronting the administration was whether
to escalate or negotiate a settlement with the communists, which would
effectively doom the government of South Vietnam. Johnson chose escala-
tion, in large part because he feared letting South Vietnam fall would under-
cut domestic support at home and thereby cripple his ability to construct his
Great Society. From early 1965 to the end of 1967, the U.S. troop presence
shot up from 23,000 to 485,000 (Herring, 2002, p. 182). Despite this rise and
a sustained bombing campaign against enemy positions, the NLF and North
Vietnam matched U.S. escalation at every turn. Seeking to allay the concerns
of an increasingly anxious public, the Johnson administration launched a
public relations offensive in November 1967. Vietnam Commander William
Westmoreland returned home to shore up public support for the war, telling
reporters “We are making real progress” and “We have reached an important
point where the end begins to come into view” (Herring, 2002, p. 221). Vice
president Hubert Humphrey toed the administration line a few days later:
“there has been progress on every front in Vietnam. . . . There is no military
stalemate” (DeGroot, 2008, p. 282).
Policy Cleavages in Historical Context 33
This sense of optimism was shattered 3 months later on January 31, 1968
when the communists launched the Tet Offensive in scores of urban areas
throughout South Vietnam thought to be immune from danger, the most
notable example of which was the assault on the U.S. embassy in Saigon.
Overcoming early setbacks in the field, U.S. and South Vietnamese regulars
regrouped to inflict punishing losses on enemy personnel. Although suf-
fering a bruising military defeat, communist forces scored a major political
victory by helping tip U.S. mass and elite opinion against the war. During
the February 27 broadcast of the CBS evening news, venerable anchorman
Walter Cronkite concluded “we are mired in a stalemate.” Public support for
Johnson’s handling of the war dropped precipitously (Page & Shapiro, 1992,
pp. 57, 232–233). Several weeks later in a nationally televised speech on March
31, Johnson announced a bombing halt, reaffirmed his commitment to peace,
and stunned the nation by withdrawing from the presidential race. On April
9 the administration announced a troop ceiling and the gradual transfer of
responsibility for waging war to South Vietnamese forces.
The war fomented deep divisions in American society. Antiwar dissent
was confined initially to college campuses but eventually spread to other
sectors of society. A number of clergy spoke out against Vietnam, Martin
Luther King, Jr. most notably. Celebrities criticized the war as well, such as
the pediatrician Dr. Benjamin Spock, the actress Jane Fonda, and heavyweight
boxing champion Muhammad Ali. Mass demonstrations, including a 1967
march on the Pentagon, gave vent to public frustration with the war. Johnson
probably could have weathered such criticism, but when dissent spread to
the Democratic Party establishment his position became increasingly ten-
uous. In 1966 Senator William Fulbright, in his capacity as chairman of the
U.S. Senate Foreign Relations Committee, held televised hearings critical of
the war. Adding to Johnson’s woes, Senator Robert Kennedy, a potential rival
for the 1968 nomination, later called for a negotiated settlement with Ho Chi
Minh’s government. Finally, during the 1968 New Hampshire primary, peace
candidate Eugene McCarthy shocked the country by capturing 42% of the
vote (compared to the 49% write in vote for President Johnson). At its chaotic
national convention in Chicago that August, the Democrats’ hawk and dove
factions clashed openly on the convention floor. The postwar consensus in
favor of military power was collapsing within the Democratic Party.
After the election, Vietnam became Nixon’s war. Adopting his predeces-
sor’s “Vietnamization” policy as his own, the new commander in chief sought
to transfer the combat burden from American to South Vietnamese troops
while increasing U.S. aid to its ally. Given the likelihood that U.S. withdrawal
might occur before South Vietnamese forces were ready to take up the fight,
34 ON VOTER COMPETENCE
Questions about the utility of military power remained after Nixon resigned
in disgrace in 1974, but party positions would not crystallize fully until the
early 1980s. Gerald Ford initially supported the Nixon and Kissinger policy
of détente. However, hard liners in both parties attacked the policy, while
relations with the Soviet Union deteriorated throughout the 1970s. Despite
the increasing tensions, arms control talks continued and ultimately culmi-
nated in the 1979 SALT II treaty, which Jimmy Carter and Brezhnev signed in
June. But by the end of the year, NATO had consented to the deployment of
new missiles in West Europe and the Soviet Union had invaded Afghanistan.
Carter withdrew SALT II from Senate consideration, imposed a grain embargo
against the Soviets, and called for rapid defense spending increases. Détente
was dead. In addition to inflamed Cold War tensions, the Iranian hostage cri-
sis fed public perceptions of U.S. vulnerability in an increasingly dangerous
and unstable world.
During the 1980 presidential campaign, Ronald Reagan attacked Carter for
being weak on national security and promised to rebuild American military
strength and restore U.S. prestige in the world. Throughout much of his first
term, Reagan damned the Soviets. For example, in his first press conference he
condemned Soviet leaders for reserving “unto themselves the right to commit
any crime: to lie, to cheat” and in a 1983 speech called the U.S.S.R. an “evil
empire” (Collins, 2007, p. 195; Patterson, 2005, p. 194). Reagan sought “peace
through strength” by overseeing the largest peacetime military buildup in
U.S. history. NATO’s 1983 deployment of the next generation of cruise mis-
siles on West European soil aroused intense opposition from the Soviets and
revitalized the peace movement. The administration supported a number of
anticommunist movements throughout the world, including the contras in
Central America and the mujahedeen in Afghanistan. U.S. military force was
deployed in the 1983 invasion of Grenada and the 1986 bombing of Libya.
Finally, Reagan’s support for the controversial Strategic Defense Initiative
(SDI), a space-based defensive shield against intercontinental missiles, accen-
tuated the president’s commitment to military strength (defensive in this
case). Of course, Reagan saw SDI as a means to transcend nuclear conflict, but
that position was not always clear to the public at the time. All these signs left
little doubt where Reagan, and by extension the GOP, fell on the hawk–dove
continuum. Consistent Democratic opposition to the Reagan defense posture
helped clarify party positions on this dimension.
Moving beyond the military buildup and saber rattling of his first term,
Reagan pursued a different tact after reelection. Reagan and Soviet leader
Mikhail Gorbachev held summits in Geneva in 1985, Reykjavik in 1986,
Washington in 1987, and Moscow in 1988. Reagan, who was genuinely
36 ON VOTER COMPETENCE
issues played little role in his campaign message. In office Clinton sought
to capitalize on a “peace dividend” by calling for defense spending cuts to
free up resources for domestic priorities. But, as always, foreign policy cri-
ses demanded presidential attention. Although U.S. national security was not
seriously threatened at any point during the decade, Clinton used military
force on a number of occasions, foremost in humanitarian and peacekeeping
efforts in Somalia and the Balkans and air and missile strikes against targets in
the Balkans, Iraq, Afghanistan, and the Sudan. In any case, because the Cold
War was over and no ground wars occurred on Clinton’s watch, national secu-
rity took a back seat to domestic policy throughout the 1990s.
Things changed dramatically after the terrorist attacks on September 11,
2001 killed nearly 3000 innocents on American soil. Osama bin Laden, the
leader of the al Qaeda terrorist organization based in Afghanistan, claimed
responsibility for the attacks. President George W. Bush promised swift ret-
ribution against al Qaeda and those who harbored its members. When the
Taliban movement refused to surrender the terrorist organization’s leaders,
and meet several additional demands, war quickly ensued. Bombing began
in early October. Six weeks later the Taliban were driven from power and
al Qaeda’s forces scattered along Afghanistan’s mountainous eastern border
with Pakistan.
The war in Afghanistan was the first battle in a broader “global war on
terror” in which the Bush administration promised a new approach to defend-
ing U.S. security. The full meaning of this became clear during the president’s
commencement address at West Point in June 2002 where he unveiled what
came to be known as the Bush Doctrine. During the speech the president
asserted the right of the United States to initiate preventive war: “We must
take the battle to the enemy, disrupt his plans, and confront the worst threats
before they emerge” (Ricks, 2006, p. 38). Discarding the decades old practice
of deterrence, Bush proposed the most aggressively militant foreign policy
in American history. Throughout the summer and fall top administration
officials conflated al Qaeda, the 9–11 terrorist attacks, and Saddam Hussein,
implying the Iraqi dictator posed an immediate threat to the United States.
The most frightening charge was that Hussein might furnish terrorists with
weapons of mass destruction to attack targets on American soil. In a speech
that October the president laid out the case for taking action: “America must
not ignore the threat gathering against us. Facing clear evidence of peril, we
cannot wait for the final proof, the smoking gun that could come in the form
of a mushroom cloud” (Ricks, 2006, p. 61). Although some diplomatic efforts
were undertaken, there seemed little doubt the administration was preparing
for war.
38 ON VOTER COMPETENCE
After efforts to peacefully resolve the crisis failed, combat began on March
20, 2003. A combination of highly effective precision bombing and a lighten-
ing quick ground invasion decimated Iraqi resistance in short order. In a May
1 televised address, Bush triumphantly—and prematurely—declared that
“Major combat operations in Iraq have ended. In the battle of Iraq, the United
States and our allies have prevailed” (Ricks, 2006, p. 145). As it turned out, the
real war was just beginning. A series of U.S. mistakes, including the failure to
develop an adequate postwar plan to secure the country; the de-Baathification
order in which some 35,000 Sunnis, many of whom had committed no crimes,
lost their jobs; the dissolution of the Iraqi army, which left hundreds of thou-
sands of armed Iraqi men unemployed and angry; and the decision to indefi-
nitely table the creation of an interim Iraqi government, ignited the insurgency.
From the summer of 2003 to the spring of 2007 the war went poorly. By the
time the president made the gutsy decision to back a “surge” strategy in early
2007 much blood and treasure had been squandered. During the 2008 election
Barack Obama attacked the Bush administration for its “misguided war” and
promised to bring it to an end. GOP rival John McCain defended his support
for the war, if not the initial war plan, and insisted the United States had to
finish the job. From 2003 to 2008, then, there was little ambiguity about which
party “owned” Iraq.
This completes my historical review. To reiterate, my purpose in this sec-
tion has been to provide a synopsis of major developments in the national
security domain in the post-World War II era. The fundamental issue driving
debate in this domain has centered on the role of military power. From the late
1940s to the mid-1960s bipartisan consensus backed the idea, codified initially
in Truman’s containment policy, that a strong military and the will to use it
were essential for protecting U.S. national security and geopolitical interests.
But when victory in Vietnam could not be achieved at an acceptable cost, elite
consensus began to unravel in the Democratic Party. For the first time, estab-
lishment politicians questioned whether hard power advanced or undermined
America’s foreign policy goals. After 1968, Vietnam became Nixon’s war and
the parties began to polarize on the conflict with Democrats becoming more
dovish and Republicans increasingly hawkish (Berinsky, 2009). Yet American
involvement in Vietnam ended in 1973 and normalization of relations with
China and détente with the Soviets belied claims that Nixon was an ardent
hawk.
If the Nixon–Kissinger–Ford commitment to détente temporarily clouded
the picture of where the parties stood on the hawk–dove dimension, Reagan’s
election clarified matters. Reagan’s first term mixed the largest peacetime
expansion of military power in American history, bellicose anti-Soviet rhetoric,
Policy Cleavages in Historical Context 39
varying degrees of support for anticommunist forces across the globe, and the
commitment to expand nuclear weapons into space, leaving little doubt about
where the president and his party stood. And though Reagan used diplomacy
to reduce U.S. and Soviet tensions during his second term, the parties quickly
reverted to form when the 1988 election rolled around as Bush hit Dukakis
for being too soft on national security. The first and second Bush presidencies
included a major military operation in Panama, two wars with Iraq, a war in
Afghanistan, and a global war on terror. Coupled with consistent reluctance
by Democrats to deploy military force—especially boots on the ground—
throughout much of this period, party positions on this cleavage were plain to
see. In short, the military power question has been salient for much of the past
60 years and the parties have, generally speaking, taken opposing positions on
this dimension since the early 1980s.6
6
The Vietnam War led some to question not only the efficacy of force, but the
purpose of U.S. foreign policy more broadly, especially in the post-Cold War era
when the issues of humanitarian intervention and nation building arose dur-
ing the 1990s (Halberstam, 2001). Although such questions were of paramount
importance to U.S. policymakers and the foreign policy elite, the discussion
probably eluded most in the mass public who continued, as always, to care pri-
marily about U.S. security (Page & Bouton, 2006).
40 ON VOTER COMPETENCE
offend southern sensibilities and did little to advance the cause of civil rights.
Thereafter, the civil rights movement pressured the national government to
move toward full racial equality. Truman, Kennedy, albeit reluctantly, and
Johnson, with great courage and conviction, employed a combination of
moral persuasion, executive orders, and legislative pressure to ultimately dis-
mantle the Jim Crow system of discrimination and segregation and secure the
franchise for blacks. Democrats paid a steep political price for doing the right
thing. Commenting on passage of the Voting Rights Act, Johnson despaired to
an aid: “I think we’ve just handed the South over to the Republican Party for
the rest of our lives” (Halberstam, 2001, p. 61). He was right. The GOP became
the party of racial conservatism in order to enhance its appeal among whites
in the south and elsewhere. Although it would be going too far to claim such
appeal was based primarily on racial considerations, inasmuch as the party’s
stance on social welfare, cultural, and national security issues attracted con-
servative support, race was an important factor (Abramowitz, 1994; Carmines
& Stimson, 1989). As the party grew in strength in the Sunbelt region and its
ranks of moderates thinned, Republican support rose among whites and all
but disappeared among blacks. Hence, with respect to party coalitions the
centrality of race is beyond dispute. Indeed, if we define cleavages in terms
of social groups rather than policy dimensions, as I do here, race is probably
the preeminent division in American politics (Abramson, Aldrich, & Rohde,
2009).
In terms of issues, racial controversies have sometimes generated divisive
public argument since the mid-1960s. For instance, conflict flared over forced
busing in the 1970s and affirmative action in the late 1970s and mid-1990s.
Periodic battles over the reauthorization of the Voting Rights Act and federal
efforts to ensure equal opportunity and fair treatment have arisen at various
intervals. Immigration has proven especially nettlesome in recent years. These
significant problems affect the lives of millions of people. But given the rela-
tively short lifespan of these issues, it seems fair to conclude that no durable
racial policy cleavage exists along the lines of strong versus limited govern-
ment, traditional standards of moral conduct, and the utility of military force.
Put otherwise, these three cleavages have impacted presidential politics and
the party system to a much greater extent than battles over affirmative action,
minority set asides, or, after 1965, voter registration.
Although overt racial conflict has not dominated the American political
scene since the 1960s, implicit racial issues have mattered a great deal. Racially
conservative politicians rarely make explicit antiblack appeals because doing
so violates the norm of racial egalitarianism endorsed by most whites, and
thus risks provoking backlash at the ballot box and, more speculatively,
Policy Cleavages in Historical Context 41
CONCLUSIONS
In this chapter I provided a historical sketch of the origin and evolution of the
bedrock cleavages in the economic welfare, cultural issues, and foreign policy
domains. For each, I suggested that particular controversies at given points
in time should be viewed as manifestations of a deeper underlying conflict.
Economic welfare issues, such as poor relief, unemployment insurance, Social
Security and Medicare, food stamps, and health care reform, reduce to phil-
osophical differences over what the government in Washington should do to
help those struggling in the market economy. In the cultural domain, issues
such as school prayer, abortion, gay marriage, pornography, and stem cell
research are linked to broader concerns about the preservation of traditional
norms of moral behavior. Finally, national security issues, such as U.S.–Soviet
relations, rapprochement with China, shooting wars in Korea, Vietnam, Iraq,
and Afghanistan, smaller military operations in places such as Grenada and
Panama, and missile strikes against terrorist camps and state sponsors of ter-
rorism, reflect the more basic issue of the role military strength should play in
advancing U.S. foreign policy objectives.
This chapter’s second takeaway point is that each policy cleavage has occu-
pied a central position on the national political agenda for a very long time.
Battles over the size and scope of the welfare state date back to the 1930s.
Concerns about moral decline and restoration have shaken American society
42 ON VOTER COMPETENCE
since the 1960s. Finally, the hard versus soft power tradeoff has structured for-
eign policy discourse since the Cold War began in the late 1940s. This is not to
say all other issues are irrelevant. Some disputes have generated passionate
debate at various moments in time, such as civil rights for African Americans
in the 1950s and 1960s or clashes on immigration over the past several years.
In my judgment, no other policy cleavage can match the visibility and dura-
bility of the three identified here.
The final point I stress is that each policy cleavage has become deeply
entwined with presidential politics and the party system. The major parties
and their standard bearers have taken contrasting positions on government
activism since the New Deal, on traditional morality since the rise of the coun-
terculture, and on the hawk–dove continuum since the early years of the
Reagan revolution. My reading of the historical record is that Democrats are
the party of activist government whereas Republicans stand for moral tradi-
tion and military strength. Public opinion research shows that citizens stereo-
type the parties and their presidential candidates precisely along these lines
(Campbell et al., 1960; Green, Palmquist, and Schickler, 2002; Hayes, 2005;
Lewis-Beck et al., 2008; Petrocik, 1996).
Of course, the historical record is more complicated than this so far as
a party’s presidential candidates do not always fall on the same point on a
given policy dimension. The economic welfare reform programs advanced by
Roosevelt and Johnson were far more comprehensive than anything proposed
by Clinton and Obama. Likewise, Eisenhower and Nixon did not contest the
legitimacy of the New Deal like Goldwater and Reagan. Be that as it may, it
seems safe to presume that when marking ballots voters rarely dwell on his-
torical comparisons within parties. When the time to decide arrives, the com-
parison that matters is where the nominees on the ballot before them fall on
a given cleavage. In 2008, the fact that Obama’s domestic policy agenda was
not as ambitious as Roosevelt’s surely meant little to most people. What mat-
tered is that a vote for Obama was a vote for more government than the coun-
try would get under McCain.
To conclude, I have argued that conflict in the primary issue areas on the
American political agenda reduces to more basic dimensions, that each dimen-
sion has been plain to see for decades, and that the parties and their lead-
ers have taken clear and consistent positions opposing one another on each
dimension. The genesis and evolution of policy cleavages have implications
for how citizens come to understand the political world and render electoral
choices. In the next two chapters, I define policy principles as they exist in the
belief systems of everyday people and explain how citizens acquire and sub-
sequently use them to inform their presidential votes.
CHAPTER 3
43
44 ON VOTER COMPETENCE
examples, someone may dislike Barack Obama a bit, feel very positively
toward the ideas of grace and self-reliance, and find conservatism more appeal-
ing than liberalism. The key point, again, is that evaluations of the object are
expressed in summary terms.
Finally, we have the “tendency” attribute of the definition. To understand
what this means, it helps to think about how attitudes form. The process often
begins with conscious perception of some entity (implicit attitudes matter too,
but I do not consider these here). Upon detection, people decide how they feel
about it. The first time this happens, a mental residue is stored in long-term
memory. This “evaluative knowledge” predisposes the individual to react to
the entity in a similar fashion when it is encountered again. The more often
this process repeats, the stronger the object evaluation association becomes
(Fazio, 2007). For an attitude object that has been considered many times, the
net result is an enduring disposition to evaluate the entity in a consistent man-
ner. For other entities, typically obscure topics, no prior evaluative tendencies
exist. Instead, when the person perceives the object he or she will need to con-
struct a temporary judgment. If the object is beheld no more, this evaluative
knowledge fades quickly from memory. It is as if the attitude never existed.
And if per chance the object is encountered again after the attitude has disap-
peared, the object evaluation association must be constructed anew. In short,
the tendency attribute in the Eagly and Chaiken model accommodates both
attitudes as dispositions (or crystallized attitudes) and attitudes as construc-
tions (or uncrystallized attitudes).
To recap, an attitude represents a psychological tendency to evaluate a
given entity with some degree of favor or disfavor. The entity can refer to any-
thing perceived, global evaluations can be based on some absolute standard
or relative to another object, evaluations can range from extremely positive to
extremely negative, and the tendency can vary from immovable to nonexis-
tent. The Eagly and Chaiken definition has much value because it incorporates
attitudes as dispositions, attitudes as temporary constructions, and attitudes
lying between these extremes. As we shall see in the next section, this inclu-
sive definition applies readily to the study of policy attitudes.
POLICY ATTITUDES
I now use the Eagly and Chaiken framework to define liberal–conservative
attitudes, issue attitudes, and attitudes toward policy principles. Each con-
cept represents a psychological tendency to evaluate a given political entity
with some measure of liking or disliking. What distinguishes them is the
entity being evaluated. Standard terminology in political science uses the
46 ON VOTER COMPETENCE
Liberal–Conservative Ideology
The study of ideology is complicated for two reasons. First, scholars often con-
flate liberal–conservative attitudes and liberal–conservative ideology. The two
are not the same. To avoid confusion, I elaborate what liberal–conservative
ideology means before turning to liberal–conservative attitudes. The second
factor muddying the analysis of liberal–conservative attitudes lies in the great
variety of meanings people attach to ideological labels. Hence, care must be
taken in specifying what these terms signify to different perceivers. I do that
after ideology has been defined.
At the broadest level, ideologies are best described as cognitive networks
of attitudes and beliefs organized by superordinate principles in an area
of human activity such as religion, economics, or politics (Kerlinger, 1984).
In the political domain, ideology can be defined as “a particularly elabo-
rate, close-woven, and far-ranging structure of attitudes” organized by the
“liberal-conservative continuum—the ‘right’ and the ‘left’ of a political spec-
trum” (Campbell et al., 1960, pp. 192–193). Although other capstone principles
might bring some coherence to political attitudes, the liberal–conservative
continuum has monopolized scholarly attention because it represents the key
dimension dividing political elites (McCarty, Poole, & Rosenthal, 2006). This
continuum centers on the role the national government should play in the
economic, social, and political life of the country. Per Campbell et al. (1960):
“Differences between liberal and conservative tend to focus upon the degree
to which the government should assume interest, responsibility, and control
over these sectors of endeavor” (p. 194). By this account, citizens’ views on
various political controversies, their preferences on multiple issues, and their
evaluations of elected officials, political candidates, and the parties are encap-
sulated by points on the liberal–conservative continuum.
To provide context for assessing what ordinary citizens know about the
continuum, I elaborate what liberalism and conservatism mean to those versed
in the argot of ideology. To delineate liberal and conservative doctrines, I draw
on Kerlinger (1984) and McClosky and Zaller (1984). To begin, contemporary
liberal philosophy maintains that public officials should harness the power
of the federal government to improve conditions for people in general, and
Policy Attitudes and Political Sophistication 47
Political elites use ideological labels along these lines. But what about non-
elites whose busy lives leave little time for politics? Do they fit this profile?
Scholars have utilized multiple techniques to see if survey respondents
treat liberal and conservative labels as political shorthand. One common
48 ON VOTER COMPETENCE
many southern conservatives and the GOP had a fair share of moderates and
a sprinkling of liberals; and when Lyndon Johnson and Sam Rayburn ran the
Senate and House, respectively, from the political center, this point rings true.
In a more ideologically charged climate, citizens might become more adept at
liberal–conservative thinking.
Following this intuition, Hans-Dieter Klingemann used 1968 NES data
to calculate the proportion of ideologues in the public. Given the Johnson–
Goldwater election in 1964, civil rights and Great Society legislation, escala-
tion and quagmire in Vietnam, and cultural change and social unrest at home,
the possibility that the number of ideologues rose seemed plausible. A rise
did occur, but a paltry one at that. The percentage of ideologues increased
from 2.5% in 1956 to 6% in 1968 (as cited in Converse, 1975, p. 102). Although
deployment of ideological labels was over twice as high by 1968, it remained
confined to a sliver of the electorate.
Given the time it takes for information in elite political discourse to diffuse
throughout the general public, it may have been asking too much to expect
ideological conceptualization to have increased substantially between 1956
and 1968. A much longer timeframe may be needed before major advances
in ideological thinking can be detected. Two long-term changes seem espe-
cially likely to have facilitated a rise in the share of ideologues. First, levels
of education have exploded over the past several decades. According to U.S.
census data, 7.7% of the public aged 25 and older held a 4-year college degree
in 1960 compared to 25.5% in 2000 (U.S. Census Bureau, 2011). Since education
promotes ideological thinking, we would expect the ideologues’ share of the
electorate to have risen over time.
In addition, the Democratic and Republican parties have polarized sharply
since the 1970s. The rightward shift of the GOP, the decimation of the Boll
Weevil Democrats, and the virtual disappearance of moderates from both
sides of the aisle have pushed the national parties further apart on the lib-
eral–conservative continuum than at any time since the Gilded Age (McCarty
et al., 2006). Given ideological polarization at the elite level, perhaps many
citizens followed suit by adopting liberal–conservative mindsets. In short, the
American public is more educated than ever before and the ideological cues
it receives from public officials have rarely been clearer. In conjunction, these
factors should have augmented ideological thinking well beyond the levels
observed during the Eisenhower administration.
Lewis-Beck and his colleagues (2008) used NES data to estimate the propor-
tion of ideologues in the public during the 2000 election season. As in previous
efforts, lenient criteria were used: “Generally speaking, we gave respondents
the benefit of the doubt when placing them in the levels of conceptualization.
50 ON VOTER COMPETENCE
The result is that many of our ideologues definitely do not possess a fully
articulated personal political philosophy” (p. 264). The authors report that
only 10.5% of the electorate made it into the ideologue category. Now, com-
pared to the 2.5% figure in 1956, 10.5% is demonstrably better. But although
this fourfold increase is notable in relative terms, it fails to impress in absolute
terms. What we see is a modest uptick rather than a fundamental change in
the character of the electorate. Dramatically higher levels of education cou-
pled with elite polarization over the past 40 plus years have coincided with
incremental growth in the portion of the electorate that actively thinks about
politics in liberal–conservative terms. To sum up, my first cut at the evidence
suggests few citizens use liberal–conservative principles to structure their
views about the candidates and parties. Even those in the most ideological
category lack a sophisticated understanding of what these terms mean. Recall
that the comment “the Democratic Party tends to favor socialized medicine”
earns classification as an ideologue.1
Though used by many scholars, the levels of conceptualization index
has not escaped criticism (Smith, 1989). Hence, it would be foolish to rely
1
The next level in the conceptualization index contains “near ideologues.”
Respondents in this category did not seem to explicitly recognize liberal–
conservative terminology but evaluated the candidates and parties using other
abstractions that hinted, however remotely, at the possibility of ideological
thought. To take one example from The American Voter, an Ohio respondent liked
Republicans because “they play up to individual rights, which is good. That’s
good—it makes a person feel more independent.” The respondent disliked the
GOP because “They believe in big industry, utilities, etc. They’ve passed a lot of
labor bills I don’t approve of.” Next, a New York respondent disliked Democrats
because “The Communists linked to Roosevelt and Truman. Corruption. Tax
scandals. I don’t like any of those things” and approved of Republicans because “I
also like the conservative element in the Republican Party” (Campbell et al., 1960,
pp. 230–232). Campbell and his coauthors found that 9% of the sample qualified as
near ideologues in 1956. Lewis-Beck and colleagues (2008) report a comparable fig-
ure for the 2000 electorate. Scholars who use the index often lump ideologues and
near ideologues into a single category. Reports put this figure in the 20–25% range
(Pierce & Hagner, 1982). This coding procedure is helpful for some purposes, but it
has had the unintended consequence of conveying the misleading impression that
20% or more of the public thinks about politics in liberal–conservative terms. This
is not so. Instead, 20% or so of the electorate has a few clues regarding the symbols
associated with the labels and applies these to the candidates and parties in very
simple ways. This is something, but it is definitely not a “particularly elaborate,
close-woven and far-ranging structure of attitudes.”
Policy Attitudes and Political Sophistication 51
Liberal–Conservative Attitudes
Although few people possess political ideologies, many more could hold
liberal–conservative attitudes. Surveys routinely show that when asked to
describe their political views on a standard liberal–conservative self-place-
ment scale, two-thirds or more of the public obliges. This begs the question of
what the labels mean to these respondents. Because the labels reflect neither
abstract political beliefs nor policy preference summaries, self-placements
must reflect something else. But what exactly?
Policy Attitudes and Political Sophistication 53
2
Liberal–conservative attitudes are embedded within liberal–conservative ide-
ologies, but they are not the same thing. Hence, the application of the Eagly and
Chaiken framework to liberal–conservative attitudes dispenses with the need to
think about these in terms of organized preferences.
54 ON VOTER COMPETENCE
Issue Attitudes
Scholars have focused a lot of attention on standalone issue preferences,
which represent the second face of policy attitudes animating electoral behav-
ior research. I define an issue attitude as a psychological tendency to evaluate
a specific policy proposal or rival proposals with some degree of positive or
negative affect. A policy proposal denotes an idea about an explicit course of
action the federal government could take to address some identifiable prob-
lem. The issue may center on something unique to a given campaign, such
as the death penalty in 1988, health care reform in 1992 or 2008, or the par-
tial privatization of Social Security in 2004. Other issues recur over a series
of elections, such as abortion and tax cuts. In contrast to liberal–conservative
attitudes, which are based on perceptions of diverse entities that vary across
respondents, issue preferences are tied to the same stimuli.
During election season, presidential candidates bombard the voting public
with multiple policy proposals. The fluidity of the campaign further complicates
things for voters as new controversies emerge and old ones fall off the agenda.
