Government Policy and Citizen Passion: A Study of Issue Publics in Contemporary America Author(s): Jon A.

Krosnick Reviewed work(s): Source: Political Behavior, Vol. 12, No. 1, Cognition and Political Action (Mar., 1990), pp. 59-92 Published by: Springer Stable URL: http://www.jstor.org/stable/586285 . Accessed: 11/02/2012 15:55
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Springer is collaborating with JSTOR to digitize, preserve and extend access to Political Behavior.

http://www.jstor.org

Political Behavior, Vol. 12, No. 1, 1990

AND POLICY CITIZEN GOVERNMENT PASSION: A Studyof Issue Publicsin Contemporary America
Jon A. Krosnick
This article describes the findings of a program of research exploring the cognitive and behavioral consequences of passionate concern about government policy issues. American citizens vary a great deal in terms of the personal importance they attach to their attitudes on particular policy issues. Citizens whose policy attitudes are especially important to them are likely to think frequently about those attitudes, to perceive competing candidates as being relatively polarized on the issue, and to form presidential candidate preferences on the basis of those attitudes. Also, policy attitudes that citizens consider personally important are highly resistant to change and are therefore especially stable over long periods of time. The American public appears to be structured into many small issue publics, each composed of citizens who are passionately concerned about a single issue. Most Americans fall into very few issue publics, the particular ones being determined by each individual's unique self-interests, social identifications, and cherished values. The implications of these findings for the workings of democracies are discussed.

INTRODUCTION According to many political theorists, democratic governments maintain stability and legitimacy partly because their citizens elect representatives who implement government policies that they favor (e.g., Dahl, 1956; Pennock, 1979). This is presumed to occur because voters' candidate preferences are determined in part by the match between their attitudes toward government policies (i.e., their policy attitudes) and their perceptions of candidates' attitudes toward those policies. This notion, referred to as policy voting, is consistent with the many social psychological theories that assert that social attraction is based in part on attitudinal similarity (Byrne, 1971; Festinger, 1954; Heider, 1958). It is also consistent with the results of many studies of voting behavior in recent American presidential elections. In addition to affiliations with political parties,
Jon A. Krosnick, Departments of Psychology and Political Science, Ohio State University, 1827 Neil Avenue, Columbus, OH 43210. 59
0190-9320/90/0300-0059$06.00/0 ? 1990 Plenum Publishing Corporation

60

KROSNICK

assessments of candidates' personality traits, the emotions candidates evoke from voters, and evaluations of incumbent presidents' performance in office, candidate preferences are shaped by voters' policy attitudes and their perceptions of candidates' policy attitudes (see Kinder and Sears, 1985). Thus, citizens' policy preferences appear to play important roles in the maintenance and change of democracies. Although this conclusion is widely accepted among political scientists, there is much less agreement regarding the details of precisely how policy issues affect citizens' thoughts and behavior. During the past eight years, I have been conducting a program of research designed to clarify the social and psychological processes involved. My work has tested and refined a general theory of the role of policy attitudes in Americans' political cognition using two complementary methodologies: (1) secondary analysis of data from face-to-face and telephone surveys of representative national samples of American adults, and (2) controlled laboratory experiments, usually conducted with undergraduate subjects. In this article, I describe the theory and review the existing evidence attesting to its validity. AND AMERICAN CITIZEN POLICY ATTITUDE IMPORTANCE THE Since the early days of attitude research, psychologists have distinguished between attitudes in terms of their importance (Festinger, 1954, 1957; Newcomb, 1956, 1961), centrality (Converse, 1970; Katz, 1960), ego involvement (Krech and Crutchfield, 1948; Sherif and Cantril, 1947), and salience (Lemon, 1973; Smith, Bruner, and White, 1956). These properties are highly related conceptually, have often been used interchangeably (see, e.g., Scott, 1968, pp. 206-207; Sherif, 1980, pp. 2-4), and are usually defined in one of two ways. Some definitions propose that important, central, ego-involved, and salient attitudes are those that individuals are especially interested in and passionately concerned about (e.g., Converse, 1970; Freedman, 1964; Smith, Bruner, and White, 1956). Other definitions assert that important, central, ego-involved, and salient attitudes are those that are closely linked to individuals' basic values, needs, and goals (e.g., Converse, 1970; Katz, 1960; Lewin, 1951; Newcomb, Turner, and Converse, 1965; Sherif and Cantril, 1947). It seems most sensible to define attitude importance as the degree to which a person is passionately concerned about and personally invested in an attitude, and to view linkage between an attitude and values, needs, and goals as one possible cause of importance. Not surprisingly, American citizens vary a great deal in terms of the personal importance they attach to their attitudes on public policy issues. As the figures in Table 1 illustrate, the personal importance of various policy attitudes varied a great deal across Americans in 1968, 1980, and 1984. Only