Over the past half dozen or so elections candidate proposals have addressed
many topics, including abortion, business regulation, business taxes, climate
change, deficit reduction, energy policy, economic stimulus, estate taxes, fam-
ily leave, federal bailouts, gun control, health care reform, homeland security,
income taxes, investment taxes, Medicaid and Medicare, missile defense, pri-
vate school vouchers, education reform, Social Security, stem cell research, ter-
rorism, welfare reform, the war in Afghanistan, the war in Iraq, and so on.
Clearly, this is a lot to keep track of. The next chapter addresses questions about
the breadth and depth of issue attitudes that citizens hold.
POLICY PRINCIPLES
What They Are
For more than 50 years, liberal–conservative predispositions and issue prefer-
ences have dominated the study of voter choice. More recently, a third class
of policy attitudes has emerged as a focal point in the fields of political psy-
chology and public opinion, if not electoral behavior. These orientations have
gone by various names, including core principles, domain-specific principles,
and policy-related predispositions (Feldman, 1988; Hurwitz & Peffley, 1987;
Miller & Shanks, 1996). Despite some subtle differences, these terms describe
the same basic construct. I refer to this class of attitudes as policy principles.
A policy principle is a psychological tendency to evaluate with some
degree of favor or disfavor a general claim about the proper course of action
Policy Attitudes and Political Sophistication 55
3
My thinking has been influenced by Peffley and Hurwitz (1985; Hurwitz &
Peffley, 1987), Feldman (1988), Feldman and Zaller (1992), and Miller and Shanks
(1996).
56 ON VOTER COMPETENCE
require government sanction. And, of course, other people hold positions that
strike a balance between these competing views. Lastly, note that whereas tra-
ditional views are influenced by religious backgrounds, they are not reducible
to religious beliefs. Many who are not especially religious endorse traditional
values.
Military strength represents the final posture under consideration. Attitudes
toward this principle reflect the degree to which someone feels American
military power and the will to use it best serve American foreign policy and
national security objectives. Hawks believe military power is vital for defend-
ing and advancing U.S. geopolitical interests in the international arena. In
contrast, doves are cooler toward military strength because they see it as
less effective and more dangerous than the tools of statecraft (Mueller, 1973).
Between these extremes lie foreign policy moderates who believe a mix of
hard and soft power serves U.S. goals. Hawk and dove principles are broader
than views of hot and cold wars, conflicts with particular foreign adversaries,
specific weapons systems, defense spending, and other policy controversies
(e.g., United Nations dues).
or decline. Lastly, in the foreign policy and security arena, the problem of how
much weight should be attached to military strength has structured public
discourse since the late 1940s. Here, too, issues tied to various adversaries,
whether nuclear armed superpowers such as the Soviet Union and China or
other threats such as the National Liberation Front (NLF) in South Vietnam
and Saddam Hussein in Iraq, often come down to the relative merits of hard
versus soft power. Put simply, divisions over the role of government in the
economic welfare domain, what set of moral standards should guide behavior
in American society, and the projection of American military power abroad
have been pervasive in American politics.
Importantly, these cleavages have taken on partisan hues for nearly as
long. As we have seen, presidential candidates and the national parties have
adopted clear positions on these dimensions for many years. From the 1930s to
the present, Democrats have backed strong government and the Republicans
have not. Since the 1970s, Republicans have proven more conservative than
Democrats on moral standards. Partisan divisions on the hawk–dove dimen-
sion started emerging after Nixon assumed responsibility for the Vietnam War
in 1969, but the 1973 Paris Peace Accords, rapprochement with China, and
détente with the Soviets clouded the picture for a time. When Ronald Reagan
won in 1980, party positions on this divide crystallized. At the presidential
level, then, the Democratic Party has generally fallen to the left of the GOP on
limited government, moral standards, and military power.
The second reason I analyze one posture per issue area centers on the psy-
chological principle of cognitive efficiency. Although other general attitudes
and beliefs might guide political judgment in the three domains, my emphasis
on the aforementioned principles comports well with abundant evidence that
citizens are cognitive misers whose innate ability and motivation to process
political information are greatly constrained (Lavine, Johnston, & Steenbergen,
2012; Sniderman et al., 1991). Given this, it seems reasonable to focus on the
principles especially likely to matter in the various policy domains (cf. Peffley
& Hurwitz, 1985). Of course, this presumes a strong case can be made for
identifying the germane principles. To the extent my historical synopsis is
on the mark, limited government, moral traditionalism, and military power
are eminently reasonable focal points. Moreover, empirical research suggests
these principles exert more consistent and powerful effects on some political
judgments than other plausible claimants, such as economic individualism or
free market capitalism in the social welfare realm, authoritarianism in the cul-
tural issues domain, and isolationism in foreign policy (Feldman, 1988; Goren,
2004; Hetherington & Weiler, 2009; Hurwitz & Peffley, 1987). To sum up, the
enduring visibility of the three great policy conflicts, the manner in which they
58 ON VOTER COMPETENCE
have been incorporated into presidential and party images, and the inherent
cognitive limitations that confront all human beings underlie my claim that
limited government, traditional moralism, and military strength function as
central heuristics in mass belief systems. I adduce a good deal of empirical
support for this claim in Chapters 5 through 8.
POLITICAL SOPHISTICATION
Having defined policy principles and distinguished them from liberal–
conservative and issue attitudes, I turn now to political sophistication. This
variable represents a critical moderator in the study of political choice and
it features heavily throughout the book. The importance of sophistication
was established in Converse’s seminal 1964 essay. Despite the theoretical and
empirical force of his argument, investigations as to how and why sophistica-
tion affects political judgment were largely neglected until the publication of
three important books in the 1990s.4 These were Reasoning and Choice by Paul
Sniderman, Richard Brody, and Philip Tetlock in 1991, The Nature and Origins
of Mass Opinion by John Zaller in 1992, and What Americans Know about Politics
and Why It Matters by Michael Delli Carpini and Scott Keeter in 1996. Because
of these works, the “sophistication interaction” model of political judgment
is now standard fare in the fields of political psychology, public opinion, and
electoral behavior.
Following prior work, I see political sophistication as a combination of
factual and associational political knowledge stored in long-term memory
(Converse, 1964; Luskin, 1987; Neuman, 1986). Factual knowledge refers to
correct information about national politics, including leading public offi-
cials, the parties, political institutions, government processes, and domestic
and foreign policy issues. Politically sophisticated individuals have many
discrete political facts at their disposal whereas the unsophisticated do not.
Associational knowledge denotes the organization of these bits of data into
broader knowledge structures.5 These structures help people manipulate new
4
Stimson (1975), Chong, McClosky, and Zaller (1983), and Neuman (1986) repre-
sent notable exceptions. These efforts aside, most work that engaged Converse
sought to refute his claims by arguing that the public as a whole was more
sophisticated than he allowed, thus sidestepping questions about the effects of
sophistication variance.
5
Note that associational knowledge structures as described here are distinct from
liberal–conservative ideologies as described earlier in the chapter. Knowledge
structures refer to connections between political facts whereas ideological struc-
tures denote connections between political attitudes and beliefs.
Policy Attitudes and Political Sophistication 59
political information in an efficient and effective manner. That is, among the
politically sophisticated incoming information can be understood, stored,
integrated, and retrieved without much trouble. In contrast, the unsophisti-
cated experience great difficulty processing whatever information they stum-
ble across. Moreover, they cannot readily store nor recall the information to
which they have been exposed (Fiske, Lau, & Smith, 1990; Lau & Redlawsk,
2006). A final point. Political sophistication is best seen as varying along an
underlying continuum. People are politically knowledgeable or ignorant to
a greater or lesser degree. To put it another way, sophistication differences
reflect differences in degree rather than differences in kind. This point should
be kept in mind when I discuss the behavior of sophisticated and unsophisti-
cated groups—and experts and novices—later on in the book.
For many analysts, political sophistication is the lynchpin of democratic
citizenship. In light of this, the obvious question is just how sophisticated is
the American public. Regardless of how the concept has been measured, the
answer is always the same. Most people are not terribly informed about gov-
ernment and politics. Just as importantly, there exists considerable variation in
what they know. Some are deeply informed about politics, others know little
or nothing, and others fall at points between (Delli Carpini & Keeter, 1996).
This point can be illustrated using public opinion data from the 2004 NES
survey that contains factual items covering a range of political topics. To take
some examples, one question asked citizens to place the Republican Party on
a seven-point scale ranging from extremely liberal to extremely conservative;
66% put the GOP in the conservative end of the scale. Although very easy by
the standards of political elites, it stumped a third of the public during the
heat of an ideologically charged election. And this is as good as it gets. When
asked which party had the most members in the Senate before the election,
51% of the sample correctly said the Republicans. Of course, if everyone were
guessing randomly we would arrive at the same figure. To take one last exam-
ple, a question asked whether George W. Bush felt it was more important to
6
To many readers of this book, these are laughably easy questions. However,
the point to recognize is that the items do a fine job in discriminating between
knowledge-rich and knowledge-poor citizens. If respondents are given harder
questions most will fail to answer correctly, thereby leaving analysts with too lit-
tle variance to permit systematic study. My point is not to criticize the public, but
rather to emphasize the fact that when we talk about knowledge, respondents
who score high are not very well informed by the standards of those who prac-
tice, work, teach, or study politics for a living.
60 ON VOTER COMPETENCE
10
6
Percent
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Number of Correct Responses
Mean = 8.22, Stdev = 4.28
Source: 2004 NES
protect the environment or preserve jobs. Only 36% of the sample saw Bush as
conservative on the issue.6
To see what the distribution of knowledge in the American public looks
like, I use information items from the 2004 NES to create a knowledge scale
and plot the results. Each respondent’s score represents the number of correct
responses he or she gave to 15 questions. The histogram in Figure 3.1 shows
the distribution. Two points stand out. First, the public as a whole performs
rather poorly, as indicated by a mean score of 8.22 correct answers out of 15
(or 55%, a failing grade). Second, we see a lot of variation across the scale (the
standard deviation = 4.28 right answers). A sizable share of the public got
most questions right, many others turned in a middling performance, and
a nontrivial segment performed abysmally. To be specific, about 17% of the
sample provided three or fewer correct answers while another 21% got 13 to
15 answers right.
These are typical results, a point affirmed by looking at data from the 1992
NES. This instrument contains items asking about the job or political office
held by Dan Quayle and Tom Foley, which branch of government is respon-
sible for determining the constitutionality of a law, and additional items. The
sample performed best on the Quayle item with 88% identifying him as Vice
President. Beyond this, the public fared much worse. For instance, 58% knew
the responsibility for determining the constitutionality of a law rests with the
Supreme Court. Only 26% knew Foley was House Speaker at the time. The
mean number of correct responses on the 11-item scale was 5.67 (about 52%).
Policy Attitudes and Political Sophistication 61
14
12
10
Percent
0
0 1 2 3 4 5 6 7 8 9 10 11
Number of Correct Responses
Mean = 5.67, Stdev = 3.07
Source: 1992 NES
Figure 3.2 shows the 1992 knowledge distribution. Once again, there is exten-
sive variation across the scale (standard deviation = 3.07). About 20% of the
sample did poorly, scoring two or less, while another 21% got nine to eleven
answers right. When it comes to political knowledge in the American public,
the mean is low and the variance is high (Converse, 1990). The conventional
view is that such variation has profound implications for political judgment
and voter competence.
CONCLUSIONS
This chapter has defined the key concepts upon which my theoretical edifice
rests. To review, I applied the Eagly and Chaiken definition of attitudes to
three classes of policy attitudes. Liberal–conservative attitudes represent eval-
uations of liberal–conservative labels, which are tied to politically relevant
symbols or social groups that vary across perceivers. Issue preferences cor-
respond to evaluations of specific government plans to address some readily
identifiable problem. Policy principles are evaluations of claims about what
should be done in the primary issue areas that define American politics. Policy
principles are similar to liberal–conservative attitudes and issues attitudes in
that they represent psychological tendencies to evaluate some aspect of public
policy favorably or unfavorably, but they differ with respect to the attitude
object at hand. Policy principles are more focused than liberal–conservative
attitudes and more abstract than concrete issues.
62 ON VOTER COMPETENCE
Ordinary citizens vary widely in terms of what they know about public affairs.
The customary view among scholars is that such differences have profound
implications for understanding the quality of policy attitudes and how these
shape electoral choice. According to the sophistication interaction model,
sophisticated citizens hold a wider range of more fully crystallized, tightly
connected policy attitudes than the unsophisticated. This approach posits fur-
ther that ideological predispositions and issue preferences shape candidate
choice to a much greater degree among the politically aware. In conjunction,
these propositions sustain the claim that the sophisticated better approximate
the standards of political competence than the unsophisticated. The findings
from this perspective thus pose a puzzle: how can citizens who are too unso-
phisticated to vote based on liberal–conservative and issue attitudes come to
rely on policy principles?
My task in this chapter is to explain how and why policy principles guide
candidate choice for practically all voters. I do so by specifying a set of condi-
tions individuals must satisfy to policy vote and then applying this frame-
work to liberal–conservative attitudes, issue attitudes, and policy principles.
The three conditions are availability, meaning the policy attitude is stored in
voters’ memories; centrality, which means it functions as a critical heuristic
in political judgment; and position matching, which implies citizens figure
out which candidate lies closer to them on a given policy dimension and vote
for that candidate. To preview the argument, I posit that political sophistica-
tion enhances the likelihood that citizens meet these conditions for liberal–
conservative and issue attitudes. In contrast, both the politically aware and
unaware should prove capable of satisfying these conditions for the use of
limited government, traditional moralism, and military strength. In short, the
first two faces of policy voting are conditional on sophistication, but the third
face is not.
63
64 ON VOTER COMPETENCE
Before proceeding, readers should note that policy voting, as I use the
phrase, stands for choosing based on liberal–conservative attitudes, issue atti-
tudes, and policy principles. That is, policy voting covers the three classes of
policy attitudes rather than applying to discrete issues alone.
1
“The second condition requires that there be some sense of the importance of an
issue” (Campbell et al., 1960, p. 170).
2
“[A]mong persons who are familiar with an issue, what variation can be
observed in a quantitative measure of intensity, extremity, or strength of
opinion?”(Campbell et al., 1960, p. 177).
3
“For a given problem of government, intensity of opinion will depend on the
importance of the values involved” (Campbell et al., 1960, p. 177).
The Three Faces of Policy Voting 65
other parties.” If voters hold a real attitude on some policy and correctly per-
ceive where the Democratic and Republican nominees stand, they can deter-
mine who falls nearer to them on this dimension. Armed with this knowledge,
voters can be ready to back that candidate. Note my emphasis on “real” atti-
tude. Policy voting occurs only when the person holds a durable evaluative
disposition about the subject, recognizes where the candidates stand, and
chooses accordingly. People who take a position on an issue may perceive
that their favored candidate is closer to them than the candidate actually is, or
may adjust their position to fit their perception of the favored candidate’s pos-
ition. In neither case can we conclude that position matching has transpired.
Instead, the “match” results from voter projection or candidate persuasion.
To reiterate, Campbell et al. described three conditions citizens must sat-
isfy for policy voting to take place. People must hold real attitudes on some
controversy, have strong feelings about it, and see which candidate is in closer
proximity to their own views. If these requirements are met, policy voting
should happen. In what follows, I elaborate the first and third conditions,
which I denote as availability and position matching, respectively, and substi-
tute a second condition which, per Converse (1964, 1970), I call centrality.
Attitude Availability
The first policy voting requirement is that a policy evaluation be available in
the person’s belief system. Availability means the attitude is encoded some-
where in long-term memory—“cognized in some form”—and hence that it
has the potential, once activated, to shape electoral choice (Higgins & King,
1981). The obvious question at this juncture is how do policy evaluations wind
up in memory in the first place?
Social scientists have developed an array of complementary models
accounting for the acquisition and maintenance of political attitudes. The
essential points from these models can be condensed into a four-step process.4
Policy attitudes, again, defined broadly to include liberal–conservative orienta-
tions, issue attitudes, and policy principles, come into being when people are
exposed to information about some policy entity, comprehend what they have
been exposed to, render a bottom line evaluation about the object, and store the
impression in their memory banks. Let us consider each step in the process.
Exposure occurs when someone spends time, however limited, contem-
plating some entity. Exposure cannot be reduced to mere perception. It is not
4
My perspective draws heavily on Converse (1964), Chong et al. (1983), and
Feldman and Zaller (1992; Feldman 1988; Zaller 1992).
66 ON VOTER COMPETENCE
enough for people to notice something in passing and then immediately forget
about it, nor to perceive it at a subconscious level. Rather, the individual must
be aware of the stimulus and mull it over a bit. The degree of engagement with
the entity need not be extensive; indeed, it can be quite brief. But there must be
a conscious recognition of the object. Of course, the more salient a policy entity
is in public debate, the greater the likelihood an individual will come across it
at some point. Ongoing controversies are especially likely to attract notice.
Next, comprehension means the information received is readily under-
stood. For a stimulus requiring neither elaboration nor explanation, com-
prehension is instantaneous. Examples of readily understood topics include
a declaration of war, a race riot, or the death penalty. Because people know
what these phrases signify in a literal sense, they can be evaluated directly. For
obscure topics comprehension will be limited to those who possess the back-
ground knowledge needed to decipher the message. Some examples include
the Central America Free Trade Agreement (CAFTA), capital gains tax cut pro-
posals, or cap and trade pollution credits.
Exposure and comprehension represent the critical stages for developing
political attitudes, because evaluation and internalization usually happen once
the first two requirements have been met. To illustrate, consider the principle
of limited government. The question of how much the national government
should do for those in need has been a recurring theme in political discourse
since the 1930s. People are likely to come across this idea at some point early
in the political life cycle. Exposure may come from stories on the front page of
the local paper, hearing a political discussion between their parents or taking
part in one at school, or watching campaign advertisements during commer-
cial breaks from their favorite television shows. Because the meaning of this
idea is transparent, a person can form an impression and encode it in memory
without difficulty. With additional exposure to the idea, the attitude is evoked
anew, leading the person to think about government power some more. As
this is done, the principle evaluation association grows stronger. Over time, a
summary judgment about limited government becomes embedded within the
person’s belief system.
To wrap up, when a person perceives some policy entity, knows what it
means, renders a global evaluation of it, and stores the judgment in memory,
an attitude forms. Subsequent exposure to the entity reinforces the object eval-
uation association, pushing the attitude a bit closer to the disposition end of
the attitude–nonattitude continuum. The attitude now resides in the voter’s
mind, ready to guide information processing and decision making. I leave
until Chapter 7 the issue of what factors predispose citizens to render positive
or negative evaluations of the object.
The Three Faces of Policy Voting 67
Attitude Centrality
The second requirement for policy voting is that the attitude functions as a cen-
tral heuristic in someone’s mind. A number of available attitudes might con-
ceivably influence the vote, but it seems improbable all will do so. Presumably,
some attitudes are more consequential than others, so they should manifest
stronger effects on political judgment. The question on hand is what charac-
teristics of different classes of attitudes render them powerful with respect to
political choice. This is where centrality comes into play.
Early work in the field of psychology presumed that attitudes guide
behavior related to the object across many situations. To invoke Allport
again (1935), “Attitudes determine for each individual what he will see
and hear, what he will think, and what he will do” (p. 806). Today, psy-
chologists recognize that the power attitudes wield over behavior is con-
tingent and have identified factors that moderate this relationship. These
include attitude accessibility, which refers to how quickly an object evalua-
tion association can be moved from long-term to working memory; attitude
importance, which denotes how much personal significance an individual
attaches to the matter; attitude certainty or the amount of confidence a per-
son has in his or her attitude; and attitude ambivalence or the degree to
which the object evokes positive and negative reactions (Visser, Bizer, &
Krosnick, 2006, pp. 3–4).
Strength-related properties tell us a lot about how, why, and under what
conditions a given attitude influences behavior. For instance, accessible atti-
tudes shape behavior more than inaccessible attitudes, personally important
concerns have larger effects than unimportant concerns, and so on (see Petty
& Krosnick, 1995, for an excellent treatment of this topic). However, my pri-
mary theoretical interest lies in examining how different types of policy evalu-
ations guide electoral choice vis-à-vis one another. Put otherwise, I am more
interested in the relative influence of liberal–conservative attitudes, issue pref-
erences, and policy principles on the vote than in how variation in strength-
related attributes for a single attitude conditions choice. Attitude centrality
represents the critical attribute on this score.
To see what centrality is and why it matters, I must first explain how polit-
ical attitudes are organized in the minds of citizens. Recall that an attitude,
an association between some discernible entity and an evaluation, is a dis-
crete mental construct. For example, evaluations of Barack Obama are psy-
chologically distinct from evaluations of African-Americans. So, too, are
feelings toward government health care reform and the minimum wage. But
attitudes do not exist in isolation. Instead, they are linked to one another in
68 ON VOTER COMPETENCE
5
I am referring to interattitudinal structures, which are different from intraattitu-
dinal structures. The latter refer to the affective and cognitive bases that underlie
a single attitude.
The Three Faces of Policy Voting 69
Health
Minimum School War in Defense
care Tax Cuts Abortion
wage prayer Iraq spending
reform
Said another way, a few long-term postures give rise to an array of short-term
evaluations such as new issues, presidential job performance assessments, and
so on. Figure 4.1 illustrates the logic. In this diagram, the three abstract prin-
ciples function as central attitudes that vertically constrain issue preferences
in each domain. For example, limited government shapes attitudes toward
health care reform, the minimum wage, and tax cuts without being recipro-
cally shaped by them. Likewise, moral traditionalism guides the positions citi-
zens take on abortion and school prayer, but these issues do not influence the
broader principle. This is a strong assumption. I test it in Chapter 6.
The next theoretical step requires explaining how centrality facilitates can-
didate choice. This can be accomplished by considering how people make
political judgments in general. I begin by noting, as many have before, that
social life is bewilderingly complex. When thinking about some matter or
weighing alternatives, cognitive misers utilize decision-making strategies to
reduce this complexity to a manageable level. Moreover, they do so in the
face of innate limitations on the cognitive and motivational resources they can
bring to bear on the task (Fiske & Taylor, 1991).
Heuristics help us make sensible choices under these constraints. Heuristics
are rules of thumb used to simplify judgmental tasks in ways that often yield
acceptable, though not necessarily optimal, decisions (Chaiken, 1987). Rather
than attending to all relevant bits of information that might conceivably matter
70 ON VOTER COMPETENCE
for the decision, people search out diagnostic cues that can act as effective sub-
stitutes for complete information. Heuristics thus serve as computational aids.
According to Sniderman et al. (1991):
overload is ever present. They cannot evaluate every policy datum they encoun-
ter, encode each one, and recall them all when comparing candidates. They need
to simplify the choice by employing cognitive strategies that facilitate decision
making efficiency. Moreover, to be handy these strategies must have a reason-
able likelihood of yielding a choice consistent with fully informed preferences.
The challenge, again, is to make a good choice with minimal effort.
This is where central policy attitudes come into play. Because people
already use these to construct other judgments, they can employ these trust-
worthy guides in the area of candidate choice. Instead of attending to all pol-
icy proposals raised during the campaign or studying the candidates’ prior
records, voters use central attitudes as rough proxies to guide their selection.
For instance, if attitudes toward traditional standards of right and wrong
are more central than attitudes toward abortion, sex education, and school
prayer, the general principle should have a stronger effect on candidate eval-
uation than these issue attitudes. Indeed, the existence of general principles
dispenses with the need to think about most specific issues.
To sum up, the use of central policy attitudes is economical and likely to
produce the same choice that would result if a more comprehensive and delib-
erative approach were taken. In essence, central policy attitudes are akin to
partisan identities, transcending elections and candidates, ready to be acti-
vated during the campaign. They are probably not as powerful as partisan
identities in shaping electoral choice, but that does not diminish their impor-
tance. To the degree that central attitudes guide the construction of short-term
political judgments, they should prove consequential for voter decision mak-
ing. But for this to happen, one final hurdle must be overcome.
Position Matching
Per The American Voter, citizens must recognize which candidate will better
represent their position on a given policy dimension. This occurs when vot-
ers compare their position to those of the rival candidates and go for the one
whose position is more similar. This is called position matching. The key mat-
ter here concerns the mechanisms by which a voter learns candidate positions.
There are several channels through which a voter can acquire the requisite
information. First, the voter may know where the presidential hopefuls stand
at the outset of the campaign. For the most part, early learning is a function
of extant sophistication. Those who are politically knowledgeable in general
are likely to know (or quickly find out) where the nominees stand when elec-
tion season kicks off. Second, voters who lack this information at the start of
the campaign have numerous opportunities to acquire it from an information
environment flush with multiple overlapping cues about candidate positions
72 ON VOTER COMPETENCE
(Rahn et al., 1990). Key campaign events, hard and soft news in all sorts of
media, and ubiquitous campaign commercials all supply a steady stream of
redundant cues.
Likewise, party labels can fill in missing information about candidate posi-
tions. A voter can leverage prior knowledge about the parties to infer where the
candidates lie on the central dimensions. This knowledge may take the form
of correct perceptions of party positions, stereotypical beliefs about the party
and its coalition, or party issue handling reputations (Conover & Feldman,
1989; Green et al., 2002; Petrocik, 1996). Because party images and positions
hold steady over time, a voter can make some reasonable assumptions about
party stances and use these to fill in the blanks about candidate positions.
Finally, citizens may learn candidate views through informal channels.
Individuals are plugged into social networks that furnish some political infor-
mation during election season. Whether through political discussion or inci-
dental learning that occurs as a byproduct of social interactions at work, with
neighbors, with friends, or in religious or social organizations, policy cues can
reach those who plan to show up on Election Day (Huckfeldt & Sprague, 1995;
Popkin, 1994).
I use some examples to illustrate position matching. First, consider those
for whom liberal–conservative placements are an important heuristic. Upon
learning of the ideological leanings of the nominees, it is a simple matter for
political sophisticates to see which one lies closer to them and, other things
equal, throw their support to that candidate. Second, assume someone
favors traditional morality but does not know what the candidates believe.
If a voter knows the GOP is more committed to traditional moral values than
Democrats, he or she can deduce the same holds true for the respective nomi-
nees. Or perhaps this person stereotypes Republicans as the party of religious
orthodoxy. Here, too, this cue facilitates inferences about candidate positions
on moral standards. Finally, the voter may overhear a conversation at work
whereby one colleague is complaining to another that Republicans are too
narrow minded and intolerant on moral issues. Whether via direct knowl-
edge, campaign learning, party-based inference, social contact, or incidental
learning, voters can usually meet the position matching condition. Insofar as
they do, voting based on central policy attitudes should occur.
attitudes cannot impact candidate choice. Second, discrete policy attitudes are
organized into hierarchical structures. Within these structures, central policy
attitudes vertically constrain peripheral attitudes. As these central heuristics
enable the construction of short-term evaluations, a voter can call upon them
when casting a presidential ballot. Therefore, given a set of available policy
attitudes, those occupying the central nodes within political belief systems
are better positioned to guide citizen choice than those at the periphery (the
“centrality” condition). Third, the voter must have a good idea of where the
candidates fall on a given policy continuum. They may learn this during the
campaign, use party schema to deduce candidate positions, or acquire it as
a byproduct of interactions within various social networks. An accurate per-
ception of candidate stances engenders policy-informed voting (the “position
matching” condition).
In the rest of this chapter, I examine the degree to which citizens can be
expected to satisfy these conditions for liberal–conservative attitudes, issue
preferences, and policy principles, paying especially close attention to the
role sophistication plays in the process. Because liberal–conservative voting
and issue voting have been studied extensively, my discussion of the first
two faces of policy voting is grounded in theoretical reasoning and empirical
evidence. In contrast, the lack of research on domain-specific principles and
electoral behavior means my discussion is guided by theoretical logic alone.
Miller and Shanks (1996) present a rare exception, but they do not analyze the
role sophistication plays.
We hear a lot of talk these days about liberals and conservatives. Here is a
seven-point scale on which the political views that people might hold are
The Three Faces of Policy Voting 75
6
Authors sometimes employ education as a proxy for political sophistication,
an admittedly crude, though empirically defensible approach as the correlation
between education and knowledge scales typically lies in the neighborhood of .50.
The Three Faces of Policy Voting 77
These results suggest that when it comes to issues, instability rules, and
hence that issue preferences fall near the constructionist end of the attitude–
nonattitude continuum. However, correlations between opinions in successive
panel waves cannot reveal whether change is systematic or random. If such
movement is systematic, the presumption that people hold genuine issue atti-
tudes may be defensible. If this is so, the correlations between adjacent panel
waves should exceed the correlations between the first and third waves. This
is not what Converse found. Instead, correlations between the 1956–1958 and
1958–1960 waves were similar to those for 1956–1960. These results debunked
the claim that observed opinion change reflects levelheaded responses to
evolving events. Instead, issue preferences fluctuate without much rhyme or
reason.
Converse probed further to determine what mix of respondents might
account for these curious patterns. The best fitting model partitioned the
sample into two groups. The first consisted of respondents whose attitudes
never changed, whereas the second, much larger group was composed of
subjects whose opinions varied randomly over time. This “black and white”
model provided a nearly perfect fit to the data on one of the eight issues—the
“power and housing” item. He found that less than 20% of the public held
“real and stable attitudes” on this issue; the remaining 80% either lacked such
preferences or held preferences that moved randomly over time (Converse,
1970, p. 176). Of the remaining seven items, the black and white model fit the
data reasonably well, though not quite as good as the power and housing
item. Evidently, many respondents constructed opinions on the spot because
no evaluative knowledge was available for recall from long-term memory.
Converse (1964) concluded famously “that large portions of an electorate do
not have meaningful beliefs, even on issues that have formed the basis for
intense political controversy among elites for substantial periods of time” (p.
245). This position is known as the “nonattitudes” thesis. Subsequent work on
response instability confirmed these were not anomalous findings.7
Given the dispiriting implications these findings have for citizen compe-
tence, many have challenged the nonattitudes thesis. Achen (1975) has articu-
lated the most prominent critique. He speculated that response instability may
be a function of inadequate questions rather than inadequate respondents.
7
Converse and Markus (1979) conducted similar analyses using data from the
1972–1974–1976 NES panel and found that classic guns and butter issue prefer-
ences were as wobbly in the 1970s as they were in the 1950s. The authors also
found party identification to be exceptionally stable and that new cultural issues
were moderately stable as well.
80 ON VOTER COMPETENCE
requirements, some make it past these hurdles. For whom should issues mat-
ter? Scholars have advanced two complementary theories. To begin with,
the sophistication interaction model applies once more. Because the politi-
cally aware think longer and harder about public affairs, they are more apt to
develop firm attitudes on at least some issues and compare their positions to
those espoused by the candidates (Delli Carpini & Keeter, 1996). Another the-
ory focuses on single issue voters (Krosnick, 1990). People who care intensely
about an issue attach a great degree of personal importance to their evalu-
ations. The personal importance of the issue motivates these voters to learn
what the candidates believe so the right one can be supported. Finally, some
work shows single issue voting is especially pronounced among knowledge-
able citizens (Anand & Krosnick, 2003).
This is not to say that issues never matter. Occasionally, an issue gener-
ates enough controversy to shape candidate choice throughout the electorate.
For instance, Abramowitz (1995) has shown that attitudes toward abortion
weighed heavily on voter decision making in 1992, a plausible finding in light
of Patrick Buchanan’s “culture war” speech at the GOP convention and Bill
Clinton’s vigorous defense of abortion rights. Nevertheless, the broader point
to take away is that issues are unlikely to manifest consistently strong effects
on electoral choice for most voters in most elections. Given the large number
of issues raised during presidential campaigns, the innate cognitive limita-
tions that burden voters’ processing capabilities, the peripheral status of issue
preferences in mass belief systems, and the difficulty in learning candidate
positions, issue voting is perforce limited.8
8
Recent work by Ansolabehere et al. (2008) suggests that issue preferences are
stable for nearly everybody and issue voting is not conditional on sophistication.
Their critique of prior work is grounded in methodological rather than theoreti-
cal considerations. I address it in Chapter 5, which takes up measurement.
The Three Faces of Policy Voting 83
staring vainly at their shoes in search of cues to help in their vote decision is in
all likelihood not a hyperbole” (Neuman, 1986, p. 173).
Such pessimism is unwarranted. The possibility that American voters
incorporate an alternative class of policy evaluations into their electoral calcu-
lus must be explored before we conclude that policy voting lies beyond their
reach. This brings us to the third face of policy voting, selecting based on pol-
icy principles, which I defined as global evaluations about the proper course
of action to take in the major issue areas in the public sphere. As always, policy
attitudes should guide candidate choice when the standard conditions apply.
First, attitudes toward limited government, traditional morality, and military
strength must be available in memory. Second, each must function as a central
heuristic in political judgment. Third, voters must identify the candidate who
lies closer to them on each dimension. If these conditions are met, principle-
guided choice should ensue.
To begin with availability, citizens may hold real attitudes toward each
principle if they are exposed to information about the idea, comprehend and
evaluate it, and assimilate these feelings. Because the manifest content of each
principle is transparent, citizens do not need ongoing contact with elite debate
to learn what the claims entail. All the same, they need some exposure, and
because many care little about public affairs even minimal levels of exposure
cannot be assumed (Zaller, 1992).
As my earlier discussion has shown, the bulk of the electorate knows little or
nothing about liberalism and conservatism. Likewise, large slices of the elector-
ate remain ignorant about particular issues. This holds true for moderate to low
salience political battles, which is not terribly surprising. It also holds for high
profile issues such as the budget deficit, the Bush tax cuts, and the war in Iraq
(Bartels, 2008; Delli Carpini & Keeter, 1996; Kull, Ramsay, & Lewis, 2003/2004).
But as we have seen in the historical chapter, public discourse in the main issue
areas usually reduces to a more fundamental division. People do not need to
stay abreast of day-to-day developments in a given domain to acquire policy
principles. All that is necessary are occasional encounters with some aspect of the
major policy cleavages that animate public debate over the long haul (Feldman
& Zaller, 1992). This exposure accrues through a combination of secondary and
postsecondary education; the hullabaloo accompanying presidential campaigns;
hard news coverage of economic and political crises; coverage that spills over
into soft news outlets; viral sensations on the Internet; social interactions with
family, friends, colleagues, and other acquaintances; and incidental learning.
Through multiple channels, voters can encounter policy principles.
Given exposure, can we expect individuals to understand what these
principles mean? The critical point here is that the idea underlying each one
84 ON VOTER COMPETENCE
to enduring party images. Given that the principle party linkages are few
in number and have long been visible, they are easy to learn and remember
(Carmines & Stimson, 1980). Very large shares of the electorate know that
Democrats “want to help the middle class and the poor” while Republicans
stand for “traditional morals” and “strong defense” (Green et al., 2002; Lewis-
Beck et al., 2008; Petrocik, 1996). As such, people can manipulate party labels to
make reasonable guesses about where the nominees stand on first principles.
Lastly, information about policy principles flows through multiple social
networks. Some might encounter cues as a byproduct of interacting with fam-
ily members, friends, neighbors, co-workers, fellow parishioners, and so on
(Huckfeldt & Sprague, 1995; Popkin, 1994). They may learn during a brief
exchange with a co-worker at the water cooler or on the shop floor, from a fel-
low power lifter or runner at the local gym, from another parent at their son’s
gymnastics practice or their daughter’s debate tournament, or from a friend
at the weekly pool hall gathering or bingo session at church. Even those who
pay relatively little attention to the campaign and have limited knowledge
stores likely receive some exposure to information about candidate positions
on general principles.
Overall, multiple sources saturate the information environment with rein-
forcing signals about where the candidates and parties stand on the fundamen-
tal policy cleavages that define American politics. Although no one receives
all campaign messages, most probably encounter a few from time to time from
different sources. And because much of this information is repeated over and
over, those who miss it at one point have plenty of chances to hear it later on
(Just et al., 1996; Rahn et al., 1990).
To reiterate, my argument maintains that nearly everyone holds genu-
ine attitudes toward limited government, moral traditionalism, and military
strength, thereby satisfying the availability condition. This is because both
political culture and campaign debate repeatedly highlight the fundamen-
tal policy cleavages, making it easy to absorb and evaluate these ideas. Next,
these principles function as central elements in political attitude hierarchies
where they vertically constrain short-term political evaluations, thus satisfy-
ing the centrality condition. Insofar as voters have experience calling upon
these heuristics to guide other short-term judgments, the ability to apply them
to electoral choice follows naturally. Finally, given an information-rich cam-
paign environment flush with repetitive cues, most of those who go to the
polls should have a decent idea of where the candidates stand on basic dimen-
sions, thereby satisfying the position matching condition. Since abstract pol-
icy principles are widely available, central, and linked to candidate positions,
citizens should rely on these heuristics when casting their presidential ballots.
The Three Faces of Policy Voting 87
In contrast to the first two faces of policy voting, restricted as they are to the
politically sophisticated, the third face of policy voting should elude relatively
few citizens.
CONCLUSIONS
For over half a century two research questions have dominated the study of
policy voting. First, to what extent do liberal–conservative orientations shape
electoral choice? Second, how much do issue preferences guide the vote? The
main finding from this body of work is that for the most part policy attitudes,
so defined, matter little. If we stop here, it would seem that policy voting
lies beyond the grasp of the unsophisticated bulk of the American elector-
ate. Before accepting this troubling conclusion, a systematic examination of
the interplay between policy principles, political sophistication, and candi-
date choice must be undertaken. Insofar as core principles are available, cen-
tral, and matched to candidate positions for citizens across the sophistication
divide, a third face of policy voting may exist. To date, no such examination
has been performed.
I hypothesize that practically all citizens hold genuine attitudes toward lim-
ited government, traditional morality, and military strength; that each oper-
ates as a central heuristic in their belief systems; and that the sophisticated
and unsophisticated alike are adept at matching positions to those held by the
candidates and grounding their votes in abstract principles. The remainder of
this book tests these claims. I begin in the next chapter by analyzing the degree
to which policy principles are present in the minds of political sophisticates
and novices. Chapter 6 examines attitude centrality and heuristic reasoning. I
then take up the origins of general principles in Chapter 7. Finally, Chapter 8
presents the vote choice analyses.
This page intentionally left blank
CHAPTER 5
89
90 ON VOTER COMPETENCE
observable indicator designed to reflect it. The issue then becomes what meas-
urement procedures should be used to establish that the attitude exists. A
number of studies have examined this matter with great care (Aldrich et al.,
1989; Campbell et al., 1960; Delli Carpini & Keeter, 1996; Lewis-Beck et al.,
2008). These efforts gauge availability via opinion-holding measures. When a
survey respondent answers a policy query, we call the response an opinion.
To illustrate, consider the following item that appears regularly on National
Election Study (NES) surveys.
There is much concern about the rapid rise in medical and hospital costs.
Some people feel there should be a government insurance plan which
would cover all medical and hospital expenses for everyone. Others feel
that all medical expenses should be paid by individuals, and through pri-
vate insurance plans like Blue Cross or other company paid plans. Where
would you place yourself on this scale, or haven’t you thought much
about this?
The response options lie along a seven-point scale with the endpoints labeled
“government insurance plan” or “private insurance plan.” When someone
chooses a point on the scale, the presumption is that the response reflects a
meaningful attitude. In contrast, when someone admits that he or she “hasn’t
thought much about this” or says “I don’t know,” analysts conclude that no
evaluation resides in memory, that the attitude is unavailable.
Although opinion holding seems like a reasonable way to explore attitude
availability, it has serious flaws. For starters, social desirability pressures com-
promise the technique’s utility. Respondents sometimes report opinions on
issues to which they have given little or no thought. They do so to convey
the impression that they care about the issue, that they are neither apathetic
nor uninformed. Although “pseudo-opinions” are prevalent on many obscure
and hard issues, they may come into play on standard political issues as well
(Bishop, Oldendick, Tuchfarber, & Bennett, 1980). Inasmuch as survey respon-
dents dissemble in this fashion, opinion-holding measures overstate the prev-
alence of policy attitudes.
To discourage pseudo-opinions, researchers and pollsters rely on “no opin-
ion” (NO) filters, such as the one above in the health care question. Common
sense suggests NO filters provide a legitimate out for respondents to admit
that they have no preference. Unfortunately, the filters do not work as adver-
tised. In a comprehensive review of their use, Krosnick (2002) has shown
that NO filters dissuade some who hold genuine attitudes from expressing
them. This may occur because respondents are ambivalent about the issue,
The Availability of Policy Principles 91
presumably exists for everyone, but the approach generalizes to multiple atti-
tudes and groups. If a set of policy attitudes is present in the minds of groups
at various levels of sophistication, comparable factor structures and loadings
should emerge across the groups. When the evidence follows this pattern,
measurement invariance holds (Bollen, 1989). However, if policy principles
exist only in the minds of the politically aware, evidence of structure will be
found there and nowhere else.
To test for measurement invariance or equivalence, I trichotomize my
samples by political sophistication, specify a CFA model that links the latent
variables to multiple indicators, and determine whether the indicators tap
the latent factors to a similar degree across the groups. If so, the data should
reveal good model fit in the samples, which would imply analogous factor
structures. Likewise, similar factor loadings would suggest the unsophisti-
cated do as well as the sophisticated in translating unobservable attitudes into
observable opinions. I focus on standardized factor loadings that reflect the
correlation between latent factors and observed indicators because they have
an intuitively clear interpretation. Insofar as the model fit and item factor cor-
relations are similar across samples, I can infer measurement invariance holds.
If I find discrepant model fit or widely varying loadings, invariance is not pre-
sent. Before carrying out these tests, I describe my data and measures.
1
The combination of a multistage area probability design, the use of the face-to-
face interviewing mode, and high response rates make NES samples the gold
standard in data quality for electoral behavior research. There are at least two
downsides to using these data. First, the questions often diverge from researcher
preferences. Second, necessary measures are sometimes unavailable.
The Availability of Policy Principles 93
lie on a 0–1 scale. The final knowledge index is standardized this way as well,
with 0 indicating no questions answered correctly and the lowest interviewer
rating and 1 signifying all correct answers plus the highest interviewer rating.
There are 11 to 17 items in these scales, depending on the year.
I use a mix of easy, middling, and hard items. Some ask interviewees
to place the presidential candidates and parties on the correct side of vari-
ous scales, such as liberal–conservative or left–right continua, government
spending, defense spending, and so on. To get credit, respondents must place
Democrats on the liberal side and Republicans in the conservative end. I also
employ items that ask about the offices or positions held by various public
figures.2 Lastly, questions on political institutions and procedures, such as
which party controls the House and Senate, are utilized when available (see
the Appendix for the items).
This strategy can be justified as follows. To begin, these scales tap the fac-
tual knowledge attribute of the sophistication concept in a highly discriminat-
ing manner. The ability to answer 9 or 10 questions correctly suggests far more
information holding than implied by these 10 questions. Similarly, someone
who cannot answer more than a couple would be hard pressed to do better
if given more opportunities. Hence, although the number of items used to
construct each scale is limited, they are diagnostic of vast differences in latent
knowledge (Converse, 2000). Next, studies show that items such as these
reflect a single latent trait, that additive scales based on such items are highly
reliable (the Cronbach α reliability coefficient exceeds .80 for every scale in
this study), and that knowledge scales are construct valid (Delli Carpini &
Keeter, 1993; Luskin, 1987; Neuman, 1986).
With respect to the NES interviewer rating of each respondent’s informa-
tion level, these judgments are made by professional interviewers with con-
siderable experience evaluating respondents in lengthy face-to-face sessions.
Although the interviewers cannot determine objectively how informed their
companions are, their ratings are shaped by respondents’ interest in and
engagement with the survey. Additionally, reliability and validity tests carried
out here and elsewhere demonstrate that this item correlates robustly with
2
I employ the office holder items for 1996 and before, but not thereafter. This is
because “don’t know” responses were not handled consistently in later surveys.
Some respondents were encouraged to take a guess after saying “don’t know”
and others were not. To be clear, I am referring to both the split ballot “don’t
know” wording experiments used in the 2000 NES and the 2004 and 2008 sur-
veys in which no such experiments were administered. In any case, alternative
coding schemes for knowledge do not alter the results in any significant way.
94 ON VOTER COMPETENCE
this is desirable for two reasons. First, NO filters screen out some who hold true
attitudes. Second, because most respondents answer all questions I can deter-
mine whether responses reflect true attitudes or random noise.
Although no single item can effectively tap the general principle, the key
point is that when multiple indicators are combined the cumulative nature of
the measure captures the principle rather nicely. Those who want government
to do more should favor spending increases across the board; those reject-
ing the claim should favor cuts (Jacoby, 1994). Is this assumption reasonable?
The items may seem deficient because they fail to ascertain whether subjects
believe the national government should do more to improve conditions for
these groups, asking instead only whether it should spend more. Moreover,
references to federal spending seem to come uncomfortably close to the types
of preferences limited government is supposed to explain. Although such con-
cerns cannot be mitigated entirely, the spending items possess a number of
attractive features that should alleviate these worries.
First, although the ideas of government “doing more” and “spending
more” differ semantically, respondents will probably react to them in a similar
way. When asked to evaluate federal spending on the poor, the homeless, wel-
fare, and such, the answers given likely depend in large part on feelings about
government assistance more generally. Next, because various subsets of the
population may benefit from government aid, items that evoke group centric
thinking are needed. The spending items satisfy this requirement handsomely.
Some do so explicitly (e.g., poor people, the homeless) and others implicitly.
In the latter category, the welfare items evoke images of welfare recipients, the
Social Security item calls to mind the elderly, and the child care item implies
help for working parents and single mothers. Third, because group deserv-
ingness varies systematically across these measures, the items can differen-
tiate between welfare state supporters and detractors. Citizens committed to
the idea that the federal government has a responsibility to do something for
the “have nots” in society should favor spending more on both sympathetic
(e.g., the elderly) and unsympathetic (e.g., people on welfare) groups. Those
who agree up to a point should favor aid for “deserving” groups alone. And
those who reject the idea should oppose spending across the board. Fourth,
evidence presented in Chapter 6 shows that the latent variable tapped by the
spending items dynamically constrains short-term preferences on a concrete
issue (i.e., health care) without being reciprocally shaped by these, which is
what we would expect if the spending items tap a general policy orientation
rather than a specific issue. Given these virtues, the spending items can serve
as valid and reliable (albeit imperfect) measures of attitudes toward govern-
ment activism.3
96 ON VOTER COMPETENCE
3
There are problems with items that appear on NES surveys that seem to be face
valid measures of limited government. First, the seven-point scales provide
respondents with a NO filter, which, as discussed earlier, is unappealing. As
I show later in the chapter, missing data on these questions are high. Second,
some of the questions lack clear group referents. Indeed, no other set of items
can rival the spending battery in terms of breadth and depth of group coverage.
Third, there are fewer items available relative to the spending measures. In short,
although the standard seven-point scales are useful for some purposes, I believe
they are inferior to the spending battery.
The Availability of Policy Principles 97
the international arena. Hawks are comfortable with using military muscle to
advance foreign policy and national security objectives whereas doves prefer
noncoercive forms of engagement. I use whatever items appear on a given
survey to construct this variable (see the Appendix). Here are some examples.
First, a 1988 item: “The U.S. should maintain its position as the world’s most
powerful nation even if it means going to the brink of war.” Five response
options are given, ranging from “agree strongly” to “disagree strongly.”
Second, a 1992 query: “In the future, how willing should the United States
be to use military force to solve international problems . . . extremely willing,
very willing, somewhat willing, not very willing, or never willing?” Finally,
this question appears on the 2004 NES.
Some people believe the United States should solve international prob-
lems by using diplomacy and other forms of international pressure and
use military force only if absolutely necessary. Suppose we put such peo-
ple at 1 on this scale. Others believe diplomacy and pressure often fail
and the U.S. must be ready to use military force. Suppose we put them at
number 7. And of course others fall in positions in-between, at points 2, 3,
4, 5, and 6. Where would you place yourself on the scale, or haven’t you
thought much about this?
Although the items vary from year to year, the key point is that each elicits
overarching views on military power. The language is generally pitched at a
broad level with no allusion to specific military conflicts, foreign countries,
or regions in the world. Lastly, note that, except for the 2004 item, NO filters
are absent. Hence, missing data are, minus this item, very low. An important
drawback with these data is that only one indicator is available in the 1996
NES and no items appear on the 2000 and 2008 NES.
λ1 λ2 λ3 λ4 λ5 λ6 λ7
x1 x2 x3 x4 x5 x6 x7
δ1 δ2 δ3 δ4 δ5 δ6 δ7
There are, of course, alternative ways the data might be structured. Two
possibilities are especially pertinent. First, if the nonattitudes thesis is on the
mark, the opinions of the highly and moderately informed should be guided
by latent principles, but there should be little evidence of this among low
knowledge respondents. This pattern of results would suggest the latter do
not hold real principles. Second, for the most knowledgeable, responses to the
policy principle items might derive from a single left–right orientation, which
would undercut my argument that opinion is organized in a domain-specific
manner. For the sophisticated, perhaps everything reduces to a single liberal–
conservative dimension.
The available measures permit specification of three-factor models in 1988,
1992, and 2004 and two-factor models in 1996, 2000, and 2008.4 To illustrate my
approach, Figure 5.1 shows the 1992 measurement model. In this specification
4
The careful reader will recall that with the exception of the 2008 NES the lim-
ited government items have three response options. In light of this, I created
miniscales by adding responses from various pairs of items together to create
five-point measures that are used in the CFA analyses. This procedure ensures
the items better approximate interval level measurement as needed for accu-
rate estimation, although slippage obviously remains. Note also that for the
moral tradition items the agree–disagree response options are compromised
The Availability of Policy Principles 99
ξ1, ξ2, and ξ3 represent latent limited government, traditional morality, and
military strength, respectively. Each xi corresponds to the ith variable used to
measure a specified factor. The λi are standardized factor loadings that indi-
cate the correlation between the latent variable and the given indicator. Each
δi represents errors of measurement and unique sources of variation for xi.
Two-way arrows between the factors represent zero-order correlations. I esti-
mate the model at different levels of sophistication by dividing each sample
into comparably sized low, medium, and high knowledge groups. If the global
model fit statistics described below imply good fit in each sample, the con-
clusion of equivalent factor structures is justified. I then examine whether the
factor loadings are similar across the groups.
The first set of estimates appears in Table 5.1.5 The top and middle sec-
tions of the table list the latent principles and corresponding indicators in the
first column. The standardized factor loadings for the low, medium, and high
sophistication groups appear in the second through fourth columns, respec-
tively. The bottom part of the table lists the global fit statistics in column one
followed by the sample estimates in the next three columns. The robust χ2 is the
test statistic for the null hypothesis that the population covariance matrix for
the observed variables equals the covariance matrix reproduced by the meas-
urement model. A statistically insignificant result means the null hypothesis
ξ1 Limited government:
λ1 Spending scale 1 .60 .56 .56
λ2 Spending scale 2 .72 .72 .89
λ3 Spending scale 3 .48 .50 .51
ξ2 Traditional morality:
λ4 Traditional morality scale 1 .68 .79 .76
λ5 Traditional morality scale 2 .58 .60 .77
ξ3 Military strength:
λ6 Military strength item 1 .39 .52 .55
λ7 Military strength item 2 .58 .75 .77
λ8 Military strength item 3 .58 .54 .65
Mean factor loading .58 .62 .68
Model fit:
Yuan–Bentler scaled χ2 34.54 38.28 24.44
Degrees of freedom 17 17 17
p-value <.01 <.01 .11
CFI .95 .96 .99
RMSEA .05 .05 .03
SRMR .04 .04 .03
Number of observations 567 599 571
Notes: Direct maximum likelihood estimates are based on raw data. Standardized factor loadings
represent the correlation between latent and observed variables. All loadings are significant at
p < .05. Latent and observed variables are coded so that higher scores reflect more support for a given
principle; therefore, each λ should be positive. CFI, comparative fit index; RMSEA, root mean square
error of approximation; SRMR, standardized root mean square residual.
cannot be rejected and thus supports the inference that the specified model
holds in the population. Hence, in contrast to the typical practice, a statis-
tically insignificant result is desired. The χ2 performs well as a significance
test when used with large samples, when the distributional assumptions hold,
and when the model is specified correctly. Because one or more of these condi-
tions are typically violated in a given application, the χ2 test loses some value
as a significance test (Bollen, 1989). Because of this, I also report three descrip-
tive fit measures that assess global fit according to various statistical criteria.
These include the robust comparative fit index (CFI) where values close to or
in excess of .95 indicate good fit, the root mean square error of approximation
(RMSEA) in which values close to or below .06 suggest reasonable fit, and the
The Availability of Policy Principles 101
6
The CFI is an incremental fit index that assesses the proportionate improvement
in model fit relative to a baseline where all observed covariances are zero. The
RMSEA and SRMR are absolute fit measures that assess how well the posited
model reproduces the sample data (the former includes a parsimony correction).
102 ON VOTER COMPETENCE
ξ1 Limited government:
λ1 Spending scale 1 .45 .56 .66
λ2 Spending scale 2 .63 .70 .72
λ3 Spending scale 3 .54 .58 .58
ξ2 Traditional morality:
λ4 Traditional morality scale 1 .52 .76 .74
λ5 Traditional morality scale 2 .85 .86 .89
ξ3 Military strength:
λ6 Military strength item 1 .65 .75 .90
λ7 Military strength item 2 .36 .52 .44
Mean factor loading .57 .68 .70
Model fit:
Yuan–Bentler scaled χ2 28.44 14.09 31.05
Degrees of freedom 11 11 11
p-value <.01 .23 <.01
CFI .95 1.00 .98
RMSEA .05 .02 .05
SRMR .04 .03 .03
Number of observations 766 699 751
Notes: Direct maximum likelihood estimates are based on raw data. Standardized factor loadings
represent the correlation between latent and observed variables. All loadings are significant at p < .05.
Latent and observed variables are coded so that higher scores reflect more support for a given principle;
therefore, each λ should be positive. NES sampling weight has been used. CFI, comparative fit index;
RMSEA, root mean square error of approximation; SRMR, standardized root mean square residual.
the low and high samples (remember that an insignificant result is preferred),
the descriptive fit measures meet or surpass the fit thresholds in every group.
The next concern is whether the factor loadings are strong within and compa-
rable across the samples. The mean loading equals .57 for the less informed, .68
for the moderately informed, and .70 for the most informed. Like the 1988 esti-
mates, the item factor correlations are substantial in all three groups, albeit more
so among the better informed. Evidently, most folks receive sufficient exposure
to these ideas in the broader political environment to form genuine evaluations.
The 1996, 2000, 2004, and 2008 estimates appear in Tables 5.3–5.6, respec-
tively. The results from all four surveys reinforce the conclusions reported
above. To summarize, model fit ranges from good to excellent in each sample
in each year, the mean item factor correlations are all substantial (varying
The Availability of Policy Principles 103
ξ1 Limited government:
λ1 Spending scale 1 .42 .56 .60
λ2 Spending scale 2 .64 .69 .77
λ3 Spending scale 3 .54 .66 .71
ξ2 Traditional morality:
λ4 Traditional morality scale 1 .92 .81 .76
λ5 Traditional morality scale 2 .31 .72 .89
Mean factor loading .57 .69 .75
Model fit:
Yuan–Bentler scaled χ2 2.42 12.95 9.67
Degrees of freedom 4 4 4
p-value .66 .01 .04
CFI 1.00 .98 .99
RMSEA .00 .07 .06
SRMR .01 .03 .03
Number of observations 498 529 457
Notes: Direct maximum likelihood estimates are based on raw data. Standardized factor loadings
represent the correlation between latent and observed variables. All loadings are significant at p < .05.
Latent and observed variables are coded so that higher scores reflect more support for a given principle;
therefore, each λ should be positive. NES sampling weight has been used. CFI, comparative fit index;
RMSEA, root mean square error of approximation; SRMR, standardized root mean square residual.
from .56 to .79), and the item factor correlations are always larger in the high
group relative to the low group (on average by .20). Taken in total, the CFA
results confirm my theoretical expectations. The evidence reveals the speci-
fied multidimensional model always fits the data, whereas the factor loading
comparisons imply political sophisticates and novices do well in reporting
their positions (but again, the aware do better). My findings suggest that
although individuals vary greatly in what they know about politics, every-
one possesses policy principles, and, thus, satisfies the first condition of pol-
icy voting.
A Unidimensional Alternative?
Evidence presented above indicates that three distinct policy predispositions
reside in the political attitude structures of everyday people. This allows me
to discount the nonattitudes thesis, which maintains that such attitudes are
absent from the minds of political novices. However, as discussed above, a
TA B L E 5 . 4 . POLICY PRINCIPLE MEASUREMENT MODELS BY LEVEL OF
SOPHISTICATION, 2000
Standardized factor loadings Low Medium High
ξ1 Limited government:
λ1 Spending scale 1 .47 .56 .66
λ2 Poor people item .74 .74 .76
λ3 Spending scale 2 .51 .43 .67
ξ2 Traditional morality:
λ4 Traditional morality scale 1 .60 .69 .79
λ5 Traditional morality scale 2 .50 .75 .87
Mean factor loading .56 .63 .75
Model fit:
Yuan–Bentler scaled χ2 7.22 20.26 10.76
Degrees of freedom 4 4 4
p-value .12 <.01 .03
CFI .98 .95 .99
RMSEA .04 .09 .06
SRMR .03 .04 .03
Number of observations 509 553 490
Notes: Direct maximum likelihood estimates are based on raw data. Standardized factor loadings
represent the correlation between latent and observed variables. All loadings are significant at p < .05.
Latent and observed variables are coded so that higher scores reflect more support for a given principle;
therefore, each λ should be positive. NES sampling weight has been used. CFI, comparative fit index;
RMSEA, root mean square error of approximation; SRMR, standardized root mean square residual.
ξ1 Limited government:
λ1 Spending scale 1 .78 .73 .77
λ2 Spending scale 2 .48 .52 .74
ξ2 Traditional morality:
λ3 Traditional morality scale 1 .69 .67 .86
λ4 Traditional morality scale 2 .53 .78 .86
ξ3 Military strength:
λ5 Military strength item 1 .61 .65 .81
λ6 Military strength item 2 .66 .79 .91
λ7 Military strength item 3 .30 .50 .61
Mean factor loading .58 .66 .79
Model fit:
Yuan–Bentler scaled χ2 8.51 16.95 34.35
104
TA B L E 5 . 5 . (Continued)
Degrees of freedom 11 11 11
p-value .67 .11 <.01
CFI 1.00 .98 .98
RMSEA .00 .04 .08
SRMR .03 .03 .05
Number of observations 346 368 325
Notes: Direct maximum likelihood estimates are based on raw data. Standardized factor loadings
represent the correlation between latent and observed variables. All loadings are significant at p < .05.