GOVERNMENT POLICY CITIZEN AND PASSION

61

TABLE 1. NES 1968, 1980, and 1984: Distribution of Policy Attitude Importance in the Amerian Public Low Importance 1 High Importance 4 D 't Total

Issue 1968 Urban Unrest

Know

17.2% 14.2%

30.8% 42.0% 29.3% 46.3%

8.6% 8.4%

1.5% 100% 1.8% 100%

1980 Unemployment
(N = 3471)

8.8% 7.4% 9.2% 10.8% 16.8% 10.9% 14.8%

24.1% 33.0% 32.9% 32.4% 22.7% 35.6% 38.6%

11.2% 13.1% 14.6% 13.6% 13.7% 14.8% 13.0%

8.5% 15.8% 13.0% 12.7% 21.6% 12.2% 12.5%

47.3% 30.6% 30.3% 30.4% 25.3% 26.6% 22.2%

100% 100% 100% 100% 100% 100% 100%

Defense spending
(N = 3590)

Govt. services (N = 3521) Relations with Russia

Guaranteed full employment

Aid to minorities
(N = 3184)

1984 Govt. services

2.1% 5.0% 9.4% 4.2%

27.9% 32.3% 30.7% 25.0% 38.8% 23.1% 30.8% 33.1%

20.5% 14.0% 11.5% 17.0%

17.2% 100% 25.3% 100%

Involvement in Central America

Aid to women
(N = 2221)

17.3% 100% 15.0% 100%

Guaranteed full employment

Note: Most of the respondents the "Don'tKnow"categoryin the 1980 and 1984 surveys in were not askedthe importance questionbecausethey saidthey didn'tknowwhere they stood on the issue, or they said they didn'tknowwhere the federalgovernment stoodon the issue.

relatively small groups of citizens indicated extremely high levels of importance for each attitude (see Krosnick, 1986, for details). Social psychological theories suggest a number of possible consequences of this variation in policy attitude importance. For example, according to balance theory, cognitive imbalance occurs when one dislikes a person who holds an attitude similar to one's own, or when one likes a person who holds

62

KROSNICK

a contrasting attitude (Heider, 1958). The intensity of the discomfort that results from this imbalance presumably increases as the personal importance of the attitude increases (e.g., Singer, 1968; Festinger, 1957). Therefore, when an important attitude is involved, the consequent noxious state should be quite powerful and should demand swift reparation (see Cacioppo and Petty, 1981). Because this imbalance can be resolved by adjusting one's sentiment toward the other person, people probably come to like others whose attitudes are similar to their own important attitudes, and to dislike others whose attitudes conflict with their own important attitudes. Unimportant attitudes are less likely to serve as a basis for interpersonal sentiment. This reasoning has clear implications for the political candidate evaluation process: The impact of a policy attitude on a citizen's candidate preference should depend on the personal importance of the policy attitude to the voter. Important attitudes ought to have powerful impact, whereas unimportant attitudes ought to have little impact. There are at least three reasons why important policy attitudes should have more impact than unimportant ones on candidate evaluations. In order for a citizen to choose between competing presidential candidates on the basis of a policy attitude, two conditions must be fulfilled. First, the attitude must be cognitively accessible when candidates are evaluated; that is, the attitude must come to mind as a basis for choosing between candidates. Higgins and King (1981) suggest that the chronic accessibility of an attitude is determined by three factors: frequency of activation, distinctiveness, and the extent of links between it and other psychological elements. Because important attitudes are frequently subjects of conscious thought (Wood, 1982), are typically extreme (Brent and Granberg, 1982), and are probably extensively linked to other psychological elements (Judd and Krosnick, 1989; Newcomb, Turner, and Converse, 1965), these attitudes are likely to be highly accessible. Important attitudes are therefore more likely to come to mind as a criterion with which to evaluate political candidates. No matter how accessible a policy attitude is, though, a citizen cannot use it to chose between competing candidates unless he or she is aware of their stands on the issue and perceives them to differ from one another. Voters with important policy attitudes might therefore be expected to attend closely to candidates' public statements of their attitudes toward the policy so as to detect differences between them. If public statements do not reveal between-candidate differences, voters for whom a policy attitude is important may be especially likely to infer differences using cues such as party platforms, affiliations with individuals or groups known to have particular policy attitudes (e.g., endorsements by labor unions), and ideological labels with which candidates are described in the news media