Latent and observed variables are coded so that higher scores reflect more support for a given principle;
therefore, each λ should be positive. NES sampling weight has been used. CFI, comparative fit index;
RMSEA, root mean square error of approximation; SRMR, standardized root mean square residual.
ξ1 Limited government:
λ1 Poor people item .77 .82 .79
λ2 Welfare item .52 .54 .73
λ3 Child care item .57 .62 .71
λ4 Social security item .40 .47 .50
ξ2 Traditional morality:
λ5 Traditional morality scale 1 .54 .56 .87
λ6 Traditional morality scale 2 .57 .88 .81
Mean factor loading .56 .65 .74
Model fit:
Yuan–Bentler scaled χ2 19.70 17.11 23.72
Degrees of freedom 8 8 8
p-value .01 .03 <.01
CFI .97 .99 .99
RMSEA .05 .04 .05
SRMR .03 .03 .03
Number of observations 682 708 697
Notes: Direct maximum likelihood estimates are based on raw data. Standardized factor loadings
represent the correlation between latent and observed variables. All loadings are significant at p < .05.
Latent and observed variables are coded so that higher scores reflect more support for a given principle;
therefore, each λ should be positive. NES sampling weight has been used. CFI, comparative fit index;
RMSEA, root mean square error of approximation; SRMR, standardized root mean square residual.
105
106 ON VOTER COMPETENCE
Notes: Cell entries represent mean Pearson product-moment correlation between all observed
variables.
1988:
Yuan–Bentler scaled χ220 216.42 223.18 376.45
p-value <.01 <.01 <.01
CFI .47 .61 .29
RMSEA .14 .14 .18
SRMR .11 .13 .12
Mean standardized factor loading .30 .34 .35
1992:
Yuan–Bentler scaled χ214 221.38 365.71 288.75
p-value <.01 <.01 <.01
CFI .41 .56 .77
RMSEA .15 .19 .17
SRMR .10 .13 .10
Mean standardized factor loading .28 .39 .53
1996:
Yuan–Bentler scaled χ25 38.74 180.37 145.95
p-value <.01 <.01 <.01
CFI .76 .60 .79
RMSEA .12 .26 .25
SRMR .07 .13 .10
Mean standardized factor loading .37 .52 .64
2000:
Yuan–Bentler scaled χ25 46.30 174.62 289.34
p-value <.01 <.01 <.01
CFI .75 .48 .52
RMSEA .13 .26 .35
SRMR .08 .13 .18
Mean standardized factor loading .37 .43 .51
2004:
Yuan–Bentler scaled χ220 89.32 125.90 213.17
p-value <.01 <.01 <.01
CFI .39 .69 .80
(Continued)
108 ON VOTER COMPETENCE
TA B L E 5 . 8 . (Continued)
Notes: Direct maximum likelihood estimates are based on raw data. Sample sizes are reported in
Tables 5.1–5.6. NES sample weights have been used. CFI, comparative fit index; RMSEA, root mean
square error of approximation; SRMR, standardized root mean square residual.
high knowledge respondents, the χ2 values are all extremely large and
highly significant (p < .00001), every CFI lies below the .95 threshold, all
RMSEA values exceed the .06 cutoff, and the SRMR returns mediocre val-
ues. When the unidimensional estimates are compared to their multidi-
mensional counterparts, the latter’s superiority is self-evident. To take an
example, the 2004 CFI for the high group is .98 in the three-dimensional
model versus .80 in the unidimensional model. Likewise, the RMSEA
equals .08 for the three-factor model compared to .21 for the one-factor
model. The mean factor loadings are also lower in the unidimensional
model compared to the multidimensional model (.66 < .79). Things are
no different in the medium and low knowledge samples. Here too the
multidimensional model bests the unidimensional specification. This, of
course, comes as no surprise given the ideological innocence of much of
the mass public.
To sum up the measurement model results, citizens with various degrees
of political knowledge hold real attitudes toward federal support for strug-
gling Americans, standards of moral rectitude, and the use of military power.
Moreover, my evidence calls into question two rival perspectives. Contrary
to the nonattitudes thesis, the policy opinions of the least informed are struc-
tured coherently and sensibly. Against the ideological sophistication thesis,
the supposition that a global left–right orientation drives opinion for the most
informed receives no support.
The Availability of Policy Principles 109
7
One important point to note is that the standard deviations of the variables are
usually larger in sophisticated groups, which contributes to higher continu-
ity correlations. These differences reflect the tendency of the less informed to
cluster near or at the middle of the scale whereas more informed respondents
report more extreme positions. The assumption is that these distributions arise
because the less informed hold weak or nonexistent attitudes and choose middle
responses to hide this fact (Converse, 1964, 1970; Jennings, 1992).
110 ON VOTER COMPETENCE
rather than know nothing respondents (Achen, 1975; Krosnick & Berent,
1993). Inasmuch as question-induced error is misattributed to survey par-
ticipants, raw correlations represent negatively biased estimates of attitude
stability. The solution is to apply statistical methods that purge the opin-
ion items of random error to generate corrected continuity correlations that
reflect “true” stability.
I argued that random error is a joint product of flawed respondents and
flawed questions. Given this, it seems unwise to insist uncorrected or cor-
rected correlations are superior. The former method underestimates attitude
stability and the latter overstates it. I take a balanced approach here by report-
ing uncorrected and corrected correlations and then taking the mean of these
estimates, which in effect partitions the error equally between respondents
and questions. To the extent that response instability is disproportionately
affected by respondents or questions, my “split the difference” approach is
potentially misleading. Nevertheless, this strategy remains preferable because
it takes more information into account than standard approaches. In brief, the
continuity correlation averages should yield more accurate estimates of atti-
tude stability than either uncorrected or corrected measures.
I rely on data from the 1992–1994–1996 NES panel to estimate the corre-
lations. Each wave contains a sufficient number of limited government and
morality indicators to conduct the tests. The welfare state measure is com-
posed of several of the items used earlier. I tap moral attitudes with the same
items described above. Correlations are calculated for the 1992–1994 and
1994–1996 intervals.
The NES panel does not contain repeated measures of abstract hawk–dove
orientations. Fortunately, Peffley and Hurwitz (1993) have undertaken such
an analysis. They administrated a two-wave panel survey to residents of the
Lexington, Kentucky metropolitan area over a 13-month period and reported
a corrected correlation of .97. An observed correlation was not reported. The
authors did not stratify the sample by political knowledge, educational attain-
ment, or some comparable variable, so it remains unclear whether stability
varies by level of sophistication. However, given the high value of the correla-
tion, any expert novice discrepancies would be marginal.
8
The use of the seven-point scale is widely accepted, but the same cannot be
said for the feeling thermometer items. Some maintain that feeling thermom-
eter scores reflect short-term evaluations rather than attitudinal dispositions.
However, some work suggests the feeling thermometers represent valid and
reliable indicators of deeper partisan attachments. CFA results imply the seven-
point scale and feeling thermometers tap a single underlying factor (Goren, 2005).
Other efforts show that responses to the feeling thermometers nearly match the
112 ON VOTER COMPETENCE
temporal stability of the seven-point scale (Green et al., 2002). Collectively, these
studies suggest that the feeling thermometers serve as good proxies of latent
partisanship.
The Availability of Policy Principles 113
Party identification
1992–1994 .68 .85 .88 .20
1994–1996 .78 .86 .93 .15
Limited government
1992–1994 .56 .67 .68 .12
1994–1996 .52 .65 .67 .15
Traditional morality
1992–1994 .55 .67 .73 .18
1994–1996 .55 .66 .80 .25
as these as supporting the nonattitudes thesis, no one believes the less aware
lack genuine partisan orientations. Unless we are willing to argue a gap of
.20 means partisan identities are doorstep opinions for the unsophisticated
even though their continuity correlation equals .68, the conclusion that gen-
uine attitudes vary in durability across levels of sophistication seems hard to
dispute. Said differently, the fact that partisan identities are sturdier among
the politically sophisticated does not imply that the unsophisticated lack such
ties. Instead, it means partisan orientations are somewhat less crystallized
among the less knowledgeable. We should keep this in mind when evaluating
the remaining correlations.
How does limited government stack up? For 1992–1994, Table 5.9 reveals
correlations of .56, .67, and .68 in the low, medium, and high groups, respec-
tively. The comparable figures for 1994–1996 are .52, .65, and .67. These cor-
relations are moderately strong in the low group and somewhat better in the
higher groups. The high–low discrepancy is .12 in the first interval and .15 in
the second. As expected, then, sophistication promotes stability. But the key
point to take away is that attitudes toward limited government, measured via
a lower bound estimate, appear reasonably solid for most persons. Note lastly
that attitudes toward government activism fluctuate more than partisan iden-
tities, a finding that should surprise no one.
The traditional morality estimates appear in the bottom row of the table.
For 1992–1994, the correlation equals .55 in the low sample, .67 in the middle
sample, and .73 in the high sample. For 1994–1996, the respective marks are
.55, .66, and .80. These results reflect moderate stability in the bottom group
114 ON VOTER COMPETENCE
Party identification
1992–1994 .71 .91 .95 .24
1994–1996 .87 .91 .97 .10
Limited government
1992–1994 .90 .91 .89 –.01
1994–1996 .64 .86 .89 .25
Traditional morality
1992–1994 .70 .83 .85 .15
1994–1996 .86 .83 .93 .07
and considerable stability in the higher two. Next, we can see that the high–
low difference equals .18 in 1992–1994 and .25 in 1994–1996. These gaps imply
that knowledge enhances the stability of moral attitudes, just as it did for the
partisan and economic welfare predispositions. Finally, the moral attitude cor-
relations also lag party identification. Overall, the key points to take away
from Table 5.9 are that partisan identities represent the most durable predis-
position, policy principles hold steady as well, and sophistication promotes
stability. These results reinforce the claim that policy principles are widely
available in the public mind.9
Table 5.10 reports the corrected estimates, which reflect the correla-
tion between latent factors free of measurement error. Here, the operating
assumption is that error arises from problematic questions. Because respon-
dents are absolved of blame, disattenuated correlations can be viewed as
upper bound estimates of attitude crystallization. For starters, latent par-
tisanship proves remarkably enduring. For 1992–1994, the corrected cor-
relation equals .71 in the low group .91 in the middle group, and .95 in the
top group. Over the latter waves, the respective correlations are .87, .91,
and .97. Next, the high–low stability difference is .24 in 1992–1994 and .10
in 1994–1996. These differences suggest knowledge does indeed promote
partisan stability.
9
Part of the difference is because there is more variance in the opinions of aware
respondents.
The Availability of Policy Principles 115
Moving on, I find impressive stability for limited government in all sam-
ples. For 1992–1994, the corrected correlation is .90 among the least informed
portion of the electorate, .91 among the moderately informed, and .89 for the
most informed. In an unanticipated finding, there is no high–low difference
to speak of. For the second period, the correlations are .64, .86, and .89 for the
respective groups. As predicted, the correlation grows larger in more aware
samples. But the most noteworthy discovery is that the correlations are gen-
erally high, which suggests limited government behaves more like an evalu-
ative disposition than a temporary construction. The final point to stress is that
limited government proves less stable than latent partisanship in the second
panel wave, but not the first. For traditional morality9294 the low, medium, and
high continuity correlations are .70, .83, and .85, respectively. For 1994–1996,
the figures are .86, .83, and .93. I draw three conclusions from these results.
First, evaluations of moral behavior are stable for all. Second, political aware-
ness promotes crystallization. Third, the stability of moral views trails that of
party.
Overall, the corrected correlations demonstrate that latent partisanship,
government power, and traditional moralism are sturdy at each level of
sophistication. This, in turn, reaffirms my claim that these are real attitudes
for most of the public. Party identification may be the most enduring orien-
tation, but limited government and moral traditionalism prove durable as
well. Before proceeding, I offer some commentary to put these results in per-
spective. Critics of these techniques sometimes argue that because the unso-
phisticated provide less reliable responses than the sophisticated, corrected
correlations purge more random error from the responses of the less aware
and thus artifactually minimize stability differences. A superficial comparison
of some results in Tables 5.9 and 5.10 lends credence to this view. For instance,
the high–low difference for limited government9294 is .12 for the uncorrected
case and –.01 for the corrected case. But other examples show the high–low
gap widens after correcting for measurement error. This happens for limited
government9496 (uncorrected difference = .15 vs. corrected difference = .25).
When all high–low differences in each table are taken into account, the mean
difference equals .18 for the uncorrected correlations versus .13 for the cor-
rected ones. Although the gap narrows to some degree, the method of purging
correlations of random error does not obliterate these differences. As such, the
corrected correlations should not be dismissed.
As argued above, the attenuated and disattenuated correlations are biased
measures of attitude stability. The uncorrected estimates are biased down-
ward because responsibility for measurement error is placed squarely on
116 ON VOTER COMPETENCE
respondents even though the questions deserve a share of the blame. The cor-
rected estimates are biased upward because the locus of error is presumed
to lie with faulty questions even though individuals are partly responsible.
Given the limits of both methods, my solution is to calculate the mean con-
tinuity correlation. In essence, this move divides the error equally between
respondents and measures. The average correlations in Table 5.11 stand as my
best estimates of attitude stability.
I begin with party identification. For 1992–1994, the party mean assumes
a value of .70 in the low sample, .88 in the medium sample, and .92 in the
high sample. Turning to 1994–1996, the respective correlations are .83, .89, and
.95. In terms of magnitude, five of the six estimates are extremely robust: the
sixth (low9294 = .70) is also impressive. The next point to note is that the high–
low difference equals .22 in period one and .12 in period two. Hence, partisan
identities endure for citizens across the board and, most especially, for the
most informed.
For limited government9294 the mean correlation is .73 in the low group and
.79 in the two higher groups. The corresponding 1994–1996 estimates are .58,
.76, and .78. These are, for the most part, very solid correlations. The high–low
difference is .06 in the first period and .20 in the second, showing anew that
crystallization rises with sophistication. Overall, it seems that for most people
attitudes toward government aid to those with varying degrees of need are
stable belief system elements. I finish up with moral traditionalism. The 1992–
1994 correlation is .63 in the low sample, rises to .75 in the medium group, and
Party identification
1992–1994 .70 .88 .92 .22
1994–1996 .83 .89 .95 .12
Limited government
1992–1994 .73 .79 .79 .06
1994–1996 .58 .76 .78 .20
Traditional morality
1992–1994 .63 .75 .79 .16
1994–1996 .71 .75 .87 .16
Notes: Cell entries represent the mean of the attenuated and disattenuated continuity correlations
from Tables 5.9–5.10.
peaks at .79 in the high group. In 1994–1996 these values are .71, .75, and .87,
respectively. The high–low gap is .16 in both periods, showing that moral pre-
dispositions are firmer in the highest strata. Broadly speaking, moral attitudes
are reasonably to highly stable for the least sophisticated and very stable for
the most sophisticated.
The stability tests have covered a lot of ground, so now is a good time
to tie everything together. First, by the standards used to judge continuity
correlations, my estimates suggest that limited government and traditional
morality represent long-term evaluative dispositions in mass belief systems.
These results imply policy principles are psychologically available regardless
of how much or how little political information citizens have. Second, stabil-
ity is higher for knowledgeable groups. Therefore, although policy principles
reside in attitude hierarchies, their degree of crystallization varies predict-
ably across cognitively heterogeneous populations. Third, party identification
emerges, as always, as the most stable political predisposition. A final caveat
is in order. Readers should note that impressive stability does not mean per-
fect stability. Hence, I make no claim that policy principles are fixed and thus
invulnerable to change. They are persistent, not immutable. They may evolve
over time, but such change occurs at a deliberate pace.
and (7) women’s role. In the top knowledge category 79% answered all seven
questions versus 67% in the medium tercile and 40% in the low knowledge
sample. Comparable patterns exist in other NES surveys. Clearly, item nonre-
sponse is rampant among less informed respondents.
Analysts sometimes address this problem using statistical algorithms
to impute opinions for the policy questions respondents failed to answer.
Ansolabehere et al. (2008) rely on this strategy in their study on attitude sta-
bility and issue voting. Using issue scales constructed from up to 24 items,
they found stability levels are much higher than previously reported and that
sophistication-based differences are marginal. They took respondents who
answered at least 75% of the issue items and imputed preferences for their
unanswered questions. In light of the missing data patterns just described,
it seems likely that preference imputation occurs most frequently among the
least aware. A critic might charge that their stability estimates are at least in
part an artifact of the imputation process. To the extent that such respondents
do not belong on the scale at all, the imputation of preferences may be contro-
versial (Converse 1964, 1970).
Rather than taking a position on the merits of this particular imputation
strategy, the points I wish to stress are that, with one exception, levels of non-
response are low for the items used in my analyses and, most critically, are
largely unaffected by sophistication. Table 5.12 reports the percentage of each
sample in each year taking positions on all measures. Some examples illu-
minate the point. First, in 1992 84% in the low knowledge group answered
all 12 items compared to 91% in the middle and high groups. Now, there is
a seven-point difference between the high and low groups, but it is a small one
at that. Moreover, it is tiny compared to the 39 point difference for the seven
issue scales described above. 1992 is not unique. In 2008, 92% of the low and
medium knowledge respondents answered the eight policy principle ques-
tions versus 94% in the high sample. The only exception to this pattern is in
2004 where 93% of the high group answered all 11 items versus 65% for the
low group, a 28-point difference. The problem here is that a hawk–dove item
contains a “don’t know” filter that many of the unsophisticated utilized. In
any case, Table 5.12 shows that subjects answer most of the questions and, as
seen earlier in the chapter, when they do the responses of the uninformed are
as meaningful as those of the informed.
To sum up, my supposition that political novices hold genuine principles is
not compromised by excessive missing data. In conjunction, the factor analy-
sis estimates, the continuity correlation results, and the opinion holding rates
point to the same conclusion. Core principles are very widely available in
political belief systems.
The Availability of Policy Principles 119
1988: 13 items 80 86 91 11
1992: 12 items 84 91 91 7
1996: 10 items 93 95 96 3
2000: 9 items 92 90 91 –1
2004: 11 items 65 86 93 28
2008: 8 items 92 92 94 2
Notes: Cell entries represent percentage of respondents giving valid responses to all policy principle
items.
CONCLUSIONS
If policy attitudes are to shape electoral choice, they must reside in long-
term memory. In Chapters 2–4 I argued that policy principles are embed-
ded deeply in political debate and the party system, are invoked regularly
during presidential campaigns, and are readily comprehensible. Because of
this, the American public receives sufficient exposure to evaluate the ideas,
encode their impressions in memory, and retrieve them later on. This chap-
ter examined whether groups at different levels of sophistication hold such
principles. My analysis of opinion data from multiple NES surveys covering
a 20-year period yields an unequivocal verdict. The unsophisticated, mod-
erately sophisticated, and highly sophisticated hold real attitudes toward
limited government, conventional morality, and military power. In the Eagly
and Chaiken framework, policy principles are better characterized as dura-
ble evaluative dispositions rather than temporary evaluations constructed
on the spot. In the language of Converse, core principles fall much closer
to the attitude end of the attitude–nonattitude continuum. Because most
folks hold policy principles, they satisfy the availability condition for policy
voting.
This does not mean sophistication has no impact on the quality of domain-
specific principles. Because the politically sophisticated think about govern-
ment and politics more often than the unsophisticated, their policy attitudes
should be more crystallized. This is precisely what I have found. The opinion
reports of the informed correlate a bit more robustly with the latent attitudes
they are designed to measure. Likewise, although these attitudes are stable for
almost everyone, stability is higher among the more knowledgeable. Without
120 ON VOTER COMPETENCE
Limited Government
Density
Traditional Morality
Density
Progressive Traditional
Note: Range [0–1], Mean = .56, Stdev = .22
Military Strength
Density
Dove Hawk
Note: Range [0–1], Mean = .60, Stdev = .22
Source: 2004 NES
gainsaying these two findings, the takeaway point is that policy principles are
long-term forces in the minds of all citizens.
Before proceeding, it should prove helpful to view the distribution of
public opinion on each dimension. Figure 5.2 stacks smoothed histograms
(kernel densities) for collective opinion on limited government, traditional
morality, and military strength in 2004. All variables lie on a 0–1 range and
The Availability of Policy Principles 121
Limited Government
Density
Traditional Morality
Density
Progressive Traditional
Note: Range [0–1], Mean = .59, Stdev = .22
Military Strength
Density
Dove Hawk
Note: Range [0–1], Mean = .56, Stdev = .22
Source: 1992 NES
are scored so that higher values reflect right-wing opinion. For limited gov-
ernment, Figure 5.2 reveals a left leaning public, one that on average prefers
more government to less. This can be seen in the shape of the distribution,
the bulk of which falls in the liberal end of the scale, and the limited govern-
ment mean, which equals .31. Next, note the American public leans to the
122 ON VOTER COMPETENCE
123
124 ON VOTER COMPETENCE
making up their minds about political choices” (p. 25). Zaller (1992) concurs:
“the impact of people’s value predispositions always depends on whether
citizens possess the contextual information needed to translate their values
into support for particular policies or candidates” (p. 25). Whereas Sniderman
and colleagues emphasize liberal-conservative orientations, Zaller concep-
tualizes value predispositions more broadly to include ideological and core
principles.
What does Zaller find? For the most part he lacks direct measures of
domain-specific principles on the NES surveys he uses to study opinion
change. To get around this problem he relies on liberal–conservative scales
as proxy measures for value predispositions under the assumption that
ideological self-placements correlate with values (Zaller, 1992, pp. 26–28,
344–345). The potential difficulty with this approach is that, as seen in
Chapter 4, the unsophisticated have little idea what ideological labels mean,
so it seems doubtful that liberal–conservative scales reflect value disposi-
tions for these respondents. This is precisely what I found in prior work
(Goren, 2001, pp. 162–163).
To underscore the point using the data at hand, Table 6.1 reports Pearson r
correlations between respondents’ self-placements on the seven-point liberal-
conservative scale and my three policy principle scales for low, medium,
and high sophistication respondents in the 1988–2008 NES surveys. Higher
values reflect increasingly conservative opinion on all measures, so positive
associations should be the rule. If the Zaller measurement strategy is defen-
sible, the correlations should be about the same across samples. The data
show instead that the associations grow stronger moving from less to more
informed samples. For instance, in 1988 the mean correlation between the lib-
eral–conservative and policy variables increases from .14 in the low group to
.21 in the middle group to .49 in the high group for a high–low difference of
.35. Likewise, in 2004 the respective correlation means equal .07, .36, and .66,
yielding a high–low gap of .59. Given the feeble values in the low knowledge
group, critics might question whether the lack of principle-based reasoning
among the unaware in Zaller’s study results from their lack of knowledge
or the lack of viable measurement instruments. In short, although this work
demonstrates that awareness motivates reliance on liberal–conservative orien-
tations, it cannot speak as directly to the question of whether the same holds
true for other abstract principles.
What happens when the sophistication interaction model is tested
using valid measures of core principles? Studies in this vein, which rely
on cross-sectional data, suggest political novices ground preferences
in crowning postures (Goren, 2004; Pollock, Lilie, & Vittes, 1993). This
The Centrality of Policy Principles 127
1988:
Limited government .14 .19 .53 .39
Traditional morality .18 .30 .46 .28
Military strength .10 .14 .48 .38
Mean .14 .21 .49 .35
1992:
Limited government .03 .27 .57 .54
Traditional morality .08 .35 .55 .47
Military strength .03 .20 .23 .20
Mean .05 .27 .45 .40
1996:
Limited government .13 .36 .61 .48
Traditional morality .07 .43 .61 .54
Military strength .06 .07 .16 .10
Mean .09 .29 .46 .37
2000:
Limited government .00 .34 .51 .51
Traditional morality .18 .46 .55 .37
Mean .09 .40 .53 .44
2004:
Limited government .02 .24 .55 .53
Traditional morality .20 .45 .74 .54
Military strength .00 .39 .70 .70
Mean .07 .36 .66 .59
2008:
Limited government –.04 .19 .49 .53
Traditional morality .05 .30 .60 .55
Mean .01 .25 .55 .54
variable, controlling for the simultaneous influence of the other. Next, the
structural coefficients β2, β4, β6, and β8 denote the effect of one latent fac-
tor on another at a given point in time, ceteris paribus. If policy principles
drive issue preferences, limited governmentt will predict health caret, con-
trolling for health caret – 1 whereas health caret will not affect limited gov-
ernmentt, controlling for limited governmentt – 1. Conversely, if issues shape
principles, health caret will constrain limited governmentt, controlling for
limited governmentt – 1 and limited governmentt will not sway health caret,
holding issue prefrencest – 1 constant. All variables vary from zero to one,
which means the coefficients can be interpreted as the percentage change in
the dependent variable given movement across the full range of the inde-
pendent variable, holding the other factor constant. Finally, note that equa-
tions (1)–(4) reflect a synchronous specification. Comparable results emerge
when lagged values of principles (issues) are used to predict current values
of issues (principles).
Table 6.2 reports the model fit statistics and unstandardized parame-
ter estimates for the structural equations. To avoid clutter, measurement
model coefficients are omitted. The first column identifies the stability
coefficients, the structural coefficients, and fit statistics. The 1992–1994
estimates appear in column two, followed by the 1994–1996 estimates in
column three. To begin with global fit, the specified model reproduces the
data exceptionally well. The χ2 statistic is insignificant and the compara-
tive fit index (CFI), root mean square error of approximation (RMSEA), and
standardized root mean square residual (SRMR) values surpass the desired
cut points.
Moving onto the parameter estimates, I find that limited government is
quite stable over the first wave (β1 = 0.74) and remarkably stable over the
second (β3 = 0.93). In contrast, health care preferences fluctuate a lot between
1992 and 1994 (β5 = 0.48) before stabilizing thereafter (β7 = 0.69). Formal testing
reveals limited government is probably more durable than health care over
the first panel wave (p = .06) if not necessarily the second (p = .12). In terms
of the structural coefficients, limited government affects health care attitudes
in 1994 but not 1996.1 In 1994, the strongest advocates for small government
1
The health care reform issue received tremendous media coverage from 1992 to
1994, only to be ignored afterward. As a colleague and I report elsewhere, 561
health care reform stories appeared on major television networks during the first
wave of the NES panel versus 10 during the second (Dancey & Goren, 2010).
Given these coverage patterns, the finding that people updated only during
a time of intense political debate makes sense.
130 ON VOTER COMPETENCE
1992–1994 1994–1996
Stability coefficients:
Limited government 0.74* 0.93*
(.12) (.20)
Health care preference 0.48* 0.69*
(.12) (.13)
Structural coefficients:
Health care → Limited government 0.06 0.05
(.08) (.13)
Limited government → Health care 0.72* 0.24
(.21) (.25)
Model fit:
Yuan–Bentler scaled χ2 25.18
Degrees of freedom 18
p-value .12
CFI .99
RMSEA .03
SRMR .03
Number of observations 412
score far more conservative on health care than government defenders, hold-
ing lagged health care preferences constant (p < .01). The limited government
coefficient remains sizable in 1996, but falls short of statistical significance. In
contrast, health care shapes limited government in neither interval. The struc-
tural coefficients cannot be reliably distinguished from zero and are substan-
tively trivial. Put simply, limited government dynamically constrains health
care preferences without being constrained by them. The broader implica-
tions should be clear. Attitudes toward limited government lie closer to the
belief system core whereas health care preferences probably reside further
out in the periphery.
The Centrality of Policy Principles 131
TA B L E 6 . 3 . TRADITIONAL MORALITY–ABORTION
OPINION MODELS, 1992–1994–1996
1992–1994 1994–1996
Stability coefficients:
Traditional morality 0.78* 0.94*
(.06) (.07)
Abortion 0.98* 0.88*
(.06) (.06)
Structural coefficients:
Abortion → Traditional morality –0.05+ 0.03
(.04) (.04)
Traditional morality → Abortion –0.02 –0.16*
(.08) (.08)
Model fit:
Yuan–Bentler scaled χ2 18.79
Degrees of freedom 20
p-value .54
CFI 1.00
RMSEA .00
SRMR .01
Number of observations 490
+
p < .10; *p < .05 (one-tailed).
(Zaller, 1992). The less extreme version of the hypothesis agrees that sophisti-
cation sharpens principle-opinion ties, but leaves open the possibility that the
unaware manipulate core principles, albeit with less felicity than their more
sophisticated counterparts.
One seemingly straightforward way to test these ideas would be to rees-
timate the dynamic constraint models for respondents at different levels of
sophistication. Despite its intuitive appeal, this is not a viable strategy for
a few reasons. First, the samples contain about 400–500 observations and
thus are of marginal utility for generating subgroup estimates via struc-
tural equation modeling techniques. Were I to trichotomize my data as
before, the samples would contain about 150 or so cases per group, too few
to generate stable and efficient parameter estimates. Second, the absence
of hawk–dove items precludes testing in the foreign policy and national
security domain. Fortunately, because the results reported in Tables 6.2
and 6.3 point to the temporal priority of abstract principles, I can use the
1988–2008 NES cross-sectional surveys to estimate the statistical models
without having to worry about simultaneity bias.
My analyses revolve around relationships between principles and issue
preferences for which the temporal ordering has been established above or
elsewhere (Peffley & Hurwitz, 1993). Although I would anticipate observing
comparable results for other issues, space limitations dictate a more focused
approach in these pages. The first test features the limited government to
health care link. To begin with the dependent variable, I tap health care opin-
ion using the seven-point scale described above with higher scores denoting
preferences for private insurance plans. Next, limited government is measured
using whatever combination of spending items appears on a given survey.
After identifying the items, I calculate a “person mean” score for respondents
who answered over half the questions used to tap a given attitude (minimum
of three questions asked). To illustrate, five applicable items appear on the
1988 NES. For individuals who answered three of the five, their limited gov-
ernment score equals the mean response for the three items. Similarly, par-
ticipants who answered four questions receive a score based on the average
of those four responses, whereas those who answered all five questions have
their positions calculated accordingly. Respondents who answered less than
half the items are treated as missing data. This procedure allows me to preserve
a small number of cases that would otherwise be dropped from the analysis.
The person mean approach works well when the items tap a single latent con-
struct, scale reliability is high, and over half the available items are employed
(Bernaards & Sijtsma, 1999; Schafer & Graham, 2002). All three conditions
134 ON VOTER COMPETENCE
are satisfied here.2 Because higher values on limited government and health
care reflect increasingly conservative sentiments, the variables should be pos-
itively related in the statistical models, which are estimated using ordinary
least squares (OLS) regression.
A standard set of control variables is incorporated into the equations. First,
I code “65 or older” 1 for individuals who meet this criterion and 0 otherwise.
Because most such respondents have Medicare coverage, they should react
coolly to private market schemes. The regression coefficient should thus be
negative. Next, “unemployed” is coded 1 for unemployed respondents and 0
otherwise. Because the unemployed typically lack health insurance, they too
might express qualms about private insurance. A negative coefficient is antici-
pated here as well. Third, party identification is gauged using the seven-point
self-categorization scale. Higher values denote stronger GOP attachments, so
the regression coefficient should be positive. Next, liberal–conservative atti-
tudes are tapped with the seven-point liberal–conservative self-placement
item. As I argued in Chapter 4, sophistication should drive the use of liberal–
conservative attitudes. To test this, the knowledge indicator and its multiplica-
tive product with the left–right scale are added to the model. The coefficient for
the constitutive liberal–conservative term summarizes this variable’s impact
on health care opinion when the sophistication variable equals zero—the
minimum knowledge score.3 Because the least informed usually do not hold
meaningful liberal–conservative attitudes (see Chapter 4) this term should be
insignificant. In contrast, the multiplicative term should be positive and signif-
icant, indicating that the bond between conservative affinities and preferences
for private insurance strengthens with sophistication. Fifth, I tap political trust
using the standard multi-item scale. Following Hetherington (2005), I expect
2
I apply this coding rule to all other multiple indicator attitude scales composed
of three or more items. In each case, the items tap a single concept, the scale is
reliable, and over half the available items are employed in the construction of the
scale. Recall from Chapter 5, Table 5.12, that rates of item nonresponse are, for
the most part, comparable across levels of sophistication. Therefore, the person
mean approach does not routinely drop less sophisticated respondents from the
estimation samples. Note finally that the health care preference variable, along
with the defense spending item used later on, include no opinion filters that
screen out some less sophisticated respondents.
3
Constitutive terms are often interpreted as the main or average effect of the
predictor in question. This is a mistake. These coefficients, again, estimate the
predictor’s effect on the dependent variable when its interaction partner equals
zero, ceteris paribus (Friedrich, 1982).
The Centrality of Policy Principles 135
that increasing trust in government will undercut support for private insur-
ance, so the trust coefficient should be negative.
The key question, to reiterate, is whether the sophisticated alone call
upon policy principles to guide their issue evaluations. To test this, a lim-
ited government × sophistication multiplicative term is added to the base-
line to create a second model. If the stronger version of the sophistication
interaction hypothesis applies, the coefficient for the multiplicative term
will be positive and significant whereas that of the constitutive limited gov-
ernment variable will be insignificant. The latter result would suggest that
the least aware fail to ground health care preferences in abstract postures
about government. If the weaker sophistication prediction holds, the con-
stitutive and multiplicative terms will jointly achieve significance. In that
case, the unaware would make use of limited government, but not to the
same degree as the aware.
Table 6.4 reports the 1988–1996 model estimates. Table 6.5 does the same
for 2000–2008.4 Model 1 includes all variables except the limited government ×
sophistication term, which appears in model 2. Because all variables are scored
on a 0–1 range, the coefficient for each additive term can be interpreted as the
percentage difference on health care opinion between those with the mini-
mum and maximum value on the predictor in question, holding all else con-
stant. The constitutive terms should be interpreted as described above. The
multiplicative terms are elaborated below.
To start with the control variables, age and unemployed perform poorly,
whereas party has the predicted effect in five of the six equations. The trust
variable is usually significant as well, though signed incorrectly, the posi-
tive coefficients implying political trust motivates support for private plans.
Next, as predicted, sophistication stimulates ideological evaluation. The
politically engaged usually rely more on ideological leanings than the unen-
gaged. When conditional slope estimates are calculated for respondents one
4
The NES uses a complex, multistage area probability design to sample respon-
dents from the population. Because such designs affect the accuracy of parameter
estimates and standard errors and the conventional formulas for estimators and
standard errors assume the data are collected using simple random sampling
techniques, adjustments should be made to get the point estimates and standard
errors right (Heeringa, West, & Berglund, 2010). I do so here by incorporating
postelection weight, cluster, and strata variables into the estimation. There is no
weight variable in the 1988 NES, but clustering and strata information are taken
into account.
TA B L E 6 . 4 . LIMITED GOVERNMENT AND HEALTH CARE OPINION, 1988–1996
+
p < .10; *p < .05 (one-tailed).
Notes: Ordinary least squares (OLS) regression estimates. Standard errors are in parentheses (adjusted for complex sample design). All variables are on a 0–1
range. Higher values on health care reflect conservative opinion.
Constant 0.12 +
0.12 +
0.24* 0.29* 0.17* 0.18*
(.09) (.09) (.08) (.08) (.07) (.08)
65 or older 0.03 0.03 0.06* 0.07* 0.04+ 0.05+
(.03) (.03) (.03) (.03) (.03) (.03)
Unemployed –0.03 –0.03 –0.12* –0.12* –0.02 –0.03
(.06) (.06) (.04) (.04) (.04) (.04)
Party identification 0.14* 0.14* 0.04 0.04 0.17* 0.17*
(.04) (.04) (.05) (.05) (.05) (.05)
Liberal–conservative identification 0.03 0.03 –0.04 0.01 0.05 0.05
(.15) (.16) (.14) (.13) (.13) (.13)
Political trust 0.10* 0.10* 0.14* 0.14* 0.09* 0.09*
(.05) (.05) (.04) (.04) (.03) (.04)
Limited government 0.39* 0.38* 0.22* –0.08 0.37* 0.31*
(.05) (.12) (.06) (.15) (.07) (.14)
Political sophistication 0.04 0.04 –0.17* –0.25* –0.18* –0.20*
(.13) (.12) (.10) (.10) (.10) (.11)
Liberal conservative id × sophistication 0.24 0.23 0.59* 0.46* 0.37* 0.35*
(.22) (.23) (.19) (.19) (.18) (.18)
Limited government × sophistication 0.01 0.50* 0.09
(.17) (.20) (.22)
F statistic 27.82* 24.59* 49.18* 48.72* 37.32* 32.90*
R2 .20 .20 .24 .24 .27 .27
Number of observations 1289 1289 1039 1039 861 861
Notes: OLS regression estimates. Standard errors are in parentheses (adjusted for complex sample design). All variables are on a 0–1 range. Higher
values on health care reflect conservative opinion.
1.00
0.75
0.50
0.25
1996 2008
0.00
Oppose Favor
Limited Government
Figure 6.1. Limited Government and Health Care Opinion, Additive Effects.
Source: Table 6.4-6.5 estimates; other predictors held at mean/mode. Effects significant.
The Centrality of Policy Principles 141
1.00
0.75
0.50
0.25
Low Sophistication High Sophistication
0.00
Oppose Favor
Limited Government
Figure 6.2. Limited Government and Health Care Opinion, 1992 Interactive Effects.
Source: Table 6.4 estimates; high/low soph = +/– 1 sd from mean; other predictors held at
mean/mode. Effects significant.
142 ON VOTER COMPETENCE
government is over twice that of party (0.37 > 0.17). Of course, visual inspection
cannot determine whether such a difference exceeds what might be expected
by chance. To see if the difference is statistically significant, I constrain the two
coefficients to be equal and assess the test result.5 The test statistic indicates
the null hypothesis of equal effects can be rejected with a high degree of con-
fidence (p < .05). Put otherwise, the 0.37 coefficient for limited government is
demonstrably larger than the 0.17 coefficient for party. The same holds true for
the other 2 years in which a significant main effect arises. Limited government
has a greater impact on health care opinion than party in 1996 (0.30 > 0.14, p <
.01) and 2000 (0.39 > 0.14, p < .001). The standardized coefficient for limited
government also surpasses that of party identification for all three of the years
(.22 > .16 in 1996; .24 > .13 in 2000; .22 > .18 in 2008).
Turning next to the equations with a significant limited government ×
sophistication term, I calculate conditional slope estimates for limited
government for those plus or minus one standard deviation from the
knowledge mean and compare these to the party coefficient. Among the
less informed, the limited government coefficient exceeds that of parti-
sanship in 1988 (0.27 > 0.12) and 1992 (0.24 > 0.14), but these differences
are not quite significant (p88 = .15 and p92 = .08). In 2004 neither variable
systematically affects health care opinion among novices. The standard-
ized betas tell a similar story as limited government performs marginally
better than party (.16 > .13 in 1988, .16 > .15 in 1992, .04 < .05 in 2004). For
those with above average knowledge, the regression coefficient for lim-
ited government tops party at p < .01 in 1988 (0.47 > 0.12), 1992 (0.43 >
0.14), and 2004 (0.33 > 0.04). The standardized coefficients reinforce the
point (.29 > .13 in 1988, .29 > .15 in 1992, .24 > .05 in 2004). Collectively,
these results show that for the less sophisticated limited government
either equals or, more often, surpasses the party cue, and it always bests
party among the sophisticated. Party identification may indeed be the best
5
The test result is not influenced by the fact that movement from 0 to 1 spans
the full range of the variables. If limited government and party identification
are scaled another way (e.g., each set to a seven-point scale), the test evalu-
ates whether a one unit change on the two predictors is statistically equivalent.
Although the magnitude of the unstandardized regression coefficient will vary
depending on the variables’ ranges, for either scaling the equality constraint test
result will be identical. Some may prefer to assess the relative importance of dif-
ferent predictors using standardized regression coefficients. Although there are
problems with this approach (see Achen, 1982), I comment on the standardized
coefficients for interested readers.
The Centrality of Policy Principles 143
predictor of many things, but it is not the best predictor of health care
opinion.6 These results underscore limited government’s centrality in mass
belief systems.
Some might wonder how limited government stacks up against the
trust heuristic. As seen above, political trust significantly affects health
care opinion but in the wrong direction. Those who trust government
evaluate private insurance more favorably than the cynics. When I con-
strain the limited government and trust coefficients to be equal across the
various equations, trust always lags limited government in sophisticated
and unsophisticated samples (with one exception) at conventional levels of
significance (the same holds true when standardized betas are compared).
What about ideological labels? Here too, limited government matches or
surpasses the force of liberal–conservative attitudes among politically
informed respondents. To be precise, the limited government coefficient
is significantly larger than the liberal–conservative analogue in two equa-
tions, approaches a significant difference in a third, and is not significantly
different in the remaining three. With respect to the standardized coeffi-
cients, limited government exceeds the liberal–conservative variable, at
times substantially, from 1988 to 2000, and lags it a bit in 2004 and 2008.
In sum, limited government often exerts more influence over health care
attitudes than other notable cues.
On the whole, the evidence buttresses my argument that limited govern-
ment serves as an effective heuristic for politically sophisticated and unsophis-
ticated individuals. By implication, the strong version of the sophistication
interaction model, which posits that the unaware cannot deduce issue attitudes
from broader principles, commands little support. However, the results lend
some credence to the weaker statement of this model. Although the absence of
sophistication does not preclude the use of core principles, its presence often
helps. Said otherwise, limited government shapes health care opinion most of
the time, but sometimes carries more weight for the better informed.
6
My account focuses on direct effects alone and thus ignores the potential indirect
influence party may wield over issue preferences through its effects on limited
government. I take up the effects of party ties on policy principles in Chapter 7,
but I note here that basic human values weigh more heavily on policy principles
than partisanship.
144 ON VOTER COMPETENCE
(continued)
TA B L E 6 . 6 . (Continued)
+
p < .10; *p < 0.05 (one-tailed).
Notes: Ordered logistic regression estimates for 1988 and 1996. OLS regression estimates for 1992. Standard errors are in parentheses (adjusted for complex
sample design). All variables are on a 0–1 range. Higher values on abortion reflect pro-choice opinion.
(continued)
TA B L E 6 . 7 . (Continued)
+
p < .10; *p < .05 (one-tailed).
Notes: OLS regression estimates. Standard errors are in parentheses (adjusted for complex sample design). All variables are on a 0–1 range.
Higher values on abortion reflect pro-choice opinion.
1.00
Support for Abortion Rights
0.75
0.50
0.25
Figure 6.3. Traditional Morality and Abortion Opinion, 2004 Interactive Effects.
Source: Table 6.7 estimates; high/low soph = +/– 1 sd from mean; other predictors held at
mean/mode. Effects significant
150 ON VOTER COMPETENCE
0.75
0.50
0.25
Figure 6.4. Traditional Morality and Abortion Opinion, 2008 Interactive Effects.
Source: Table 6.7 estimates; high/low soph = +/– 1 sd from mean; other predictors held at
mean/mode. Effects significant.
+
p < .10; *p < .05 (one-tailed).
Notes: OLS regression estimates. Standard errors are in parentheses (adjusted for complex sample design). All variables are on a 0–1 range. Higher values on
defense spending opinion reflect prospending sentiment.
spending than Democrats. The female, born again Christian, and authoritari-
anism variables behave as predicted roughly 50% of the time. Second, sophis-
tication conditions reliance on liberal–conservative predispositions in three
of four models. When conditional slope coefficients are calculated, liberal–
conservative attitudes are insignificant for respondents one standard devia-
tion under the sophistication mean and highly significant for those a standard
deviation above it. As always, relatively sophisticated people take advantage
of liberal–conservative cues.
What about the hawk–dove heuristic? Here I find the sophisticated and
unsophisticated rely equally on views of American power to guide their
defense spending evaluations. Across the equations the military strength coef-
ficient is correctly signed and statistically significant in model 1 whereas the
military strength × sophistication term falls short in model 2. The estimates
lead to the conclusion that the majority of the American electorate puts eval-
uations of military power to good use. Substantively, the military strength
effect proves quite potent. To take a pair of examples, extreme hawks evaluate
defense spending 36% more favorably than extreme doves in 1988 and 46%
more positively in 2004, all else being constant. The effect sizes for 1988 and
2004 are illustrated graphically in Figure 6.5. Next, citizens rely more heavily
on principle than partisanship. In all 4 years the hawk–dove coefficient for-
mally eclipses that of party identification. Military strength also outperforms
authoritarianism and, among the sophisticated, liberal–conservative identi-
fications. Finally, the standardized regression coefficient for the hawk–dove
variable is always significantly larger than that of party, authoritarianism,
and, among respondents one standard deviation above the knowledge mean,
liberal–conservative identities (except for 1996). Put simply, military strength
represents the key ingredient in the derivation of defense spending opinion.
1.00
Support for Defense Spending
0.75
0.50
0.25
1988 2004
0.00
Oppose Favor
Military Strength
Figure 6.5. Military Strength and Defense Spending Opinion, Additive Effects.
Source: Table 6.8 estimates; other predictors held at mean/mode. Effects significant.
The Centrality of Policy Principles 155
CONCLUSIONS
The first condition of policy voting holds that if policy attitudes are to shape
candidate choice they must be present in mass belief systems. As demonstrated
in Chapter 5, attitudes toward core principles are available in the minds of
nearly everyone. The second condition holds that among available attitudes
those that play the part of central heuristics are more likely to guide electoral
choice than peripheral attitudes. So far as individuals grow accustomed to
using policy principles to craft evaluations of short-term controversies, these
attitudes may be applicable to candidate selection. In this chapter, my exami-
nation of panel data attests to the fact that policy principles shape issue pref-
erences in a top-down manner without being simultaneously influenced by
them, whereas my analysis of cross-sectional data indicates that experts and
novices rely heavily on these principles to infer issue positions. And although
sophistication sometimes promotes heuristic reasoning, a lack of sophistication
rarely forecloses it. What is more, core principles seem to wield more power
than the usual suspects in citizens’ reasoning chains. As such, my argument
The Centrality of Policy Principles 157
that policy principles function as central heuristics for individuals across the
awareness spectrum rests on a solid empirical foundation.
Having found that the demos satisfy the first two conditions of policy vot-
ing, I am now ready to model the relationship between core principles and
the presidential vote. But before doing so, I take a detour in the next chapter
to study the origins of limited government, traditional morality, and military
strength. Although such an analysis is both interesting and informative in its
own right, it serves a broader normative purpose as well. Though perhaps
willing to concede, based on evidence marshaled so far, that most people hold
real policy principles, the skeptic might charge that said principles reflect little
more than latent partisan biases. If limited government, traditional moralism,
and military strength are shaped primarily by simple-minded cues such as
these, the fact that such principles guide political judgment offers little reas-
surance about the democratic capacities of ordinary citizens. But if policy prin-
ciples reflect first and foremost personal goals, the esteem in which citizens are
held appreciates. The next chapter seeks to sort this all out.
This page intentionally left blank
CHAPTER 7
Chapters 5 and 6 have established that policy principles are widely available
and function as central heuristics in the minds of voters. Where do these prin-
ciples come from? Why do some people hold positive views of government
activism and others hold negative views? Why do some individuals favor the
preservation of traditional family ties whereas others denounce such efforts?
What impels different people to become national security hawks or doves?
In Chapter 4, I sketched a model delineating the process by which citizens
acquire policy attitudes. Such attitudes develop when citizens are exposed
to salient ideas in American political discourse, gather what the ideas mean,
evaluate them, and store their impressions in memory. This model reveals
why core principles develop, but it elides the question of what drives evalua-
tion. This chapter aims to shed some light on these matters.
I argue that two factors are paramount in this process: party identification
and personal values. I invoke work on partisan perception (Campbell et al.,
1960) and basic human values (Schwartz, 1992) to explain how and why these
forces shape policy principles. Data from three new national surveys are used
to test the hypotheses. The results suggest that party and values systemati-
cally affect policy principles for the sophisticated and unsophisticated alike,
with the influence of values outweighing that of party. After this, I pursue
an exploratory analysis of the relationship between human values and party
identification. I posit that values drive partisanship and present some evidence
consistent with this claim. The chapter concludes with a brief discussion of the
broader implications these results have for assessing citizen competence.1
1
Readers may wonder why I do not include liberal–conservative orientations
alongside party and basic values as antecedents of policy principles. There
are two reasons. First, as made plain in Chapter 3, the supposition that ideo-
logical principles structure political evaluations may hold for only 5–10% of
the electorate. A considerably larger share of the public holds attitudes toward
159
160 ON VOTER COMPETENCE
of thumb dictates that people decide based on how they feel about the source
behind the message (McGuire, 1969).
The most salient cue in politics is the party label. Sometimes cues are embed-
ded in political communications, such as when Democratic- or Republican-
sponsored campaign commercials air on television. At other times the partisan
leanings of the source are identified by a news organization, such as when
a reporter announces the positions of a Democratic or Republican Senator.
Cues also reach citizens from copartisans in their social networks. Whatever
the impetus, partisan predispositions, once activated, motivate recipients to
accept or reject messages in a manner indicative of party leanings. If the mes-
sage source and recipient share the same affiliation, the latter accepts the claim
at face value without sweating the details. But if the source and target stand
on opposite sides of the political divide, the recipient will probably reject the
message (Goren, Federico, & Kittilson, 2009).
Given the centrality of partisan identities in the minds of citizens and the
prevalence of party cues in a political environment organized as a two-party
system, Democratic and Republican politicians have considerable latitude to
shape public opinion. Party leaders regularly disseminate simple cues about
party positions that reach many people. These cues may be statements about
party principles or calls for specific government actions that signal broader
party stances. So long as party representatives furnish the electorate with a
steady stream of cues, citizens can simply toe the party line when thinking
about policy cleavages. As we have seen in Chapter 2, the Democratic Party
has long stood for government activism in the economic welfare domain and
the GOP has resisted such demands. Republicans have decried social change
and moral decline whereas Democrats have stressed, with varying degrees of
enthusiasm, the virtues of tolerance. In the national security arena, the GOP
has for the most part favored force over diplomacy; Democrats have proven
more tentative about military engagements. Given this, I hypothesize that
Republicans will evaluate limited government, orthodox morality, and mili-
tary strength more favorably than Democrats.
Stimulation Benevolence
Hedonism
Conformity Tradition
Achievement
SELF CONSERVATION
ENHANCEMENT Power Security
Figure 7.1 The Schwartz Model of Value Relations among the 10 Motivational
Domains.
164 ON VOTER COMPETENCE
2
Hedonism values are subsumed under both openness to change and self-
enhancement.
3
Personal communication with Shalom Schwartz, December 29, 2008.
The Origins of Policy Principles 165
4
Resource constraints precluded inclusion of the full set of value items across
the surveys as described below. I concentrated on what seem likely to be the key
values affecting principles.
166 ON VOTER COMPETENCE
Benevolence –
Universalism – – –
Self-direction + –
Achievement +
Power + +
Security +
Conformity/tradition + + +
Note: + means the value should promote support for the given principle; – means the value should
depress support for the principle. Blank spaces denote untested relationships.
5
I thank Stanley Feldman for bringing this point to my attention.
168 ON VOTER COMPETENCE
in wealth and power between rich and poor.” Again, these measures blend
abstract ideas with identifiable political issues.
This brief discussion indicates that personal values and core political val-
ues are not interchangeable (Schwartz et al., 2010, make the same argument).
Whereas personal values apply broadly to all situations and contexts, core
political values are confined to a single area—politics. Moreover, personal val-
ues are rank ordered in regard to personal priority but political values are
not. Schwartz’s theory of value content and structure specifies a universe of
values and how these relate to one another; political science offers no ana-
logue. Finally, as I will show before long, personal values are measured at a
higher level of abstraction and, thus, unlike political value measures, are not
confounded with public policy. On grounds of theoretical breadth, concep-
tual clarity, and measurement validity, the social–psychological approach has
much to recommend it. I adopt it accordingly.6
6
Before proceeding, I want to emphasize the key insight emerging from work
on core political values. The consistent finding that those who are “innocent of
ideology” instead use political values to constrain their issue preferences repre-
sents a signal contribution to the study of political psychology and public opin-
ion. Ideological innocence does not imply political incompetence, an incredibly
important and heartening finding.
The Origins of Policy Principles 169
desire for variety, what John, Naumann, and Soto (2008, p. 120) have described
as “the breadth, depth, originality, and complexity of an individual’s men-
tal and experiential life.” Next, conscientiousness encompasses dependabil-
ity and reliability, effort, a drive to achieve, and self-control, which add up
to “socially prescribed impulse control that facilitates task- and goal-directed
behavior” (John, Naumann, & Soto, 2008, p. 120). Extraversion denotes a gre-
garious, outgoing, and active disposition, “a preference for companionship
and social stimulation” (McCrae & Costa, 1999, p. 143), and “an energetic
approach toward the social and material world” (John et al., 2008, p. 120).
Agreeableness denotes a tendency to strive for positive interactions and rela-
tions with other people, a willingness to defer to them to minimize interper-
sonal conflict and disagreement, and a “prosocial and communal orientation
toward others” (John et al., 2008, p. 120). Lastly, emotion stability has been
defined as a calm and steady disposition versus being tense, anxious, and
prone to negative emotional states.
So, how do personality traits and basic human values differ? First, traits
are innate dispositions to act in comparable ways across a variety of dif-
ferent contexts and over time, whereas personal values are enduring goals
people learn from the environment. Second, traits vary across people in
their intensity and frequency of occurrence, whereas values reflect differ-
ences in the priorities people attach to desirable end-states or modes of
conduct. Third, traits can be judged positively or negatively (e.g., conscien-
tiousness is typically viewed more favorably than neuroticism), but values
are generally seen as desirable to varying degrees. Fourth, personality traits
appear to develop before personal values in a temporal or causal sequence
(McCrae & Costa, 1999; Roccas et al., 2002). Perhaps the simplest way to
summarize the difference is this: “Traits describe what people are like; val-
ues refer to what people consider important” (Caprara et al., 2006, p. 3). Just
as personal values differ from core political values, so too do they differ
from personality traits.
My point here is not to dismiss the theoretical utility of personality traits for
the study of political judgment. On the contrary, a burgeoning line of research
demonstrates that the Big Five have far reaching implications for understand-
ing public opinion, political engagement, political identities, the strength of
those attachments, and other important variables (Carney, Jost, Gosling, &
Potter, 2008; Gerber, Huber, Doherty, Dowling, & Ha, 2010; Mondak, 2010).
Instead, my point is that personality traits and basic human values are differ-
ent constructs and that they should not be viewed as one in the same. Indeed,
had I adequate measures of trait dispositions, they would be included as inde-
pendent variables in the models of core principles.
170 ON VOTER COMPETENCE
7
KN has assembled a web-based panel in which random digit dialing techniques
are used to recruit representative samples of the U.S. voting age public to partic-
ipate in a few monthly online surveys on various topics (e.g., consumer choices,
entertainment preferences, and health care usage) in exchange for free Internet
access. Because participants are selected using probability sampling techniques
and current population data are used to create poststratification weights that
reduce the effects of nonresponse and noncoverage biases, KN samples are rep-
resentative of the broader U.S. population.
8
I thank William Chittick and Jason Reifler for inviting me to be part of this project.
YouGov uses sample matching techniques to draw “representative” samples from
nonrandomly selected pools of respondents in online access panels, which consist
of internet users who were recruited via banner ads, purchased email lists, and other
devices. The sample matching technique begins by drawing a stratified national
sample from a target population (in this case, the 2006 American Community
Survey). Rather than contacting these individuals directly, which would be pro-
hibitively expensive, YouGov utilizes matching techniques to construct a compa-
rable sample from its existing internet panel. Members of the matched sample are
then contacted and invited to participate in the survey. After selection, the sample
is weighted to match the target population on a series of demographic factors.
Although the matched sample has been drawn from a nonrandomly selected pool
of opt-in respondents, it can in some respects be treated as if it were a random sam-
ple (Vavreck & Rivers, 2008). These matched samples resemble the broader public
on a number of sociodemographic variables; however, because respondents self-
select into the original panel they may differ from the broader public on unmeas-
ured variables such as political interest and awareness.
The Origins of Policy Principles 171
“welfare,” “food stamps,” “aid to the homeless,” and “child care.” Respondents
indicated whether spending should be “increased a lot,” “increased some,”
“kept the same,” “decreased some,” or “decreased a lot.” I calculate an average
or person mean score for each respondent who answered more than half the
available items (see Chapter 6 for a review). The limited government variable is
constructed so that higher scores denote more conservative responses.
I include statistical controls for female (1 = female, 0 = male), African-
American (1 = African American, 0 = other), Hispanic (1 = Hispanic, 0 =
other), dummy variables for the middle and top income terciles (roughly),
and dummy variables for high school graduates (1 = high school graduates
and those with some college, 0 = other) and college graduates (1 = 4-year col-
lege graduate, 0 = other). I hypothesize that women, African-Americans, and
Hispanics will evaluate limited government more critically than men and
non-Hispanic whites, respectively. Moreover, respondents in the middle and
high income groups should hold stronger antigovernment views than those
in the bottom third of the income distribution. Similarly, those in the medium
and high education categories should prove more receptive to smaller govern-
ment than the least educated.
Gilens (1999) and Kinder and Kam (2008) have shown that whites who
endorse negative stereotypes about the work ethic of blacks and Hispanics are
more likely to oppose means-tested welfare programs than those who repudi-
ate stereotypes. The same may hold true with respect to the broader principle
of government activism. To test this claim, I include a stereotype variable in
the model, which is measured using the person mean score for respondents
who answered at least three of four seven-point stereotype scales. The first
two items asked respondents to place “blacks in general” and “Hispanics in
general” on a “hardworking” versus “lazy” scale; a second pair of items did
the same for a “responsible” versus “irresponsible” question (Cronbach α =
.79). Because higher scores indicate antiminority sentiments, the stereotype
and limited government variables should be positively related.
Turning to the key independent variables I tap party identification using
the standard seven-point scale, coded so that higher scores reflect GOP self-
categorization. Compared to Democrats, Republicans should evince greater
hostility toward the welfare state. To capture personal values I employ a vari-
ant of the measurement strategy developed and validated in early research by
Schwartz (1992, 1994). The following paragraph introduced the values battery
to respondents:
You are going to see some statements about things some people consider
important in life. Using a number from 0 to 10, please rate how important
172 ON VOTER COMPETENCE
each one is to you. If it’s one of the absolutely most important things to
you, give it a 10. If it’s one of the least important things, give it a 0. You’re
free to use any number between 0 and 10, but remember, the more impor-
tant something is to you, the higher the number you should give it.
174
The Origins of Policy Principles 175
TA B L E 7 . 3 . (CONTINUED)
+
p < .10; *p < .05 (one-tailed).
Notes: Ordinary least squares (OLS) regression estimates. Unstandardized coefficients are reported.