GOVERNMENT POLICY CITIZEN AND PASSION

63

(e.g., Conover, 1981; Feldman and Conover, 1983). Thus, even when two candidates appear to take very similar stands toward a policy, voters whose attitudes toward it are important seem likely to infer between-candidate differences. As a result, they may find it especially easy to choose between the candidates on that basis. According to social judgment theory (Sherif and Hovland, 1953, 1961; Sherif, Sherif, and Nebergall, 1965), people assimilate communicators' attitudes that fall within their latitudes of acceptance toward their own attitudes and contrast communicators' attitudes within their latitudes of rejection away from their own. Important attitudes are thought to be especially powerful perceptual anchors that cause individuals to see others as falling primarily into one of two groups: those with whom they agree (at one extreme of the attitude continuum) and those with whom they disagree sharply (at the opposite extreme of the continuum) (e.g., Crano, 1983; Judd and Johnson, 1981). Therefore, among voters whose attitudes toward a policy are important, and who detect or infer that one candidate favors a policy and the other candidate opposes it, assimilation and contrast processes are likely to enhance the apparent magnitude of the difference between the candidates' attitudes. Such a difference would further facilitate choosing between these candidates. There is also a third reason why important policy attitudes are likely to be potent determinants of candidate evaluations. When a citizen recognizes that he or she disagrees with a liked candidate on a policy issue or agrees with a liked candidate, the discomfort due to the apparent cognitive imbalance can be resolved by changing one's own policy attitude. That is, one can change one's policy views to match those of liked candidates or to diverge from those of disliked candidates. This process, called persuasion, would be a relatively simple way to preserve one's candidate preferences and to eliminate the existing discomfort. Whereas this repair strategy is likely to be implemented in the case of unimportant policy attitudes, it is unlikely to occur with important policy attitudes, because important attitudes are likely to be highly resistant to change. Extensive linkage to other attitudes, beliefs, values, and psychological elements through a network of associations in memory exerts stabilizing forces (Ostrom and Brock, 1969). Important attitudes are presumably accompanied by large stores of relevant knowledge in memory, which equip individuals to counterargue against attitude-challenging information (Wood, Kallgren, and Preisler, 1985). People are attracted to and associate with others with whom they agree in terms of attitudes they consider important (e.g., Tedin, 1980), so these attitudes are reinforced by social support. And people are especially likely to commit themselves to their important attitudes in public (Krosnick, 1986; Schuman and Presser,

64

KROSNICK

1981, p. 242), which increases their resistance to change (Hovland, Campbell, and Brock, 1957). Information that challenges important attitudes is therefore likely to be subjected to biased elaboration that defends the attitudes (Petty and Cacioppo, 1986, pp. 111-140), so they are unlikely to change. For all these reasons, cognitive imbalances are unlikely to be resolved by changing important policy attitudes. Instead, policy attitudes that people consider to be personally important should have substantial impact on their candidate preferences, and unimportant policy attitudes should have little impact. Empirical Evidence Candidate Evaluations Recent national survey data handsomely validate the hypothesized role of attitude importance in guiding candidate evaluations. The figures in Table 2 are unstandardized regression coefficients estimating the impact of each policy attitude on candidate preferences (for details on the analytic procedure, see Krosnick, 1988a). This impact is quite strong among people
TABLE 2. NES 1968, 1980, and 1984: Unstandardized Regression Coefficients Estimating the Association Between Policy Attitudes and Candidate Preferences Low Importance 1 1968 Urban unrest (N = 1273) Vietnam (N = 1248) 1980 Unemployment (N = 1743) Defense (N = 2364) Govt. services (N = 2319) Russia (N = 2226) Abortion (N = 2422) Jobs (N = 2157) Minorities (N = 2287) 1984 Govt. services (N = 1614) Central America (N = 1460) Women (N = 1588) Jobs (N = 1631)
* p < .05.