HC3 standard errors are in parentheses. All variables are on a 0–1 range. Higher values on limited
government reflect a conservative opinion. Number of observations = 1452. Knowledge Networks
sampling weights are used.
p < .05). Power values do not appear to constrain attitudes either way (b =
0.01, ns). Overall, the regression estimates suggest that basic human values,
universalism and conformity especially, play a central role in shaping bottom
line judgments about how much government should try to spare citizens eco-
nomic hardship.
To convey effect size, Figure 7.2 plots the predicted limited government
position across the range of scores for universalism and conformity, the dom-
inant values acting on opinion. Bear in mind that because relatively few cases
fall in the bottom end of the value distributions this plot essentially illustrates
the hypothetical maximum effect of a given value on the principle, ceteris pari-
bus (the same holds true for most plots in the chapter). We can see that the
more importance citizens attach to universalism the less they care for limited
government. Going from the lowest to highest importance rating on univer-
salism reveals that support for limited government falls some 38%. The figure
further demonstrates that those most wedded to conformity values score 20%
higher on limited government versus the least committed.
The third regression model tests whether the value principle relation-
ship grows stronger among the better educated. Here, the evidence is mixed.
The null hypothesis that the value × education terms simultaneously equal
zero can be rejected (F = 2.37, p < .001), whereas the R2 rises, albeit modestly,
from .29 to .31. Yet only two of the 10 interactions are correctly signed and
176 ON VOTER COMPETENCE
1.00
Support for Limited Government
Universalism
0.75
0.50
0.25
Conformity
0.00
Low High
Importance of Value
I will describe some people. Please tell me how much each person is or is
not like you. Is this person very much like you, like you, somewhat like
you, a little like you, not like you, or not like you at all?
+
p < .10; *p < .05 (one-tailed).
Notes: OLS regression estimates. Unstandardized coefficients are reported. HC3 standard errors are in
parentheses. All variables are on a 0–1 range. Higher values on traditional morality reflect a conservative
opinion. Number of observations = 1000. Knowledge Networks sampling weights are used.
179
180 ON VOTER COMPETENCE
Universalism
0.75
Self-direction
0.50
0.25
Conformity/tradition
0.00
Low High
Importance of Value
9
I remind readers that one of the items used to construct my dependent variable
reads as follows: “This country would have many fewer problems if there were
more emphasis on traditional family ties.” Two items in the conformity/tradition
scale mention tradition as well (“Tradition is important to her. She tries to respect
tradition” and “She thinks it is important to preserve traditional ideas of right
and wrong”). The value and opinion items differ in that the value measures ask
respondents to rate how much a hypothetical other for whom a given normative
belief serves as a guiding principle in his or her life is like them, thus tapping
personal priorities. The “family ties” question asks respondents how much they
agree or disagree with the sentiment expressed above and thereby more readily
evaluates attitudes. Because the word “tradition” appears in both the dependent
and independent variables, some might wonder if the observed value–principle
relationship is an artifact of semantic overlap. I tested this possibility by dropping
the “traditional family ties” item from the dependent variable scale and reesti-
mated the model. The coefficient for the conformity/tradition variable declines
from 0.54 to 0.49 and remains highly significant. I then dropped the two personal
value items mentioning tradition and modeled the four-item opinion scale as a
function of this pared down conformity scale and the other predictors. The value
coefficient remains powerful statistically and substantively (b = 0.34, p < .001).
The Origins of Policy Principles 181
each value × education term falls short of standard or even marginal levels of
significance. If we can place some confidence in education as a sophistication
proxy, then these results infirm the hypothesis that sophistication enhances
value-driven constraint. Instead, the evidence suggests once again that funda-
mental human values—especially universalism and conformity/tradition—
shape policy attitudes throughout the general public.
Some people think that in dealing with other nations our government
should be strong and tough. Suppose these people are at one end of this
scale, at point number 1. Others think that our government should be
understanding and flexible. Suppose these people are at the other end, at
point 7. And, of course, other people have opinions somewhere in between
at points 2, 3, 4, 5, or 6. Where would you place yourself on this scale?
Some people believe the United States should solve international problems
by using diplomacy and other forms of international pressure and use mil-
itary force only if absolutely necessary. Suppose we put such people at 1
on this scale. Others believe diplomacy and pressure often fail and the U.S.
must be ready to use military force. Suppose we put them at number 7.
Where would you place yourself on this scale?
The additive scale is coded so that the minimum score (0) represents the most
dovish position and the maximum score (1) represents the most hawkish
(Cronbach α = .48). Controls for sex, African-American, Hispanic, education,
and party identification, all measured as above, are included in the model.
I expect the force option attracts less support from women than men, from
minorities than whites, from the more educated relative to the less educated,
and from Democrats compared to Republicans.
I use the Schwartz PVP format to measure values. Each variable is com-
posed of two or more gender-matched items respondents viewed after seeing
this introduction: “Here we briefly describe some people. Please read each
description and tick the box on each line that shows how much each person
182 ON VOTER COMPETENCE
is or is not like you.” Six response options were offered: “Very much like me,”
“like me,” “somewhat like me,” “a little like me,” “not like me,” and “not like
me at all.” Item wording is as follows. For the benevolence value: “It is impor-
tant to her to be loyal to her friends. She wants to devote herself to people
close to her” and “It is important to her to respond to the needs of others. She
tries to support those she knows.” The universalism items are as follows: “She
thinks it is important that every person in the world should be treated equally.
She believes everyone should have equal opportunities in life” and “She wants
everyone to be treated justly, even people she doesn’t know. It is important to
her to protect the weak in society.” My power measure is constructed from the
following: “It is important to her to get respect from others. She wants people
to do what she says” and “It is important to her to be in charge and tell oth-
ers what to do. She likes to be the leader.” Next, I gauge security using this
pair of statements: “It is important to her to live in secure surroundings. She
avoids anything that might endanger her safety” and “Having a stable soci-
ety is important to her. She is concerned that the social order be protected.”
Tradition is tapped using the following: “It is important to her to be humble
and modest. She tries not to draw attention to herself” and “Tradition is impor-
tant to her. She tries to follow the customs handed down by her religion or her
family.” Conformity is measured this way: “She believes that people should
do what they’re told. She thinks people should follow the rules at all times,
even when no one is watching” and “It is important to her to be obedient. She
believes she should always show respect to her parents and to older people.”
As before, the conformity and tradition responses are joined together.
These items follow the recommendations made by Schwartz after exten-
sive pretesting (Schwartz, 2003). Analysis conducted elsewhere has shown
that the items tap the designated value dimensions and combine to form rea-
sonably reliable scales (Cronbach α varies from .55 to .75; see Goren, Reifler, &
Scotto, 2011). Recall from Table 7.2 that benevolence and universalism ought
to be negatively associated with military strength, whereas power, security,
and conformity/tradition values are expected to relate positively.
The YouGov survey contains a four-item measure of political knowledge.
The questions ask about the number of terms a president can serve, which
branch of government has responsibility for determining the constitutionality
of a law, the office held by Hillary Clinton, and the primary responsibility of
the United Nations (Cronbach α = .45). The mean for this scale is 3.13 correct
answers out of 4. Compared to past research on political knowledge (e.g., Delli
Carpini & Keeter, 1996) this estimate seems high. As I indicated in note 8,
the fact that respondents opted into the online panel may render them differ-
ent from the general public on political knowledge and similar unmeasured
The Origins of Policy Principles 183
+
p <.10; *p < .05 (one-tailed).
Notes: OLS regression estimates. Unstandardized coefficients are reported. HC3 standard errors are
in parentheses. All variables are on a 0–1 range. Higher values on military strength reflect a hawkish
opinion. Number of observations = 1019. YouGov sampling weights are used.
184
The Origins of Policy Principles 185
Universalism
0.75
0.50 Power
0.25 Conformity/tradition
0.00
Low High
Importance of Value
influence over core principles than party or vice versa? To address this, I test
whether the unstandardized regression coefficient for a given value is signif-
icantly larger than that of party identification. To the degree that values out-
pace party, the evidence will suggest central policy attitudes in mass belief
systems reflect far more than the residue of partisan bias.
Limited government comes first.10 My attention centers on the value dimen-
sions found to be most significant in the regression models. In Table 7.3, the
magnitude of the universalism coefficient (b = –0.38) exceeds that of the party
variable (b = 0.14) at p < .001. And although the conformity coefficient is some
40% larger than that of party (0.20 > 0.14), the null hypothesis of equal weights
cannot be rejected (p = .22). For moral traditionalism (see Table 7.4), the party
coefficient (0.11) lags universalism (–0.27), self-direction (–0.15), and conform-
ity/tradition (0.54) in absolute magnitude. The universalism and conformity
coefficients prove significantly larger than that of party (both p < .002) whereas
the self-direction and party coefficients are statistically indistinguishable. The
coefficients from the military strength equation are as follows: party = 0.19,
universalism = –0.30, conformity/tradition = 0.19, and power = 0.11. Formally,
the party effect trails that of universalism (p < .05), equals the conformity/tra-
dition effect (p = ns), and eclipses power (p < .05).11
10
Party identification is positively related to each policy principle. Some of the
personal values are positively related to a policy variable, but others carry a
negative sign. When party and value are positively related to the dependent
variable, I test the null hypothesis: bPID – bValue = 0. When party and value have
opposing signs, I test this null hypothesis: bPID + bValue = 0.
11
For limited government the standardized beta equals –.29 for universalism, .24
for party identification, and .16 for conformity. For moral traditionalism the
186 ON VOTER COMPETENCE
shape all kinds of preferences and behaviors outside of politics, it seems plau-
sible that they would exercise similar influence on those occasions when atten-
tion turns to politics. Personal values shape what we do for a living, where we
live, who we spend time with, what we spend our money on, and so on. Such
leverage could extend to the political parties with which we identify (Rokeach,
1973). To make the same point another way, values have greater motivational
efficacy than partisanship. We can easily imagine people who go into social
work, organize local food drives, and volunteer willingly and repeatedly to
watch their neighbors’ children because they are motivated by benevolence
and universalism values, by a keen desire to help those near and far. The pos-
sibility that they go into social work or lead food drives because they are a
Democrat seems farfetched. Put simply, the role values play in human judg-
ment renders them more likely to shape partisan loyalties than the reverse.
Other considerations reinforce this argument. A great deal of evidence con-
firms that everyone holds personal values, but the same does not hold true
for partisan attachments. Two bodies of work speak to this point. As reviewed
above, researchers have found that the posited value structure holds in hun-
dreds of samples across scores of countries. The structure has been discovered
in student samples, adult convenience samples, and representative national
samples on six continents over the past three decades. Discrepancies emerge
at times, but for the most part this value hierarchy seems to be a common fea-
ture of the human psyche (much like the five factor model of personality). This
holds true in countries such as the United States that have long-established
party systems, but the value structure also turns up in places where party
organizations and ties are weak or nonexistent. Because values take hold in the
minds of everyone and partisan identities do not, the former can be viewed as
more widely available in, and central elements of, belief systems. The second
body of research that speaks to the temporal priority of values deals with the
distribution of partisanship in the U.S. electorate. Studies always find a large
fraction of the public feels little or no sense of affinity with either party. In the
2008 NES, for instance, 29% of the sample identified weakly with the parties
and another 11% professed complete independence. In terms of raw numbers,
this means that approximately 92,000,000 voting age adults lacked strong
party ties.13 No one supposes that 92,000,000 Americans lack basic values.
To sum up, personal values impinge on a much wider range of judgments
and behaviors than partisan identities. The reach of values extends to almost
13
To calculate this estimate I multiplied a voting age population figure of
230,782,870 (see http://elections.gmu.edu/Turnout_2008G.html) by .40.
188 ON VOTER COMPETENCE
all aspects of life, but partisanship functions only in the political domain.
Moreover, values exist on a nearly universal basis, but partisan identities do
not. The implications are clear: personal values likely hold temporal priority
over party ties and thus are more likely to shape such ties than to be shaped
by them. Ideally, I would test this proposition using panel data containing
repeated measures of Schwartz values and party identification that would
allow me to estimate dynamic constraint models or experimental work in
which the consequences of value and partisan identity manipulations might
be explored. No such data are currently available; therefore, my assumption
that values take precedence over party identification is necessarily tentative.
It is not unreasonable, perhaps, but is tentative all the same. Readers should
keep this caution in mind as we proceed.
I now take a look at the relationship between personal values and parti-
san loyalties using data from the three surveys employed earlier. The value–
party expectations are straightforward because the groups that comprise the
party coalitions and the policies pursued by each party have clear implica-
tions for values. The Democratic Party has long represented disadvantaged
and marginalized groups in American society and sought to redress these
imbalances through political reform. The GOP has represented groups whose
positions in the economic, social, and political spheres are established more
securely; therefore, the party has frequently defended status quo arrange-
ments favored by their constituents (Gerring, 1998). Given these profiles, the
party labels should attract support from people with the following value hier-
archies. Those dedicated to self-transcendence values (i.e., benevolence and
universalism) should gravitate toward the Democratic Party more readily
than those who display little affection for these values. Moreover, those who
pledge allegiance to conservation values (i.e., security and conformity/tradi-
tion) should find the GOP brand more appealing compared to those who give
conservation goals less priority. I have no clear predictions regarding self-
direction, achievement, and power values. Both parties favor self-direction
and achievement, though they differ on how to realize these goals. Finally,
neither party is crass enough to come out and say it stands for controlling or
dominating others.
Table 7.6 arrays the regression estimates from the 2007, 2008, and 2011 surveys
in the second through fourth columns, respectively. Demographic controls are
included for sex, African-American, Hispanic, income, and education, all coded
as above. The dependent variable is coded so that higher scores denote GOP lean-
ings; hence the regression coefficients for universalism and benevolence should
be negative and those for security and conformity/tradition should be positive.
TA B L E 7 . 6 . THE ORIGINS OF PARTY IDENTIFICATION, 2007–2011
2007 2008 2011
+
p < .10; *p < .05 (one-tailed).
Notes: OLS regression estimates. Unstandardized coefficients are reported. HC3 standard
errors are in parentheses. All variables are on a 0–1 range. Higher values on party
identification reflect Republican attachments.
Sources: 2007 Knowledge Networks survey, 2008 Knowledge Networks survey, 2011
YouGov survey.
189
190 ON VOTER COMPETENCE
The estimates show the Democratic brand usually garners more support from
women than from men, from blacks and Hispanics compared to whites, and
from the less educated (marginally) and poor relative to their more educated
and affluent counterparts. None of this is surprising.
More importantly for my purposes, basic human values covary with party
identification as predicted (all joint F tests are significant at p < .001). A univer-
salism effect emerges in all three models; its coefficient is highly significant (p <
.001) and substantively powerful. To illustrate, Figure 7.5 plots predicted parti-
san self-placements across the range of universalism for each year (remember
these illustrate hypothetical maximum effects). The downwardly sloping lines
reveal that the more emphasis respondents place on the well being of others
the weaker their ties to the GOP. Across the three surveys, those who attach
the most importance to universalism score about 44–56% more Democratic
than those who attach the least importance to this value, holding demograph-
ics and other values constant. Conformity/tradition values also behave as
predicted. Each coefficient lies in the posited direction and proves to be highly
significant (p < .001). Figure 7.6 plots partisanship as a function of these goals.
Movement from the minimum to maximum score on conformity/tradition
leads to increases of 35–49% in the direction of GOP identification, all else
being constant. Overall, the regression estimates and plots suggest that uni-
versalism and conformity/tradition values powerfully shape party ties. Three
points warrant emphasis. The first is that the effect sizes are always large; the
second is that consistent effects are obtained even though different question
formats and items are used to measure values; and the third is that the results
are robust over time.
1.00
0.83
Party ID (High = Rep)
0.66
0.50
0.33
0.16
2007 2008 2011
0.00
Low High
Importance of Value
1.00
0.83
Party ID (High = Rep)
0.66
0.50
0.33
0.16
2007 2008 2011
0.00
Low High
Importance of Value
Among the remaining values only two, achievement in 2007 and security in
2011, impact party. The 0.16 achievement coefficient (p < .05) implies that the
more highly we rank personal success, the more likely we are to identify with
the GOP. Security may be negatively related to Republican partisanship in
2011, but the effect is substantively small and marginally significant (b = –0.08,
p < .10). Neither power in 2007 or 2011 nor self-direction in 2007 or 2008 nor
benevolence in 2011 matter. With the exception of benevolence, and security,
I had no firm priors about how these beliefs would affect party. Finally, I find
no evidence that the values–party link is stronger among more educated or
informed respondents (data not shown).
My bottom line conclusion is that universalism and conformity/tradition
hold considerable sway over partisan attachments. So far as my assumption
about the direction of causality is on the mark, the results imply that these two
value domains powerfully shape the positions people take on limited govern-
ment, traditional morality, and military strength. Basic human values exert
strong direct effects on general policy principles that match or, more typically,
surpass those of partisanship. Additionally, values can shape principles indi-
rectly through their influence on party identification, the effects of which also
prove quite substantial (again, experimental and panel studies are needed to
properly evaluate this hypothesis).
CONCLUSIONS
This chapter has examined two factors that correlate strongly with limited
government, traditional morality, and military strength. My evidence sug-
gests that personal values wield a good deal of influence over the positions
192 ON VOTER COMPETENCE
To this point in the book, my results establish that policy principles are
available in the minds of citizens, function as central decision-making heuris-
tics, and are infused with normative beliefs about what is important in life. All
of this seems to hold to a similar degree for politically sophisticated and unso-
phisticated individuals. My final task requires examining systematically how
policy principles shape voter choice in U.S. presidential elections. Chapter 8
takes up that charge.
This page intentionally left blank
CHAPTER 8
The key question posed by my book is whether policy principles shape the
presidential vote to a comparable degree for politically sophisticated and
unsophisticated individuals. I have argued that policy principles will guide
voter choice if three conditions are satisfied. The first is that each principle
must be present in their minds. Insofar as principles are available, they
have the potential to impact candidate choice. The measurement model
and attitude stability estimates reported in Chapter 5 establish that citizens
meet the availability condition. Second, people must use these principles to
guide their reactions to short-term developments such as issue controver-
sies. To the extent that they do so, policy principles are well placed to guide
electoral choice. The dynamic constraint and regression results summarized
in Chapter 6 suggest that individuals meet the centrality condition. Third,
voters must determine which candidate lies closer to them on a given con-
tinuum and then select that candidate. If people fulfill this position match-
ing condition, the case for the third face of policy voting will be complete.
In this chapter, I model the relationship between policy principles and voter
choice using data from the 1988–2008 National Election Study (NES) cross-sec-
tional surveys. My efforts begin with an overview of the leading theories of
electoral behavior: the partisan voter model, the retrospective voter model, and
the ideological voter model. These theories suggest a set of variables that must
be included in the candidate preference equations. Moreover, each perspective
has implications regarding voter competence. After describing the data and
measures, I estimate the statistical models and summarize the results. To fore-
shadow my findings: (1) policy principles consistently and powerfully affect
candidate preferences for all voters; (2) liberal–conservative attitudes affect
the vote only among the politically sophisticated; and (3) the effects of issues
prove inconsistent, though there is a tendency for sophistication to enhance
195
196 ON VOTER COMPETENCE
issue voting. In short, the first and second faces of policy voting elude most
people, but the third face encompasses all electors.
1
But not always. Some partisans cast a more critical eye toward their party when
conditions deteriorate under its stewardship or evaluate the opposition more
favorably when times are good (Lavine et al., 2012).
The Electoral Consequences of Policy Principles 197
Is there more or less unemployment in the country than there was four
years ago? Is America as respected throughout the world as it was? Do
you feel that our security is as safe, that we’re as strong as we were four
years ago?
et al. (2008) discovered that about 11% of the public actively used liberal–
conservative frames of reference to evaluate the presidential candidates and
major parties in 2000. Likewise, Bennett’s (1995) careful analysis of 1980–1992
NES data showed that when asked to provide up to three definitions of what
liberal and conservative mean (for a total of six), nearly half the public could
do no better than a single answer. At the high end, only 12% of the samples
furnished five or six passable definitions. These results, which differ margin-
ally from reports in seminal works, suggest that ideological labels are opaque
if not impenetrable to much of the electorate.
But this does not apply to everyone. Those who know a good deal about
public affairs associate some basic political ideas and symbols with ideo-
logical labels and thus form liberal–conservative attitudes (Sears, 2001;
Sniderman et al., 1991). For the politically aware, these labels may connote
a social group or two such as “feminists” or “Christian fundamentalists.”
For others, the labels connote a symbol such as “big government” or “free
enterprise.” Although these associations are understandable, their lack of
breadth and depth cannot be ignored. Among those who hold genuine lib-
eral–conservative attitudes, such evaluations are rarely based on overarch-
ing political philosophies or summaries of multiple issue positions. Instead,
these attitudes are simple reactions to a limited number of attitude objects.
By my reading of the evidence, perhaps 10% of the electorate acquires
a decent understanding of ideological labels. For another 25–30% or so, atti-
tudes toward liberal–conservative labels reflect little more than impressions of
a couple of groups or symbols. And for the rest of the public, there is little or
nothing. Converse is right on this score.
When it comes to evaluating presidential candidates, those who know
something about ideological labels can satisfy the position-matching condition
of policy voting. The process can be illustrated using a pair of examples from
the 1992 campaign. At the GOP convention that summer, Patrick Buchanan
described the gathering of Democratic delegates several weeks earlier as “that
giant masquerade ball . . . where 20,000 liberals and radicals came dressed up
as moderates and centrists in the greatest single exhibition of cross-dressing
in American political history.”2 Three nights later President Bush charged that
“the liberal McGovern wing of the other party, including my opponent, con-
sistently made the wrong choices.”
To those versed in the argot of ideology, references to “radicals and liber-
als . . . dressed up as moderates and centrists” and “the liberal McGovern
2
http://www.americanrhetoric.com/speeches/patrickbuchanan1992rnc.htm.
200 ON VOTER COMPETENCE
wing of the other party” had a familiar ring. Some interpreted Buchanan’s
diatribe as an accusation of ideological hypocrisy. To other viewers, Bush’s
allusion to liberal McGovernites evoked stereotypes of Democratic politi-
cians as naive peaceniks who could not be trusted to defend the nation.
Whatever meaning they ascribed to these labels, the politically aware could
see which contender was closer to them on the continuum and vote for the
better match. Nothing comparable would have occurred in the minds of
politically unsophisticated viewers. Lacking prior knowledge about what
the labels meant, they would have experienced difficulty extracting useful
information from these cues. Apart perhaps from a vague sense of unease,
allusions to “radicals and liberals . . . dressed up as moderates and centrists”
and the “liberal McGovern wing” of the Democratic Party would have
left many viewers baffled. In other words, the first face of policy voting is
a function of political sophistication.
The normative implications of the ideological voter model can be sum-
marized as follows. On the one hand, some researchers take comfort in
findings that liberal or conservative orientations guide voter choice for
a portion of the electorate (Delli Carpini & Keeter, 1996; Lau & Redlawsk,
2006; Neuman, 1986). This optimistic reading is based on the assumption
that liberal–conservative orientations are infused with a healthy dose of
policy content. In this view, the politically sophisticated use their votes to
communicate about policies they want the federal government to pursue.
But as pointed out repeatedly, for most people liberal and conservative dis-
positions function neither as abstract worldviews nor as issue preference
markers. Because liberal–conservative attitudes reflect simplistic global
judgments about a few social groups or evocative symbols, the fact that the
politically knowledgeable use them to guide their votes is not necessarily
encouraging from a normative point of view.
may surmise its nominee will embrace traditional morality with greater fervor
than the Democratic opposition.
Most voters probably learn candidate positions from the campaign or
through party-based inference. The intention to vote motivates them to do
just this. But even if information from these sources escapes some people,
communication flows through social networks help them fill in the blanks.
People who pay little attention to hard news media and have impoverished
party schemas may learn about candidate positions from a soft news pro-
gram such as Today, from a passing conversation at the lunch counter, from
a chat with neighbors or friends at the local diner, from colleagues at work,
in houses of worship, and so on (Huckfeldt & Sprague, 1995; Popkin, 1994).
Through social channels such as these, cues about candidate positions on gov-
ernment aid, moral standards, and force versus diplomacy can reach voters.
Fiorina (1990, p. 338) sums up this view quite nicely: “Citizens often receive
information in the course of doing other things. There is no question of infor-
mation costs, no question of deciding to gather information as opposed to
doing something else.”
To illustrate the possibilities of campaign learning, I return to the GOP con-
vention speeches invoked above. To begin with Buchanan, although his refer-
ence to cross-dressing radicals and liberals may have mystified millions of
viewers, there was no ambiguity later on in his speech.
Friends, this election is about more than who gets what. It is about who we
are. It is about what we believe and what we stand for as Americans. There
is a religious war going on in this country. It is a cultural war, as critical to
the kind of nation we shall be as the Cold War itself. For this war is for the
soul of America. And in that struggle for the soul of America, Clinton and
Clinton are on the other side, and George Bush is on our side.
Those who watched it live, or saw a snippet rebroadcast on the news at a later
date, or heard about it at work or at church or softball practice, could quickly
realize that President Bush stood to the right of Bill Clinton on traditional
morality. Knowing this, citizens could assess, albeit roughly, where they stood
relative to each contender, thereby increasing the odds of voting for the appro-
priate nominee.
Three nights later, President Bush chastised the Democratic Party for its
alleged spinelessness during the Cold War:
peace through strength. From Angola to Central America, they said, ‘Let’s
negotiate, deliberate, procrastinate.’ We said, ‘Just stand up for freedom.’
A bit later in the speech Bush returned to this theme by declaring that “We
must be a military superpower.” Those who knew nothing about Cold War
proxy battles in Angola and Central America could not mistake what the
president meant by “they wanted a hollow army” and “a nuclear freeze” and
“military superpower.” When encountering this information, whether dur-
ing the speech or in subsequent debates or campaign commercials or con-
versations with friends and co-workers, the implication could not be missed.
Republicans such as Bush were tougher on national security than Democrats
such as Clinton.
I offer a final comment on the normative appeal of the principled voter
model. Policy principles summarize what people want the federal govern-
ment to do in key issue areas and thereby provide a means for substantive
policy representation. Furthermore, when presidents and their allies pur-
sue actions at odds with the policy desires of voters, the latter can support
the challenger when the time comes. Hence, core principles also provide
citizens with a mechanism for holding elected officials accountable for
what they do in office. And because these policy attitudes are grounded
deeply in personal beliefs about what is important in life (see Chapter 7),
they cannot be dismissed as the byproduct of partisan biases or out-group
antipathies. So far as citizens ground candidate preferences in limited gov-
ernment, moral traditionalism, and military strength, the charge that their
votes are bereft of policy concerns, along with the corollary that they are
politically incompetent, will necessitate revision.
Expectations
The study of voting behavior has been animated by three theories. The par-
tisan voter model emphasizes the impact enduring partisan loyalties have
on electoral choice. The retrospective voter model stresses the role played
by judgments about national and international conditions. The ideological
voter model contends that liberal–conservative attitudes shape the choices of
the politically informed. Without denying the utility of these theories, I have
developed an alternative model that centers on basic policy principles.3 The
key theoretical supposition of the principled voter model is that attitudes
toward limited government, traditional moral standards, and military power
3
I take up issue voting later in the chapter.
204 ON VOTER COMPETENCE
shape candidate choice to a comparable degree for citizens across the sophis-
tication spectrum. The model’s key normative claim is that reliance on policy
principles speaks well of the political competence of regular people.
From the above frameworks, I derive the following working hypotheses.
First, the more strongly citizens identity with a political party the higher the
likelihood they will vote for their side’s nominee. Second, positive retrospec-
tive judgments will enhance the probability of a pro-incumbent vote. Third, the
more positively citizens react to an ideological label the greater their propen-
sity to support the ideologically congruent candidate, conditional on political
sophistication. Fourth, stronger left-wing positions on each policy principle
should lead to firmer support for the Democratic candidate.
4
The latter item is not available in 1996 and not used in 2004 due to a program-
ming error in which the item was not administered to 125 respondents. Rather
than losing these cases, which comprise nearly 12% of the sample, I drop the
unemployment item from the analysis.
The Electoral Consequences of Policy Principles 205
respondents who said “weaker” to serve as the reference group. Those who
believe the U.S. position has held steady or improved should favor the incum-
bent, leading me to expect positive coefficients in 1996 and 2000 and negative
coefficients in the remaining years.
I measure liberal–conservative attitudes with the self-placement item
that has respondents describe their “political views” using a seven-point
scale ranging from “extremely liberal” to “extremely conservative.” In each
survey about a quarter to a third of the sample says they “haven’t thought
much about this” or “don’t know.” These respondents receive a follow-up
question asking “If you had to choose, would you consider yourself a lib-
eral or a conservative?” Respondents who selected the “moderate, middle
of the road” option on the seven-point scale get the same probe. Given
the problems with no opinion filters (see Chapter 2), I use the probe to
locate respondents on the seven-point scale. This procedure helps ensure
that the estimation samples are not upwardly biased on the knowledge
variable since those who fail to answer the initial question are typically less
informed than those who comply. The liberal–conservative variable lies on
a 0–1 scale and is keyed so that higher values denote more conservative
leanings, which should depress Democratic voting. To test whether polit-
ical sophistication moderates this relationship, I take the product of the
liberal–conservative and sophistication variables (tapped using the multi-
item political knowledge scales employed in earlier chapters). The coeffi-
cient for the multiplicative term should be negative, which would indicate
that the ideological effect becomes increasingly pronounced at higher lev-
els of sophistication.
Policy principles are measured the usual way. To remind readers, limited
government is a multiple indicator scale composed of four to six federal spend-
ing items depending on the survey (the Cronbach α coefficient varies from .65
to .76). The scale is coded such that higher scores reflect more support for
limited government, so it should be negatively related to voting Democratic.
For moral traditionalism I use the standard quartet of items (the α coefficient
varies from .55 to .67). Because higher scores reflect increasingly conventional
outlooks, this variable should also be negatively related to a Democratic vote.
Lastly, I measure hawk–dove orientations using one to three items depending
on the survey (the α coefficient varies from .51 to .68). Unfortunately, mea-
sures of abstract ideas about military force and diplomacy do not appear in
the 2000 and 2008 surveys. Higher scores denote more hawkish sentiments.
As such, this variable should be negatively related to a Democratic choice. All
principle variables span a 0–1 range. Finally, note that person mean scores are
employed for these scales.
206 ON VOTER COMPETENCE
Notes: Logit estimates. Standard errors are in parentheses (adjusted for complex sample design).
All variables are on a 0–1 range. The partial Wald statistic tests the null hypothesis that the logit
coefficients for the additional variables simultaneously equal 0. Number of observations = 1052.
p < .001). For the individual variables each coefficient is correctly signed and
statistically significant. The negative coefficient for limited government means
that the more someone opposes federal aid for the economically dispossessed,
the lower the likelihood of a Dukakis vote. The negative coefficients for tra-
ditional morality and military strength imply that right-wing views on each
208 ON VOTER COMPETENCE
dimension lower the odds of a Dukakis ballot. These effects hold controlling
for party identification, retrospective judgments, and liberal–conservative
orientations. In short, the model 2 results affirm the proposition that citizens
ground presidential selections in general policy orientations.
How large are the effects? To address this, I use the logit coefficients to sim-
ulate the probability of a Dukakis vote for hypothetical respondents with the
minimum and maximum scores on each principle, holding the other variables
constant at their central tendencies (means for interval level variables and modes
for nominal variables) and plot the results in Figure 8.1. Moving left to right
along the horizontal axis reflects increasingly right leaning positions on each
principle. The vertical axis plots the probability of backing Dukakis. For limited
government, the probability welfare state opponents cast a Dukakis vote is .08
versus .61 for welfare state supporters. For the traditionalism variable, the prob-
ability of Dukakis support equals .57 for those comfortable with cultural change
and .28 for those who reject it. Finally, the probability that national security
doves cast a Democratic vote equals .57 compared to .26 for hawks. These simu-
lations document the power of domain-specific principles to change votes.
The final question is whether the model 2 specification obscures condi-
tionality in the policy principle–vote choice relationship. My argument holds
that principles drive voter choice for everyone. The sophistication interac-
tion model maintains that policy voting is a function of political awareness.
To adjudicate between these perspectives, I estimate a final equation that
assesses how the limited government × sophistication, traditional morality
× sophistication, and military strength × sophistication variables perform
when added to model 2. If my expectations are on the mark, the parameter
1.00
Probability of Dukakis Vote
0.75
0.50
0.25
0.00
Oppose Favor
+
p < .10; *p < .05 (one-tailed).
Notes: Multinomial logit estimates. The Bush vote is the base outcome. Standard errors are in parentheses (adjusted for complex sample design). All variables
are on a 0–1 range. The partial Wald statistic tests the null hypothesis that the logit coefficients for the additional variables simultaneously equal 0. Number of
observations = 1522.
associated with a higher likelihood of voting for Bush relative to Clinton (p <
.001) and Perot (p < .05). Liberal–conservative attitudes do little to guide the
preferences of political novices.
I turn next to policy principles. To begin, the possibility that the principles
jointly have zero effect can be rejected without hesitation (p < .001). Next, five
of the six individual coefficients are statistically significant across the candi-
date pairings, suggesting that voters differentiate between the three candi-
dates on policy grounds. To elaborate, those who favor strong government
are more likely to support Clinton over Bush compared to those who dislike
federal activism (p < .001). The odds of choosing Clinton over Perot are also
higher among backers of the welfare state (p < .001). However, these attitudes
are probably unrelated to the Perot–Bush choice (p ns). Put another way, lim-
ited government systematically affects the Clinton–Bush and Clinton–Perot
pairings but not Bush–Perot. Consider what these results tell us about voter
competence. The GOP and president Bush had a well-established record of
opposing government activism. For Perot, controlling federal spending was
a centerpiece of his campaign. This, of course, sounded like Republican boil-
erplate. In contrast, the Democratic Party had long favored federal efforts to
alleviate economic suffering. Although Clinton took moderate positions on
some issues, he promised to help the “forgotten middle class” and “work-
ing families” through new federal initiatives such as health care cost reform,
expanded health care access, family leave, and so on. In other words, Bush
and Perot were much closer to one another on the limited government con-
tinuum than to Clinton. Hence, the finding that limited government shapes
preferences for Clinton over Bush and Clinton versus Perot but fails to do so
for the Bush–Perot contrast underscores voters’ powers of discernment.
Turning to moral traditionalism, we see analogous results. For the Clinton–
Bush contrast, the negative coefficient means that those who view traditional
codes of conduct positively are less inclined to vote for Clinton than moral
progressives (p < .001). Likewise, cultural traditionalists are less likely to favor
Perot than Bush (p < .001). For Clinton versus Perot, moral attitudes exert little
pull either way (p ns). Given the issue handling reputation of the GOP and
the fact that Bush took more socially conservative positions than Clinton and
Perot, the finding that moral attitudes shape Clinton–Bush and Perot–Bush
voting but fail to emerge for the Clinton–Perot match-up suggests information
about core principles captures voter attention. Like the findings about govern-
ment power, then, these results reveal a thoughtful public, one sensitive to
differences between multiple candidates on multiple policy dimensions.
How does military strength affect candidate choice? Here, the odds of pre-
ferring Bush to either Clinton or Perot are significantly higher among hawks
The Electoral Consequences of Policy Principles 213
than doves (both p < .001). In contrast, hawk–dove orientations manifest little
influence over the Clinton–Perot choice (p ns). Given the GOP’s reputation as
the party of military strength, which had been reinforced by the decisive 1991
U.S. victory over Iraq in the first Gulf War, these findings again speak to the
policy voting capacities of everyday people. Insofar as the contrast between
Bush and his opponents was much cleaner than that between Clinton and
Perot, the finding that the hawk–dove continuum mattered only for the Bush
comparisons implies that voters are cognizant of where the candidates and
parties stand on this fundamental cleavage.
The statistical estimates validate expectations, but they fail to convey how
strongly policy principles guide candidate preference. Therefore, I simulate
the probability that voters cast a Clinton ballot across the range of each prin-
ciple, holding the other variables at their means or modes. Figure 8.2 presents
the curves. Respondents who endorse limited government are .40 less likely to
fancy Clinton (.22) than those who favor activist government (.62). For moral
standards, the likelihood of a Clinton vote equals .65 among the most progres-
sive versus .28 for the most traditional. Lastly, the probability that extreme
hawks back Clinton is .37 compared to .53 for extreme doves. These simula-
tions demonstrate that abstract principles carry a lot of weight in the minds of
voters. Indeed, movement from very negative to very positive evaluations on
each one changes how citizens decide.
But perhaps these results are wrong. Perhaps policy voting is moder-
ated by sophistication. As the interaction model estimates reveal (model 3,
Table 8.2), evidence that the politically aware make up their minds differ-
ently is thin. The likelihood that all three interaction terms have no effect
cannot be rejected (p < .50). The individual coefficients for the government
1.00
Probability of Clinton Vote
0.75
0.50
0.25
0.00
Oppose Favor
Limited Govt Traditional Morality Military Strength
5
I also ran a multinomial logit model with Perot. The conclusions in the text do
not change with Perot included. Given the greatly diminished appeal and sali-
ence of his candidacy, this is not surprising.
The Electoral Consequences of Policy Principles 215
Notes: Logit estimates. Standard errors are in parentheses (adjusted for complex sample design).
All variables are on a 0–1 range. The partial Wald statistic tests the null hypothesis that the logit
coefficients for the additional variables simultaneously equal 0. Number of observations = 962.
0.75
0.50
0.25
0.00
Oppose Favor
Limited Government
Notes: Logit estimates. Standard errors are in parentheses (adjusted for complex sample design).
All variables are on a 0–1 range. The partial Wald statistic tests the null hypothesis that the logit
coefficients for the additional variables simultaneously equal 0. Number of observations = 1040.
was better off were, oddly enough, no more likely to prefer Gore than those
who saw a faltering economy. Lastly, liberal–conservative evaluation contin-
ues to emerge conditionally. Sophisticated liberals line up behind the vice
president whereas sophisticated conservatives go for the governor of Texas.
Among the less informed swatches of the public, liberal–conservative atti-
tudes are inconsequential.
The second column contains the policy principle estimates. The 2000 NES
lacks measures of abstract hawk–dove attitudes, so their effects on the vote
218 ON VOTER COMPETENCE
cannot be assessed. For the other two principles, Table 8.4 yields unequivocal
support for my key hypothesis. As indicated by the adjusted Wald statistic
(p < .01), it is extremely unlikely that both variables have no impact. The
individual coefficients tell the same story as both are statistically significant
(p < .05). Welfare state supporters are more likely to prefer Gore than wel-
fare state critics, whereas moral traditionalists support Bush to a much greater
degree than progressives. To assess the magnitude of these effects, I generate
predicted probability curves. Figure 8.4 indicates that for limited government,
movement from the most left-wing to the most right-wing position drops the
probability of a Gore vote from .67 to .32, all else being equal. Otherwise typ-
ical voters switch sides based on their evaluations of government. The moral
traditionalism curve mirrors this pattern. For zealous traditionalists, the esti-
mated probability of supporting Gore equals .42. For uncompromising pro-
gressives, the probability of a Gore vote is .69.
Do the politically sophisticated rely more on policy principles to guide their
ballots? The interaction model estimates in the last column of Table 8.4 fail to cor-
roborate this hypothesis, as indicated by the insignificant joint test and individual
t-tests. Although sophistication promotes reliance on liberal–conservative atti-
tudes, its influence does not extend to the use of core principles.
Taken as a whole, the results imply abstract policy principles guide voter
decision making. Information about candidate positions on general principles
was dispersed throughout the information environment, thereby allowing
very large portions of the electorate to satisfy the position matching condition
1.00
Probability of Gore Vote
0.75
0.50
0.25
0.00
Oppose Favor
Limited Government Traditional Morality
of policy voting. Regardless of how much or little they knew about public
affairs in general, most voters made good use of limited government and
moral traditionalism when choosing between Gore and Bush. Note finally that
the 2000 results bear a closer resemblance to those of 1988 and 1992, in which
all three principles mattered, than to 1996, in which limited government alone
was significant.
Notes: Logit estimates. Standard errors are in parentheses (adjusted for complex sample design).
All variables are on a 0–1 range. The partial Wald statistic tests the null hypothesis that the logit
coefficients for the additional variables simultaneously equal 0. Number of observations = 724.
0.75
0.50
0.25
0.00
Oppose Favor
Limited Govt Traditional Morality Military Strength
Notes: Logit estimates. Standard errors are in parentheses (adjusted for complex sample design).
All variables are on a 0–1 range. The partial Wald statistic tests the null hypothesis that the logit
coefficients for the additional variables simultaneously equal 0. Number of observations = 1419.
suggesting that less sophisticated voters rely more heavily on limited gov-
ernment. This, of course, undercuts the sophistication interaction model. As
seen in Figure 8.6, for voters one standard deviation below the knowledge
mean movement from the minimum to maximum score on limited govern-
ment drops the probability of an Obama vote from .85 to .44, a large sub-
stantive effect. For voters a standard deviation above the mean, movement
across the limited government range does not affect the probability of an
Obama vote (the probability ticks up from .75 to .76).
The Electoral Consequences of Policy Principles 223
0.75
0.50
0.25
1.00
Probability of Obama Vote
0.75
0.50
0.25
Notes: “–” indicates a nonsignificant effect for the policy principle variable in the 1996 model, so
no issue variable was used to assess the principle’s robustness. NA indicates the military strength
variable was not available in the survey, so no issue variable was used to assess the principle’s
robustness. Number of items used to measure each opinion are in parentheses.
1992 BC 1992 RP
1988 versus GB versus GB 1996 2000 2004 2008
Notes: Logit estimates. Standard errors are in parentheses (adjusted for complex sample design). All variables are on a 0–1 range. The issue variables are listed in
Table 8.7. The issue variables are included only when the policy principle was significant in Tables 8.1–8.6. Issues are coded to be negatively related to vote. The
other variables are included in the model, but are not presented in the table to preserve space.
variables had significant main effects in four cases (one at p < .10), nonsignifi-
cant effects in another four, and sophistication interaction effects in six cases
(two at p < .10). This varied pattern of results implies that sophistication pro-
motes issue voting to some extent and, more importantly, that discrete issues
do not have nearly as much influence over candidate choice as abstract policy
principles.
To sum up, limited government, traditional morality, and military strength
have robust effects on the presidential vote for citizens across the sophisti-
cation continuum. The effects emerge in the presence of well-known predic-
tors of the vote such as party identification and retrospective judgments, they
obtain controlling for liberal–conservative orientations, and they hold when
campaign-specific issues are accounted for.6
1.00
Probability of Dukakis Vote
0.75
0.50
0.25
6
In terms of relative influence, the party variable eclipses the isolated policy
principle effects most of the time. Of course, when principles are considered en
masse, their relative impact rises accordingly.
The Electoral Consequences of Policy Principles 229
1.00
Probability of Obama Vote
0.75
0.50
0.25
CONCLUSIONS
This chapter brings my empirical study of the American voter to a close. To
recap, there exist multiple pathways through which information about can-
didate positions on policy principles can reach prospective voters. They may
230 ON VOTER COMPETENCE
breaks sharply with the dominant strains in the electoral behavior, political
psychology, and public opinion fields, which have insisted for far too long that
the ability to develop, maintain, and use policy attitudes depends on political
sophistication. This holds true when policy attitudes are defined in terms of
liberal–conservative orientations and issue preferences, but once the concep-
tual rubric of policy attitudes is broadened to encompass core policy prin-
ciples, the limits of the conventional wisdom become clear. The concluding
chapter takes up these and other broader matters.
This page intentionally left blank
CHAPTER 9
Voters are frequently mistaken for fools. Some critics offer telling anecdotes to
underscore the political stupidity of the masses. To take one example, a survey
found that 22% of American adults could name all five members of Homer
Simpson’s fictional cartoon family, whereas only 1 in 1000 identified all five
first amendment freedoms (Shenkman, 2008). Others have built a case by mar-
shalling an impressive array of evidence showing that most voters are innocent
of ideology and unsure about the issues, the net result of which is that policy
voting eludes them (Campbell et al., 1960; Converse, 1964; Neuman, 1986).
But from findings such as these it does not automatically follow that voters
are incompetent, that they lack genuine positions on the policy cleavages that
define American politics, or that they cannot judge candidates on grounds of
what really matters—the direction the national government should take in the
major policy domains. In this final chapter, I summarize my argument and
findings and situate them in broader literatures, delineate qualifications that
bear on my claims, and discuss the broader implications my results have for
evaluating the American voter.
233
234 ON VOTER COMPETENCE
of scholars believe policy means little to most voters. The dominant sophisti-
cation interaction model maintains that when it comes to choosing between
presidential candidates, only those who know a good deal about public affairs
rely on the liberal–conservative heuristic and issue preferences to guide their
choices. Insofar as the first two types of policy voting escape most people, the
general public falls short on this criterion of political competence. Instead,
the much larger share, oblivious to liberalism and conservatism and political
issues, defaults to simple cues such as party identification, party infused judg-
ments about conditions at home or abroad, and so on.
The sophistication interaction account is at once plausible and backed by
much evidence. Instead of showing once again that the first two faces of pol-
icy voting are limited to politically aware citizens, I have explored the pos-
sibility that a third face of policy voting exists, one centered on the role of
core principles. These principles reflect abstract beliefs about what should be
done in one of the three major issue areas that comprise the American political
agenda. First, in the economic welfare domain, limited government centers
on the extent to which the government in Washington should provide some
measure of economic security to those in need or otherwise vulnerable to the
vicissitudes of the market economy. Second, traditional morality, defined in
terms of the moral standards that should guide the public and private life of
the nation, operates in the cultural and social issues domain. Third, military
strength centers on the role of force versus diplomacy in the conduct of U.S.
foreign policy.
These principles were chosen for theoretical and historical reasons. On the
theoretical side, voters are cognitive misers whose natural inclinations are to
make accurate decisions without expending scarce cognitive resources; hence,
my predilection for one bedrock principle per issue area makes sense. But why
these three as opposed to other viable principles? On the historical side of
the ledger, my review of the record suggests limited government, traditional
morality, and military strength have long served as organizing principles in
their respective domains. Divisions over whether the national government is
doing too much or too little to help average Americans and those at the eco-
nomic margins of society extend back to the New Deal era. The 1960s ushered
in a series of disruptive social changes, what came to be known as the “culture
wars,” battles that continue to this day. Questions about striking the proper
balance between military muscle and the tools of statecraft have animated the
foreign policy and national security domain since the aftermath of World War
II. Although the relative visibility of these disparate conflicts has waxed and
waned over time, these divisions have defined political conflict in this country
for decades. Moreover, they have mapped onto the party system for nearly
236 ON VOTER COMPETENCE
who know relatively little about politics rely on limited government, tra-
ditional morality, and military strength as much as those whose knowledge
stores run far deeper. Core principles manifested strong substantive effects
within and across the elections, controlling for retrospective judgments, party
identification, liberal–conservative orientations, and issue preferences. Lastly,
the estimates revealed that sophistication habitually amplifies ideological
voting and promotes issue voting to a degree as well. At the end of the day,
core principles represent the only class of policy attitudes that consistently
and powerfully shapes voter choice. The first two faces of policy voting may
be confined to more sophisticated strata of the electorate, but the third face
extends largely to all. The American voter is a principled voter.
QUALIFICATIONS
These conclusions rest on a number of simplifying assumptions made through-
out the book. As such, the limits of what I have done and what I can claim
must be emphasized. To begin, the argument rests heavily on my judgment
that the policy space of American politics revolves around three fundamen-
tal cleavages. My coverage of American history from the 1930s to the present
maintains that divisions over the role the federal government should play in
insulating citizens from market risk constitutes the first great policy cleavage;
conflict over social change and the implications it has for moral standards con-
stitutes a second overriding cleavage; and that disputes over whether force,
diplomacy, or some combination thereof best serve U.S. foreign policy and
national security goals expresses a third fundamental cleavage. These strike
me as leading candidates for America’s political fault lines, but they are not
the only ones the analyst might select.
Similar to quantitative work, historical analysis relies on simplification. In
trying to reduce so much of this complexity to the essentials, we court the
risk that something important has been missed. Some readers may be trou-
bled by my neglect of the racial cleavage. But as I explained in Chapter 2,
my argument is not that race is irrelevant. There is no denying the profound
implications race as a social cleavage has had for structuring the party system.
Moreover, explicit racial issues, such as busing and affirmative action, have
moved onto the national agenda on occasion. My claim is simply that after the
1960s the issue of race qua race has diminished in relative importance. Once
politicians embraced the principle of racial equality, whether by choice or
necessity, overt racial issues receded into the background. Conservative poli-
ticians could no longer campaign openly on segregationist or other explicitly
antiblack platforms. This does not mean racial issues disappeared—far from
238 ON VOTER COMPETENCE
it. Instead, they became intertwined with other, ostensibly race-neutral issues
such as law and order and welfare (Mendelberg, 2001). Again, race matters,
but not, in my judgment, to the extent that battles over limited government,
social change, and military power do.
A second potential shortcoming lies in my assessment (or lack thereof) of
position matching. Recall this requires prospective voters to compare their
own positions on each dimension to those espoused by the candidates and
then go for the one lying in closer proximity. Although voters struggle when
it comes to placing presidential candidates on the liberal–conservative scale
and various issue continua, I have argued that the barriers to doing so are
much lower for the limited government, traditional morality, and hawk ver-
sus dove dimensions. Lacking direct measures of voter perceptions of can-
didate positions, I presumed the salience of these affairs, the persistence of
party stereotypes and issue handling reputations on first principles, and the
multiple opportunities for campaign learning help most people see where the
candidates stand on each dimension. This is a reasonable assumption, per-
haps, but an assumption all the same.
Third, I studied the relative influence of different classes of policy atti-
tudes on the vote, conditional on a single moderator—political sophistica-
tion. Other approaches suggest variation along intraattitudinal dimensions of
strength matter too. Prior work chronicles that attitude accessibility, ambiva-
lence, importance, and uncertainty influence political judgment and behavior
(Krosnick, 1990; Lavine et al., 2012). Perhaps some people attach much greater
importance to limited government than others. In this case, attitude impor-
tance could alter the magnitude of the relationship between principles and
candidate preferences. Or perhaps those sure of their positions make greater
use of principles than those bedeviled by some uncertainty. Having noted this,
the attitude stability evidence in Chapter 5 implies policy principles are strong
attitudes for most people, hence they should continue to matter as a general
rule.
A fourth issue concerns the generalizability of these results over time. The
1988–2008 period studied here occurred against a backdrop of intense elite
polarization, thereby raising questions about what happened before and what
is likely to transpire going forward [the paucity of adequate measures of pol-
icy principles on National Election Study (NES) surveys before 1988 dictated
my temporal focus]. What occurred earlier depends on when the cleavages
mapped onto the party system. Thus, the prospects of voting on the moral
dimension seem dim prior to the 1960s when widespread social conflict
emerged and the GOP sought to exploit cultural backlash for electoral gain
later that decade. Similarly, widespread hawk–dove voting seems more likely
The Exoneration of the American Voter? 239
to have occurred after the parties began to separate on this dimension in the
early 1980s than during the era of Cold War consensus. Truman and Kennedy
were Cold War warriors after all. But when attention turns to limited govern-
ment, this cleavage has probably shaped candidate choice since the New Deal
(cf. Campbell et al., 1960). Indeed, in terms of its durability over time and the
role it has played in structuring electoral choice throughout the culture wars
and various cold and hot wars, we might reasonably posit that limited gov-
ernment is the dominant cleavage in recent American history (cf. Claggett &
Shafer, 2010).
What of the future? Should we expect these cleavages to endure going
forward? Because elite polarization shows no sign of abating anytime soon,
party signals will remain clear for some time on these dimensions. The ques-
tion then becomes how salient these conflicts are likely to be. Battles over the
size and scope of government have continued to rage since Obama’s elec-
tion. Obama and congressional Democrats have clashed with the congressio-
nal wing of the GOP over the stimulus bill, health care reform, government
spending, and taxes, all manifestations of the broader conflict about the role
the federal government should play in American economic life. On the moral
dimension, the trajectory seems more difficult to chart. American society has
become more progressive on some matters over the past decade or so (e.g.,
gay rights, gay marriage, and marijuana usage). Cohort replacement, whereby
older, less tolerant citizens are replaced by younger, more tolerant counter-
parts, would seem to ensure that these liberalizing trends continue apace. But
in the short term, it is easy to imagine that traditional values will continue to
influence candidate preferences so long as a sizable share of the public presses
demands for moral restoration. Lastly, the hawk–dove dimension seems most
dependent on prevailing developments in the world. As I found in Chapter 8,
attitudes toward this dimension did not significantly affect the vote in 1996,
an understandable result given the demise of the Cold War a few years ear-
lier and the lack of a shooting war that year. If America becomes preoccupied
with domestic problems and turns inward or international problems revolve
around economic rather than security issues, the hawk–dove cleavage will
diminish in relevance. Of course, as all presidents have learned the hard way,
foreign policy crises that require difficult choices about the application of mil-
itary power often explode without warning. Another terrorist attack on U.S.
soil, a civil war in Iraq that draws in neighboring countries, or conflict with
Iran or North Korea that spirals out of control could quickly elevate the sali-
ence of this dimension.
The next qualification to stress is that sophistication conditions some
elements of political judgment and electoral choice. Throughout the book I
240 ON VOTER COMPETENCE
have made a strong case that sophistication is less important than we have
been led to believe. This should not be read as saying that sophistication is
irrelevant or that it has no bearing whatsoever. To revisit findings attesting
to its significance, the partisan identities and policy principles adopted by
the highly informed are more durable over time than those held by the less
informed (Chapter 5); the sophisticated rely more heavily than the unsophis-
ticated on limited government to some extent, and moral traditionalism to an
even greater extent, to vertically constrain their positions on concrete issues
(Chapter 6); the aware make greater use of liberal–conservative attitudes than
the unaware to construct preferences and guide votes (Chapters 6 and 8); and
sophistication enhances issue voting (Chapter 8). Put simply, sophistication
affects how people decide, but again, its absence does not preclude principled
political thought.
Finally, I must acknowledge my argument’s debt to revisionist work on
public opinion, voting behavior, and citizen competence. I drew heavily on the
domain-specific model of public opinion developed by Peffley and Hurwitz
(1985; Hurwitz & Peffley, 1987), Feldman (1988), Feldman and Zaller (1992),
and Miller and Shanks (1996). In The Reasoning Voter, Popkin (1994) argues
that citizens are adept at heuristic reasoning and learning from campaigns, a
perspective that guides the approach taken in these pages. Aldrich et al. (1989)
make many similar points. Most recently, Ansolabehere et al. (2008) come to
the defense of the American voter by showing that issue attitudes are more
stable over time and policy voting is more prevalent than suggested by the
sophistication interaction model.1
Though similar in spirit, two differences between On Voter Competence
and these efforts deserve emphasis. First, some revisionists remain wedded
to liberalism and conservatism and issue preferences. My view is that much
is gained by elaborating the conceptual and empirical distinction between
the three classes of policy attitudes. Insofar as ideological worldviews are
too abstract for most people to comprehend and there is too much informa-
tion about scores of policy controversies within and across multiple policy
domains to contend with, it is not hard to see why everyday people, beset
as they are by the demands of daily living and innate cognitive constraints,
fail to develop strong attitudes toward ideological labels and political issues.
Policy principles are different. They are neither too abstract to grasp intui-
tively nor too concrete and innumerable to keep up with. Citizens develop
1
Nie et al. (1979) offered an especially comprehensive challenge to the Michigan
model of political behavior, but subsequent work has undermined some of the
key evidence (see Smith, 1989, for a review).
The Exoneration of the American Voter? 241
policy principles without much difficulty and deploy them as the primary
policy lenses through which the political world can be evaluated. Indeed, this
approach represents an optimal use of policy heuristics in the realm of polit-
ical choice.
The second critical difference lies in the broader scope of my book. My work
alone centers on the historical, psychological, and behavioral importance of
core policy principles. And whereas most revisionist studies restrict analysis
to part of the big picture, I investigated the development, crystallization, ori-
gins, and consequences of policy principles for understanding public opinion
and the presidential vote. These differences aside, all revisionist work is ani-
mated by the contrarian spirit of V.O. Key (1966), who declared in the pages of
his final work that “The voter emerges as a person who appraises the actions
of government, who has policy preferences, and who relates his vote to those
appraisals and preferences” (pp. 58–59). Voters, in short, are no fools.
BROADER IMPLICATIONS
Doubts about the political intelligence of ordinary people go back a very long
time. Plato warned against letting the demos take hold of the ship of state and
regarded democracy as superior only to tyranny. Citizens could not be trusted
to rule; instead, those responsibilities must be reserved unto wise and benev-
olent philosopher kings. Next, though he championed representative govern-
ment, John Stuart Mill (1861/1991) was so apprehensive about its prospects
that he proposed a plurality voting system in which “the institutions of the
country should stamp the opinions of persons of a more educated class as
entitled to greater weight than those of the less educated” (p. 241). Fears that
the “passions” of ordinary people would undermine the political order and
destroy individual liberty pepper The Federalists Papers. To take a single exam-
ple from Federalist Paper 63, “the people, stimulated by some irregular pas-
sion, or some illicit advantage, or misled by the artful misrepresentations of
interested men, may call for measures which they themselves will afterwards
be the most ready to lament and condemn” (Hamilton, Madison, & Jay, 1961,
p. 384). Joseph Schumpeter, writing in 1942, scoffed at the notion of demo-
cratic governance: “the typical citizen drops down to a lower level of mental
performance as soon as he enters the political field. He argues and analyzes in
a way which he would readily recognize as infantile within the sphere of his
real interests. He becomes a primitive again” (p. 262).
The emergence of scientific opinion polling in the 1940s and 1950s seemed
to confirm this gloomy view. As I have noted throughout my book, schol-
ars have been complaining about voter shortfalls ever since. Carping about
242 ON VOTER COMPETENCE
Yet sophistication also facilitates less noble behaviors, a point not always
acknowledged by those who sing its praises. To take some examples, the politi-
cally aware are more likely to process new political information in ways that
reinforce their prior beliefs and ignore evidence that contradicts their expecta-
tions (Lodge & Hamill, 1986; Taber & Lodge, 2006); the pernicious influence of
partisan bias on subjective political judgments and objective facts grows stron-
ger at higher levels of sophistication (Shani, 2006); educated and sophisticated
individuals are especially susceptible to thinking about welfare in race-coded
terms and applying a racial double standard to claims for economic assistance
made on behalf of blacks (Federico, 2006; Goren, 2003); the sophisticated are
more likely to ape the positions taken by political leaders and even political
entertainers such as Rush Limbaugh than their less sophisticated counterparts
(Barker, 2002; Zaller, 1992).