2 .18* .10* .17* .17* .21* .08* -.03 .08* .22* .22* .14* .20* .21*

3 .32* .13* .28* .29* .35* .17* .01 .35* .32* .40* .28* .26* .36*

High Importance 4 .42* .24* .32* .30* .39* .30* .01 .36* .46* .56* .48* .43* .44*

.06 .07* .00 .14* .01 .01 -.11 .09 .07 .30 .02 .18* .14

AND PASSION POLICY CITIZEN GOVERNMENT

65

whose attitudes on a given issue are highly important to them personally, and is very weak among respondents whose attitudes are unimportant.

Attitude Accessibility

One of the probable reasons why important policy attitudes are more consequential is that they are more accessible in memory and are more likely to come to mind spontaneously in the course of thinking about presidential candidates. This hypothesis can be evaluated in two ways. First, one might examine the reasons survey respondents offer for liking or disliking presidential candidates. If a particular policy attitude is highly accessible in an individual's mind, he or she should be especially likely to mention that attitude as a basis for candidate evaluation. As the results in Table 3 illustrate, there is a strong relation between policy attitude importance and accessibility as measured in this fashion. People whose attitudes on a particularpolicy issue are more important are also more likely to mention that attitude as a reason for liking or disliking candidates (for details on the analytic method used, see Krosnick, 1988a). A second measure of the accessibility of an attitude is the length of time
TABLE 3. NES 1968, 1980, and 1984: Proportion of Respondents Mentioning Each Issue as a Reason to Vote For or Against Either of Two Presidential Candidates Low Importance 1 1968 Urban unrest (N = 1276) Vietnam (N = 1275) 1980 Unemployment (N = 772) Defense (N = 1127) Govt. services (N = 1131) Russia (N = 988) Abortion (N = 1233) Jobs (N = 955) Minorities (N = 1016) 1984 Govt. services (N = 1842) Central America (N = 1664) Women (N = 1836) Jobs (N = 1884) 2.8% 18.0% 16.0% 12.0% 4.8% 1.5% 2.6% 9.1% 0.6% 19.1% 17.1% 0.5% 20.4% High Importance 4 14.2% 37.5% 24.1% 25.4% 17.7% 2.3% 12.0% 14.0% 7.0% 31.4% 32.8% 0.4% 25.8%

2 5.5% 22.9% 17.6% 15.9% 12.8% 3.6% 3.7% 12.2% 2.4% 23.2% 21.6% 0.2% 21.6%

3 10.5% 29.2% 22.5% 16.3% 14.5% 8.9% 7.9% 16.2% 3.0% 31.6% 26.9% 0.2% 22.2%

66

KROSNICK

it takes an individual to report the attitude. Attitudes that people report quickly are highly accessible, whereas attitudes reported slowly are less accessible. In a series of laboratory studies, we had individuals report their attitudes on a series of political issues on a computer that timed their response latencies; they also reported the personal importance of each attitude. As Table 4 illustrates, response latencies were shorter among individuals for whom an attitude was more important. This result also supports the assertion that more important policy attitudes have more impact on candidate evaluations because these attitudes are more accessible in memory (for details, see Krosnick, 1989). Perceptions of Candidates' Attitudes People whose policy attitudes are important may find it easier to vote on that basis because they perceive relatively large differences between competing candidates' attitudes toward the policy. This claim can be tested by computing the proportion of people at various levels of attitude importance who perceive a large difference between competing presidential candidates' stands on various issues. These figures, displayed in Table 5, show that people who consider a policy attitude to be important are indeed more likely than those who consider it unimportant to perceive a substantial difference between competing candidates' attitudes. Table 5 provides an interesting explanation for the lack of an abortion attitude effect in Table 2. In 1980, only a small proportion of respondents in each importance group perceived a sizable difference between the TABLE 4. Attitude Self-Report Latencies (in seconds) for High- and LowImportanceSubjects Issue StudyOne Women'srights Abortion Race StudyTwo Abortion Defense Low Importance 3.71
(N = 59)

High Importance 3.31

p-value .022 .001 .071

4.59

3.68 3.98

(N = 29)

4.38

(N = 71)

4.42 (N = 48) 4.30 (N = 61)

3.89 (N = 103) 3.91 (N = 91)

.013 .028

AND PASSION POLICY CITIZEN GOVERNMENT

67

TABLE 5. NES 1968, 1980, and 1984: Proportion of Respondents Perceiving a Substantial Difference Between Two Competing Presidential Candidates' Attitudes on a Policy Issue Low Importance 1 1968 Urban unrest (N = 1304) Vietnam (N = 1313) 1980 Unemployment (N = 1777) Defense (N = 2418) Govt. services (N = 2383) Russia (N = 2311) Abortion (N = 1849) Jobs (N = 2242) Minorities (N = 2408) 1984 Govt. services (N = 1841) Central America (N = 1664) Women (N = 1832) Jobs (N = 1881) 41.2% 29.5% 35.2% 46.6% 39.3% 43.7% 11.4% 40.3% 36.0% 42.6% 36.0% 38.9% 40.2% High Importance 4 70.2% 53.6% 43.9% 59.2% 52.6% 48.9% 18.8% 53.8% 57.4% 71.6% 65.9% 60.4% 60.0%

2 51.3% 35.7% 39.1% 54.0% 45.6% 47.9% 13.5% 49.2% 45.9% 59.1% 47.4% 45.4% 45.7%

3 58.6% 44.3% 48.8% 58.9% 52.0% 54.1% 16.4% 54.9% 53.3% 72.9% 56.1% 54.1% 53.8%

candidates on that issue, which is probably why most people did not evaluate candidates on that basis. Among the other issues, the proportion of
respondents who perceived sizable candidate differences is closely related to the extent of policy voting. Across the high importance groups, the

correlation between the estimates in the fourth columns of Tables 2 and 5 is .73. The more people perceived a substantial candidate difference, the more likely they were to vote on the basis of the issue, which is consistent with the assertion that candidate perceptions partially mediate the effect of attitude importance. That is, attitude importance may enhance policy voting by causing perceptions of larger between-candidate differences. Resistance to Attitude Change Important attitudes are also likely to be more consequential in elections because these attitudes are likely to be highly resistant to change. Consistent with this hypothesis, important attitudes change less in response to persuasive communications administered in laboratory settings (Ewing, 1942; Fine, 1957; Knower, 1936; Rhine and Severance, 1970; Yankelovich, Skelly, and White, 1981). Also consistent with this hypothesis is evidence

68

KROSNICK

regarding attitude change during presidential election campaigns. As Table 6 illustrates, policy attitudes that people consider more personally important were more stable during the 1980 and 1984 U.S. presidential election seasons (for details on the method used, see Krosnick, 1988b). One likely reason for the enhanced stability of importantattitudes involves access to relevant knowledge. People are probably acutely attuned to information they encounter that is relevant to policy attitudes they consider personally important, whereas information relevant to unimportant attitudes is more likely to be ignored. Among informationthat is attended to, people are more likely to think deeply about and elaborate upon informationrelevant to importantpolicy attitudes. As a result, people probablyaccumulatelarge stores of knowledge in memory relevant to important policy attitudes and smaller amounts of knowledge relevant to unimportant policy attitudes. Supporting this hypothesis, people who say they consider an attitude to be more important also say they have more relevant knowledge stored in memory (Bradburn and Caplovitz, 1965; Hyman and Sheatsley, 1947, p. 416; Judd and Johnson, 1981; Krosnick, 1986; Vallone, Ross, and Lepper, 1985) and are able to report more of such knowledge (Wood, 1982). Another likely reason for the enhanced stability of important policy attitudes is linkage between these attitudes and core values. Conscious thought about cognitive elements linked by implicational relations enhances the consistency among these elements (e.g., Judd and Krosnick, 1989; McGuire, 1960). Because important attitudes are probably the subject of TABLE 6. NES 1980and 1984:Unstandardized RegressionCoefficientsEstimating the Stability of Attitudes After Correction for Measurement Unreliability Issue
1980