Although political sophistication is ritualistically hailed for promoting the
hallmarks of good citizenship, the dark side of sophistication has been less
widely noted (cf. Lavine et al., 2012). To the extent that sophistication promotes
selective perception, motivated reasoning, dogmatism, blind partisan loyalty,
ideological rigidity, and so on, its status as normatively sacrosanct should be
challenged. This is not to say that ignorance is preferable to knowledge, but
rather that sophistication has both positive and negative ramifications for dem-
ocratic life. The case for sophistication must be assessed civic virtue by civic
virtue and normative judgments should be withheld unless backed by careful
theoretical reasoning and rigorous empirical scrutiny. Ultimately, I would not
be surprised if sophistication’s benefits outweigh its liabilities. At the same
time, it would not surprise me if the differences between the sophisticated and
the unsophisticated turn out to be as modest for other dimensions of political
competence as I have shown them to be in the case of policy voting.
To conclude, Delli Carpini and Keeter (1996) may well be right that
“informed citizens are demonstrably better citizens, as judged by the stan-
dards of democratic theory and practice underpinning the American system”
(p. 272). At the end of the day, I have rebutted a single charge in the indictment
against the American voter. When it comes to developing and holding policy
principles, using these to guide the positions they take on short-term political
judgments, and grounding the presidential vote in such postures, the unso-
phisticated do about as well as the sophisticated. Given all that we have been
told about the cognitive, motivational, and normative deficiencies of ordinary
citizens, these are no mean achievements. But other charges have been leveled
against the people, ones that I have not addressed here. The question of which
way the scale ultimately tips remains to be settled.
This page intentionally left blank
Appendix: Measurement of Key Variables
245
246 APPENDIX: MEASUREMENT OF KEY VARIABLES
249
250 REFERENCES
Barker, D. C. (2002). Rushed to judgment: Talk radio, persuasion, and American political
behavior. New York, NY: Columbia University Press.
Bartels, L. M. (2002). Beyond the running tally: Partisan bias in political percep-
tions. Political Behavior, 24, 117–150.
Bartels, L. M. (2008). Unequal democracy: The political economy of the new gilded age.
Princeton, NJ: Princeton University Press.
Baum, M. A. (2002). Sex, lies, and war: How soft news brings foreign policy to the
inattentive public. American Political Science Review, 96, 91–109.
Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in experi-
mental social psychology (Vol. 6, pp. 1–62). New York, NY: Academic Press.
Bennett, S. E. (1995). Americans’ knowledge of ideology, 1980–1992. American
Politics Quarterly, 23, 259–278.
Berelson, B. R., Lazarsfeld, P. F., & McPhee, W N. (1954). Voting: A study of opinion
formation in a presidential campaign. Chicago, IL: University of Chicago Press.
Berinsky, A. J. (2009). In time of war: Understanding American public opinion from
World War II to Iraq. Chicago, IL: University of Chicago Press.
Bernaards, C. A., & Sijtsma, K. (1999). Factor analysis of multidimensional poly-
tomous item response data suffering from ignorable item nonresponse.
Multivariate Behavioral Research, 34, 277–313.
Bishop, G. F., Oldendick, R. W., Tuchfarber, A. J., & Bennett, S. E. (1980). Pseudo-
opinions on public affairs. Public Opinion Quarterly, 44, 198–209.
Blum, J. M. (1991). Years of discord: American politics and society, 1961–1974. New York,
NY: W.W. Norton.
Bollen, K. A. (1989). Structural equations with latent variables. New York, NY: Wiley.
Boyer, P. S. (Ed.). (2001). The Oxford companion to United States history. New York,
NY: Oxford University Press.
Brewer, M. D., & Stonecash, J. M. (2007). Split: Class and cultural divides in American
politics. Washington, DC: CQ Press.
Brewer, M. D., & Stonecash, J. M. (2009). The dynamics of American political parties.
New York, NY: Cambridge University Press.
Brewer, P. R. (2003). Values, political knowledge, and public opinion about gay
rights: A framing-based account. Public Opinion Quarterly, 67, 173–201.
Campbell, A., Converse, P. E., Miller, W. E., & Stokes, D. E. (1960). The American
voter. New York, NY: Wiley.
Campbell, D. T. (1963). Social attitudes and other acquired behavioral dispositions.
In S. Koch (Ed.), Psychology: A study of a science (Vol. 6, pp. 94–172). New York,
NY: McGraw-Hill.
Caplan, B. (2007). The myth of the rational voter: Why democracies choose bad policies.
Princeton, NJ: Princeton University Press.
Caprara, G. V., Schwartz, S., Capanna, C., Vecchione, M., & Barbaranelli, C. (2006).
Personality and politics: Values, traits, and political choice. Political Psychology,
27, 1–28.
Carmines, E. G., & Stimson, J. A. (1980). The two faces of issue voting. American
Political Science Review, 74, 78–91.
References 251
Carmines, E. G., & Stimson, J. A. (1989). Issue evolution: Race and the transformation
of American politics. Princeton, NJ: Princeton University Press.
Carney, D. R., Jost, J. T., Gosling, S. D., & Potter, J. (2008). The secret lives of liberals
and conservatives: Personality profiles, interaction styles, and the things they
leave behind. Political Psychology, 29, 807–840.
Carsey, T. M., & Layman, G. C. (2006). Changing sides or changing minds? Party
identification and policy preferences in the American electorate. American
Journal of Political Science, 50, 464–477.
Carter, D. T. (1995). The politics of rage: George Wallace, the origins of the new con-
servatism, and the transformation of American politics. New York, NY: Simon &
Schuster.
Chaiken, S. (1987). The heuristic model of persuasion. In M. P. Zanna, J. M. Olson,
& C. P. Herman (Eds.), Social influence: The Ontario Symposium (Volume 5,
pp. 3–39). Hillsdale, NJ: Lawrence Erlbaum.
Chong, D., McClosky, H., & Zaller, J. (1983). Patterns of support for democratic
and capitalist values in the United States. British Journal of Political Science, 13,
401–440.
Claggett, W. J. M., & Shafer, B. E. (2010). The American public mind: The issues struc-
ture of mass politics in the postwar United States. New York, NY: Cambridge
University Press.
Collins, R. M. (2000). More: The politics of economic growth in postwar America.
New York, NY: Oxford University Press.
Collins, R. M. (2007). Transforming America: Politics and culture during the Reagan
years. New York, NY: Columbia University Press.
Conover, P. J., & Feldman, S. (1981). The origins and meaning of liberal/conserva-
tive self identifications. American Journal of Political Science, 25, 617–645.
Conover, P. J., & Feldman, S. (1989). Candidate perception in an ambiguous world:
Campaigns, cues, and inference processes. American Journal of Political Science,
33, 912–940.
Conover, P. J., Feldman, S., & Knight, K. (1986). Judging inflation and unemploy-
ment: The origins of retrospective evaluations. Journal of Politics, 48, 565–588.
Converse, P. E. (1964). The nature of belief systems in mass publics. In D. Apter
(Ed.), Ideology and Discontent (pp. 206–261). New York, NY: Free Press.
Converse, P. E. (1970). Attitudes and non-attitudes: Continuation of a dialogue. In
E. R. Tufte (Ed.), The quantitative analysis of social problems (pp. 168–189). Reading,
MA: Addison-Wesley.
Converse, P. E. (1975). Public opinion and voting behavior. In F. I. Greenstein &
N. W. Polsby (Eds.), Handbook of political science, volume 4: Nongovernmental poli-
tics (pp. 75–169). Reading, MA: Addison-Wesley.
Converse, P. E. (1990). Popular representation and the distribution of information.
In J. A. Ferejohn & J. H. Kuklinski (Eds.), Information and democratic processes
(pp. 369–388). Urbana, IL: University of Illinois Press.
Converse, P. E. (2000). Assessing the capacity of mass electorates. Annual Review of
Political Science, 3, 331–353.
252 REFERENCES
Converse, P. E. (2006). Democratic theory and electoral reality. Critical Review, 18,
297–329.
Converse, P. E., & Markus, G. B. (1979). Plus ca change . . . : The new CPS election
study panel. American Political Science Review, 73, 32–49.
Curran, P. J., West. S. G., & Finch, J. F. (1996). The robustness of test statistics to non-
normality and specification error in confirmatory factor analysis. Psychological
Methods, 1, 16–29.
Dallek, R. 2003. An unfinished life: John F. Kennedy, 1917–63. Boston, MA: Little
Brown and Co.
Dancey, L., & Goren, P. (2010). Party identification, issue attitudes, and the dynam-
ics of political debate. American Journal of Political Science, 54, 686–699.
Davidov, E., Schmidt, P., & Schwartz, S. H. (2008). Bringing values back in: The
adequacy of the European Social Survey to measure values in 20 countries.
Public Opinion Quarterly, 72, 420–445.
DeGroot, G. J. (2008). The sixties unplugged: A kaleidoscopic history of a disorderly
decade. Cambridge, MA: Harvard University Press.
Delli Carpini, M. X., & Keeter, S. (1993). Measuring political knowledge: Putting
first things first. American Journal of Political Science, 37, 1179–1206.
Delli Carpini, M. X., & Keeter, S. (1996). What Americans know about politics and why
it matters. New Haven, CT: Yale University Press.
Downs, A. (1957). An economic theory of democracy. New York, NY: Harper & Row.
Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Fort Worth, TX:
Harcourt, Brace, Jovanovich.
Edsall, T. B., & Edsall, M. D. (1991). Chain reaction: The impact of race, rights, and taxes
on American politics. New York, NY: W.W. Norton.
Enelow, J. M., & Hinich, M. J. (1984). The spatial theory of voting: An introduction.
New York, NY: Cambridge University Press.
Fazio, R. H. (1989). On the power and functionality of attitudes: The role of atti-
tude accessibility. In A. R. Pratkanis, S. J. Breckler, & A. G. Greenwald (Eds.),
Attitude structure and function (pp. 153–179). Hillsdale, NJ: Lawrence Erlbaum.
Fazio, R. H. (2007). Attitudes as object-evaluation associations of varying strength.
Social Cognition, 25, 603–637.
Federico, C. M. (2006). Race, education, and individualism revisited. Journal of
Politics, 68, 600–610.
Feldman, S. (1988). Structure and consistency in public opinion: The role of core
beliefs and values. American Journal of Political Science, 32, 416–440.
Feldman, S. (1989). Measuring issue preferences: The problem of response instabil-
ity. In J. A. Stimson (Ed.), Political analysis: Volume 1 (pp. 25–60). Ann Arbor, MI:
University of Michigan Press.
Feldman, S, & Conover, P. J. (1983). Candidates, issues, and voters: The role of
inference in political perception. Journal of politics, 45, 810–839.
Feldman, S., & Zaller, J. (1992). The political culture of ambivalence: Ideological
responses to the welfare state. American Journal of Political Science, 36, 268–307.
References 253
Finney, S. J., & DiStefano, C. (2006). Non-normal and categorical data in structural
equation modeling. In G. R. Hancock & R. O. Mueller (Eds.), Structural equa-
tion modeling: A second course (pp. 269–314). Greenwich, CT: Information Age
Publishing.
Fiorina, M. P. (1981). Retrospective voting in American national elections. New Haven,
CT: Yale University Press.
Fiorina, M. P. (1990). Information and rationality in elections. In J. A. Ferejohn &
J. H. Kuklinski (Eds.), Information and democratic processes (pp. 329–342). Urbana,
IL: University of Illinois Press.
Fiske, S. T., Kinder, D. R., & Larter, W. M. (1983). The novice and the expert:
Knowledge-based strategies in political cognition. Journal of Experimental Social
Psychology, 19, 381–400.
Fiske, S. T., Lau, R. L., & Smith, R. A. (1990). On the variety and utility of political
knowledge structures. Social Cognition, 8, 31–48.
Fiske, S. T., & Taylor, S. E. (1991). Social cognition. New York, NY: McGraw-Hill.
Friedrich, R. J. (1982). In defense of multiplicative terms in multiple regression
equations. American Journal of Political Science, 26, 797–833.
Gaddis, J. L. (2005). The cold war: A new history. New York, NY: Penguin Press.
Gawronski, B. (2007). Editorial: Attitudes can be measured! But what is an atti-
tude? Social Cognition, 25, 573–581.
Gelman, A., & Stern, H. (2006). The difference between “significant” and “not
significant” is not itself statistically significant. American Statistician, 60,
328–331.
Gerber, A. S., Huber, G. D., Doherty, D., Dowling, C. M., & Ha, S. E. (2010).
Personality and political attitudes: Relationships across issue domains and
political contexts. American Political Science Review, 104, 111–133.
Gerring, J. (1998). Party ideologies in America, 1828–1996. New York, NY: Cambridge
University Press.
Gilens, M. (1999). Why Americans hate welfare: Race, media, and the politics of antipov-
erty policy. Chicago, IL: University of Chicago Press.
Goldberg, R. A. (1995). Barry Goldwater. New Haven, CT: Yale University Press.
Goren, P. (1997). Political expertise and issue voting in presidential elections.
Political Research Quarterly, 50, 387–412.
Goren, P. (2001). Core principles and policy reasoning in mass publics. A test of
two theories. British Journal of Political Science, 31, 159–177.
Goren, P. (2002). Character weakness, partisan bias, and presidential evaluation.
American Journal of Political Science, 46, 627–641.
Goren, P. (2003). Race, sophistication, and white opinion on government spending.
Political Behavior, 25, 201–220.
Goren, P. (2004). Political sophistication and policy reasoning: A reconsideration.
American Journal of Political Science, 48, 462–478.
Goren, P. (2005). Party identification and core political values. American Journal of
Political Science, 49, 881–896.
254 REFERENCES
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance struc-
ture analysis: Conventional criteria versus new alternatives. Structural Equation
Modeling, 6, 1–55.
Huckfeldt, R. R., & Sprague, J. D. (1995). Citizens, politics, and social communica-
tion: Information and influence in an election campaign. New York, NY: Cambridge
University Press
Hunter, J. D. (1992). Culture wars: The struggle to define America. New York, NY:
Basic Books.
Hurwitz, J., & Peffley, M. (1987). How are foreign policy attitudes structured? A
hierarchical model. American Political Science Review, 81, 1099–1120.
Jacoby, S. (2009). The age of American unreason. New York, NY: Vintage.
Jacoby, W. G. (1994). Public attitudes toward government spending. American
Journal of Political Science, 38, 336–361.
Jennings, M. K. (1992). Ideological thinking among mass publics and political
elites. Public Opinion Quarterly, 56, 419–441.
John, O. P., Naumann L. P., & Soto, C. J. (2008). Paradigm shift to the integrative
big five trait taxonomy: History, measurement, and conceptual Issues.” In O.
P. John, R. W. Robins, & L. A. Pervin, (Eds.), Handbook of Personality: Theory and
research, 3rd ed. (pp. 114–158). New York, NY: Guilford Press.
Jost, J. T. (2006). The end of the end of ideology. American Psychologist, 61,
651–670.
Jost, J. T., Federico, C. M., & Napier, J. L. (2009). Political ideology: Its structure,
functions, and elective affinities. Annual Review of Psychology, 60, 307–337.
Jost, J. T., Glaser, J., Kruglanski, A. W., & Sulloway, F. J. (2003). Political conserva-
tism as motivated social cognition. Psychological Bulletin, 129, 339–375.
Judd, C. M., Drake, R. A., Downing, J. M., & Krosnick, J. A. (1991). Some dynamic
properties of attitude structures: Context-induced response facilitation and
polarization. Journal of Personality and Social Psychology, 60, 193–202.
Just, M. R., Crigler, A. N., Alger, D. E., Cook, T. E., Kern, M., & West, D. M. (1996).
Crosstalk: Citizens, candidates, and the media in a presidential campaign. Chicago, IL:
University of Chicago Press.
Katz, D. (1960). The functional approach to the study of attitudes. Public Opinion
Quarterly, 24, 163–204.
Kennedy, D. M. (1999). Freedom from fear: The America people in war and depression,
1929–1945. New York, NY: Oxford University Press.
Kerlinger, F. N. (1984). Liberalism and conservatism: The nature and structure of social
attitudes. Hillsdale, NJ: Lawrence Erlbaum.
Key, V. O. (1966). The responsible electorate: Rationality and presidential voting, 1936–60.
Cambridge, MA: Belknap Press.
Kinder, D. R. (1998). Opinion and action in the realm of politics. In A. R. Pratkanis,
S. J. Breckler, & A. G. Greenwald (Eds.), Handbook of social psychology, 4th ed.
(pp. 778–867). Hillsdale, NJ: Lawrence Erlbaum Press.
Kinder, D. R. (2006). Belief systems today. Critical Review, 18, 197–216.
256 REFERENCES
Norpoth, H., & Lodge, M. (1985). The difference between attitudes and nonatti-
tudes in the mass public: Just measurement? American Journal of Political Science,
29, 291–307.
Oberdofer, D. (1998). From the cold war to a new era: The United States and the Soviet
Union, 1983–1991 (Updated ed.). Baltimore, MD: John Hopkins University
Press.
Packer, G. (2005). The assassin’s gate: America in Iraq. New York, NY: Farrar, Straus,
and Giroux.
Page, B. I. (1978). Choices and echoes in presidential elections: Rational man and electoral
democracy. Chicago, IL: University of Chicago Press.
Page, B. I., & Bouton, M. M. (2006). The foreign policy disconnect: What Americans
want from our leaders but don’t get. Chicago, IL: University of Chicago Press.
Page, B. I., & Shapiro, R. Y. (1992). The rational public: Fifty years of trends in Americans’
policy preferences. Chicago, IL: University of Chicago Press.
Patterson, J. T. (1994). America’s struggle against poverty, 1900–1994. Cambridge,
MA: Harvard University Press.
Patterson, J. T. (1996). Grand expectations: The United States, 1945–74. New York, NY:
Oxford University Press.
Patterson, J. T. (2005). Restless giant: The United States from Watergate to Bush v. Gore.
New York, NY: Oxford University Press.
Peffley, M., & Hurwitz, J. (1985). A hierarchical model of attitude constraint.
American Journal of Political Science, 29, 871–890.
Peffley, M., & Hurwitz, J. (1993). Models of attitude constraint in foreign affairs.
Political Behavior, 15, 61–90.
Pennock, J. R. (1979). Democratic political theory. Princeton, NJ: Princeton University
Press.
Perlstein, R. (2008). Nixonland: The rise of a president and the fracturing of America.
New York, NY: Scribner.
Petrocik, J. R. (1996). Issue ownership in presidential elections, with a 1980 case
study. American Journal of Political Science, 40, 825–850.
Petty, R. E., & Krosnick, J. A. (Eds.) (1995). Attitude strength: Antecedents and conse-
quences. Hillsdale, NJ: Lawrence Erlbaum.
Pierce, J. C., & Hagner, P. R. (1982). Conceptualization and party identification,
1956–1976. American Journal of Political Science, 26, 377–387.
Pollock, P. H., Lilie, S. A., & Vittes, E. (1993). Hard issues, core values, and vertical
constraint: The case of nuclear power. British Journal of Political Science, 23, 29–50.
Pomper, G. M. (1972). From confusion to clarity: Issues and American voters, 1956–
1968. American Political Science Review, 66, 415–428.
Popkin, S. L. (1994). The reasoning voter: Communication and persuasion in presidential
campaigns (2nd. ed.). Chicago, IL: University of Chicago Press.
Price, V., & Zaller, J. (1993). Who gets the news? Alternative measures of news
reception and their implications for research. Public Opinion Quarterly, 57,
133–164.
References 259
Rahn, W. M., Aldrich, J. H., Borgida, E., and Sullivan, J. L. (1990). A social cogni-
tive model of candidate appraisal. In J. A. Ferejohn & J. H. Kuklinski (Eds.),
Information and democratic processes (pp. 136–159). Urbana, IL: University of
Illinois Press.
Ricks, T. E. (2006). Fiasco: The American military adventure in Iraq. New York, NY:
Penguin Press.
Roccas, S., Sagiv, L., Schwartz, S. H., & Knafo, A. (2002). The big five personal-
ity factors and personal values. Personality and Social Psychology Bulletin, 28,
789–801.
Rokeach, M. (1973). The nature of human values. New York, NY: Free Press.
Scammon, R. M., & Wattenberg, B. J. (1970). The real majority: An extraordinary exam-
ination of the American electorate. New York, NY: Coward McCann.
Schafer, J. L., & Graham, J. W. (2002). Missing data. Our view of the state of the art.
Psychological Methods, 7, 147–177.
Schumpeter, J. A. (1950). Capitalism, socialism, and democracy, 3rd ed. New York, NY:
Harper and Brothers.
Schwartz, S. H. (1992). Universals in the content and structure of values: Theoretical
advances and empirical tests in 20 countries. In M. Zanna (Ed.), Advances in
experimental social psychology (pp. 1–65). Orlando, FL: Academic Press.
Schwartz, S. H. (1994). Are there universal aspects in the structure and contents of
human values? Journal of Social Issues, 50, 19–45.
Schwartz, S. H. (2003). A proposal for measuring value orientations across nations
(chapter 7). Retrieved from http://www.europeansocialsurvey.org/index.php?
Schwartz, S. H., Caprara, G. V., & Vecchione, M. (2010). Basic personal values, core
political values, and voting: A longitudinal analysis. Political Psychology, 31,
421–452.
Schwartz, S. H., Sagiv, L., & Boehnke, K. (2000). Worries and values. Journal of
Personality, 68, 309–346.
Schwarz, N., & Bohner, G. (2001). The construction of attitudes. In A. Tesser &
N. Schwarz (Eds.), Blackwell handbook of social psychology: Intraindividual processes
(pp. 436–457). Malden, MA: Blackwell.
Sears, D. O. (2001). The role of affect in symbolic politics. In J. H. Kuklinski (Ed.),
Citizens and politics: Perspectives from political psychology (pp. 14–40). New York,
NY: Cambridge University Press.
Sears, D. O., and Citrin, J. (1985). Tax revolt: Something for nothing in California
(Enlarged ed.). Cambridge, MA: Harvard University Press.
Sears, D. O., Hensler, C. P., & Speer, L. K. (1979). Whites opposition to “busing”:
Self-interest or symbolic politics. American Political Science Review, 73, 369–384.
Sears, D. O., Lau, R. R., Tyler, T. R., & Allen, H. M. (1980). Self-interest vs. symbolic
politics in policy attitudes and presidential voting. American Political Science
Review, 74, 670–684.
Shani, D. (2006). Knowing your colors: Can knowledge correct for partisan bias in
political perceptions? Unpublished manuscript.
260 REFERENCES
Shenkman, R. (2008). Just how stupid are we? Facing the truth about the American voter.
New York, NY: Basic Books.
Smith, E. R. A. N. (1989). The unchanging American voter. Berkeley, CA: University
of California Press.
Sniderman, P. M., Brody, R., & Tetlock, P. E. (1991). Reasoning and choice: Explorations
in political psychology. New York, NY: Cambridge University Press.
Sniderman, P. M., & Piazza, T. (1993). The scar of race. Cambridge, MA: Harvard
University Press.
Spini, D. (2003). Measurement equivalence of 10 value types from the Schwartz
value survey across 21 countries. Journal of Cross-Cultural Psychology, 34, 3–23.
Stimson, J. A. (1975). Belief systems: Constraint, complexity, and the 1972 election.
American Journal of Political Science, 19, 393–417.
Stokes, D. E. (1966). Party loyalty and deviating elections. In A. Campbell, P. E.
Converse, W. E. Miller, & D. E. Stokes (Eds.), Elections and the political order (pp.
125–135). New York, NY: Wiley.
Sundquist, J. L. (1968). Politics and policy: The Eisenhower, Kennedy, and Johnson years.
Washington, DC: Brookings Institution.
Sundquist, J. L. (1983). Dynamics of the party system: Alignment and realignment of
political parties in the United States (Revised ed.). Washington, DC: Brookings
Institution.
Suskind, R. (2006). The one percent doctrine: Deep inside America’s pursuit of its enemies
since 9/11. New York, NY: Simon & Schuster.
Taber, C. S., & Lodge, M. (2006). Motivated skepticism in the evaluation of political
beliefs. American Journal of Political Science, 50, 755–769.
Thompson, D. F. (1970). The democratic citizen: Social science and democratic theory in
the twentieth century. New York, NY: Cambridge University Press.
United States Census Bureau. (2011). Table A-1. Years of school completed by people 25
years and over, by age and sex: Selected years 1940 to 2010 [XLS Data file]. Retrieved
from http://www.census.gov/hhes/socdemo/education/data/cps/historical/
index.html.
Vavreck, L., & Rivers, D. (2008). The 2008 cooperative congressional election study.
Journal of Elections, Public Opinion and Parties, 18, 355–366.
Verba, S., Schlozman, K. L., & Brady, H. E. (1995). Voice and equality: Civic volunta-
rism in American politics. Cambridge, MA: Harvard University Press.
Verplanken, B., & Holland, R. W. (2002). Motivated decision making: Effects of
activation and self-centrality of values on choices and behavior. Journal of
Personality and Social Psychology, 82, 434–447.
Visser, P. S., Bizer, G. Y., & Krosnick, J. A. (2006). Exploring the latent structure of
strength-related attitude attributes. In M. Zanna (Ed.), Advances in experimental
social psychology (Vol. 38, pp. 1–67). San Diego, CA: Academic Press.
von Bothmer, B. (2010). Framing the sixties: The use and abuse of a decade from Ronald
Reagan to George W. Bush. Amherst, MA: University of Massachusetts Press.
References 261
Washington Post. (2012, February 12). Transcript: Illinois senate candidate Barack
Obama. Retrieved from http://www.washingtonpost.com/wp-dyn/articles/
A19751-2004Jul27.html.
White, T. H. (1973). The making of the president, 1972. New York, NY: Atheneum
Publishers.
Wicker, A. W. (1969). Attitudes versus actions: The relationship of verbal and overt
behavioral responses to attitude objects. Journal of Social Issues, 25, 41–78.
Wilentz, S. (2008). The age of Reagan: A history, 1974–2008. New York, NY: Harper.
Wilson, T. D., & Hodges, S. D. (1992). Attitudes as temporary constructions. In
L. L. Martin & A. Tesser (Eds.), The construction of social judgments (pp. 37–65).
Hillsdale, NJ: Lawrence Erlbaum.
Woolley, J., & Peters, G. (2011, July 25). The American presidency project. Retrieved
from http://www.presidency.ucsb.edu/.
Wuthnow, R. (1988). The restructuring of American religion: Society and faith since
World War II. Princeton, NJ: Princeton University Press.
Zaller, J. R. (1992). The nature and origins of mass opinion. New York, NY: Cambridge
University Press.
Zanna, M. P., & Rempel, J. K. (1988). Attitudes: A new look at an old concept. In
D. Bar-Tal & A. Kruglanski (Eds.), The social psychology of knowledge (pp. 315–334).
New York, NY: Cambridge University Press.
Zumbo, B., Gadermann, A. M., & Zeisser, C. (2007). Ordinal versions of coefficients
alpha and theta for Likert rating scales. Journal of Modern Applied Statistical
Methods, 6, 21–29.
This page intentionally left blank
INDEX
263
264 INDEX
Civil Rights Act, 19, 27, 39 core principles, 118, 156–57, 159
Clinton, Bill, 1, 19, 212 in policy voting, 235
abortion rights defense of, 82 sophistication interaction model
Bush, G. H. W., economic problems measuring, 126–27
charges from, 36–37 voter choice shaped by, 237
centrist position of, 20 correlation
Dole running against, 214 continuity, 112–17
military force used by, 37 corrected continuity, 114t
presidential campaign with, 209 knowledge, 76n6
Clinton, Hillary, 182 for limited government, 116–17
cognition policy principles, 127t
distinct groups with, 91 uncorrected, 109–10
-driven reasoning, 125–26 crime, fear of, 27
efficiency in, 57 Cronkite, Walter, 33
motivated social, 74 cultural liberalism, 29–30
networks with, 94 culture wars, 235–36
Colbert, Stephen, 1 abortion and, 131
Cold War, 31 over traditional morality, 27–28
Collins, Robert, 30 Supreme Court and, 24
comparative fit index (CFI), 100, 129
conditional slope estimates, 142–43 decision-making
confirmatory factor analysis (CFA), 91, heuristics, 124
98n4 limited government guiding, 216
conformity, 172, 173, 182, 191f strategies, 228–29
conscientiousness, 169 defense spending
conservation values, 165 military strength and, 151–56,
conservatism, 8, 51, 240 152t–154t
conservative doctrine, 47 Republican Party favoring, 151–54
conservative sophistication, 217 Delli Carpini, M. X., 10, 58, 76, 94, 243
constraint, 68 democratic citizenship, 59
Converse defining, 68 Democratic Party
horizontal, 68 Bush, G. H. W., chastising, 202–3
models of, 124–25 domestic policy liberalism
resource, 165n4 championed by, 19–20
vertical, 68 federal governments power and, 23
containment policy, 31–32 government capacities enlarged by,
continuity correlation, 112–17 18
conventional morality, 178 marginalized groups represented
Converse, P. E., 64, 78–79 by, 188
attitude-nonattitude continuum military power consensus collapsing
from, 44 in, 33
constraint defined by, 68 moral standards position of, 30–31
ideological labels examination of, 51 direct maximum likelihood, 105
political sophistication established dispositional model of attitudes, 43
by, 58 divorce rates, 26
voter competence reflections of, 3 Dobson, James, 28
core political values, 166–68 Dole, Bob, 214
266 INDEX