Low Importance .61 (N = 179) .19 .59 (N = 172) .58 (N = 185) .52 (N = 162) .60
(N = 203) (N = 122)

High Importance .91 (N = 115) .51 .75 (N = 167) .87 (N = 130) .85 (N = 329) .80
(N = 298) (N = 74)

services Government Unemployment Defense spending with Russia Relations 1984 services Government full Guaranteed employment

POLICY AND CITIZEN PASSION GOVERNMENT

69

frequent conscious copsideration, and are thought to be linked to individuals' core values, these attitudes are likely to be characterized by greater consistency with these values. Consistent with this hypothesis, pairs of policy attitudes are more ideologically consistent when one or both of the attitudes involved are personally important to an individual (Jackman,1977; Judd and Krosnick, 1989; Schuman and Presser, 1981; Smith, 1982). Also, as Table 7 shows, personally important policy attitudes are more consistent with liberal/conservative ideological orientations and with values for equality and individualism.
Ruling Out Random and Systematic Measurement Error

Nearly all of the apparent effects and correlates of attitude importance

Low Importance .09

High Importance .39

.22
(N = 261) .06 (N = 327)

.38
(N = 398) .60 (N = 313)

Relations with Russia Abortion Guaranteed full employment Aid to minorities Value: Equality
Racial integration Government services

.20
(N = 355)

.54
(N = 301)

.02
(N = 490)

.31
(N = 481)

.12
(N = 343)

.72
(N = 255)

.19
(N = 454) .58 (N = 70) .61 (N = 69)

.50
(N = 258 1.00 (N = 20) 1.05 (N = 50)

Value: Individualism Racial integration

.01
(N = 70) .52 (N= 69)

.59
(N = 20) .81 (N= 50)

Note: Cell entries are unstandardized regression coefficiencs estimating the effect of the value on the policy attitude.

70

KROSNICK

described above could be due to a simple methodological artifact: Perhaps survey respondents are able to report their important policy attitudes more precisely than they can report their unimportant policy attitudes. In fact, there are at least three reasons why random and systematic measurement errors in survey reports of policy attitudes may be less frequent when the attitude being measured is personally important to respondents (see Krosnick, 1988a; Krosnick and Schuman, 1988). First, because people presumably think relatively infrequently about unimportant attitudes, the internal cues associated with them may be more ambiguous, whereas important attitudes seem more likely to be represented by unambiguous, preconsolidated evaluative judgments in memory. Second, extensive thinking about an attitude object may prepare an individual to reliably impose the same meanings on ambiguous attitude questions and response choices, whereas people who have thought little about an object may be more unreliable in terms of their interpretations of questions and response alternatives. And third, the more important an attitude is to an individual, the more motivated he or she may be to provide a precise report of it. If these speculations were true, the results shown above would be far less interesting than I have portrayed them. Perhaps improved measurement precision would have eliminated most or all of the apparent effects of attitude importance. Fortunately, though, this is not the case. In a series of studies using longitudinal data, I have found no relation between attitude importance and the amount of random measurement error in policy attitude reports (Krosnick, 1988b). Furthermore, a set of 27 split-ballot experiments conducted in national surveys failed to reveal any general relation between attitude importance and systematic measurement error (or bias) in policy attitude reports (Krosnick and Schuman, 1988). These findings suggest that attitude importance is not related to the ambiguity of internal attitudinal cues, to ability to reliably interpret attitude questions, or to motivation to make precise attitude reports. Furthermore, these findings reinforce confidence that the evidence reviewed above does indeed illustrate substantive correlates and effects of policy attitude importance. IMPORTANCE AlTITUDE OF ORIGINS POLICY The empirical findings summarized above indicate that Americans vary in the personal importance they attach to their attitudes on particular policy issues. Policy attitudes that citizens consider important are highly accessible in memory, are highly resistant to change, are highly stable over time, are extensively linked to and consistent with individuals' basic values, instigate polarized perceptions of competing presidential candidates' policy attitudes, and are powerful determinants of candidate preferences. This accumulated

POLICY CITIZEN AND GOVERNMENT PASSION

71

body of evidence makes it clear that in order to fully understand the roles of policy attitudes in driving citizens' political cognition and behavior, we must recognize the impact of policy attitude importance. Given this evidence, a natural question to ask is, What leads individual citizens to consider some of their policy attitudes to be personally important and others to be personally unimportant? The political science literature offers two competing answers to this question: the educational stratification hypothesis and the issue public hypothesis.

EDUCATIONAL STRATIFICATION HYPOTHESIS Perhaps the most long-lived and widely cited theory of public opinion and political participation proposes that knowledge, power, and involvement are concentrated among a society's best educated citizens. People who are more educated are presumably more able and more motivated to attend to political events and to form relevant opinions. These individuals are equipped with more cognitive sophistication than others and with the ability to organize and manipulate abstract ideas, necessary requisites for understanding the complex world of politics. The more education one receives, the more one is trained to analyze human societies and to speculate about how life should be lived. Furthermore, formal education typically involves socializing people to value civic participation. As a consequence, this perspective argues, participation and interest in politics is concentrated among those who are best equipped and most motivated, those with the most formal education. There is much evidence to support this argument. Better educated citizens tend to vote more often in elections (Wolfinger and Rosenstone, 1980) and to be exposed to more information about politics, because they attend more to mass media sources of news (Comstock et al., 1978). Better educated individuals also seem to be more opinionated regarding government policy, because they tend to say "don't know" to policy attitude measures less often (Converse, 1977; Faulkenberry and Mason, 1978; Ferber, 1966; Francis and Busch, 1975; Schuman and Presser, 1981), and their opinions are more clearly organized according to ideological principles (e.g., Campbell et al., 1960; Converse, 1964). Better educated individuals tend to be more familiar with political issues (Campbell et al., 1960), to be more knowledgeable about political events (Neuman, 1985), to feel efficacious with regard to national politics, and to believe that government is responsive to citizens like themselves (Pomper, 1975). Better educated individuals are also more likely to derive their candidate preferences in

72

KROSNICK

presidential elections from their attitudes on a variety of policy issues (Carmines and Stimson, 1980; Miller et al., 1976; Pomper, 1975; Stimson, 1975). According to the educational stratification hypothesis, this last empirical regularity occurs because the best educated members of a society are likely to have personally important attitudes toward the wide range of government policy options being debated at any given time. Because they are exposed to a constant stream of information about a wide range of policy questions, they are especially likely to think about and form preferences on these many issues. In contrast, the least educated are unlikely to form attitudes toward any policy options, because these people lack the necessary information (for an example of this argument, see Almond, 1950). Therefore, the educational stratification hypothesis predicts that the more formal education an individual has acquired, the more personally important attitudes he or she is likely to have formed on issues of government policy. A society's least educated citizens would be expected to have the fewest important policy attitudes, and the most educated would be expected to have the most. Issue Public Hypothesis The issue public hypothesis, most prominently presented by Converse (1964), is based on three premises that contrast sharply with those of the educational stratification hypothesis. First, this perspective argues, politics is hardly the center of the vast majority of Americans' lives; once they have managed their immediate concerns of work and family, most people have few resources left to devote to careful and comprehensive study of the nation's political affairs (Lipmann, 1922, 1925). Second, the information costs entailed in becoming well informed about any given issue are substantial (e.g., Downs, 1957), so an individual can at best be expected to form highly important attitudes on a small handful of public policy issues. Third, one need not be especially cognitively sophisticated in order for attitudes toward government policies to become personally important. According to the issue public hypothesis, three classes of considerations can cause an attitude toward a government policy to become personally important to an individual, all of which are instances of an object "[gearing] into the primary goal and need structures of the individual (Converse, 1964, p. 181)." First, an attitude may become important to individuals who perceive the attitude object to be linked to their material self-interests. Self-interest-based importance presumably develops when people "expect [an object or issue] to have significant consequences for their own lives (Apsler and Sears, 1968, p. 162)," when an individual perceives that an attitude object is likely to have clear and direct impact upon his or her

POLICY CITIZEN AND PASSION GOVERNMENT

73

rights, privileges, or lifestyle (see also Key, 1961, pp. 220-227; Modigliani and Gamson, 1979, p. 11; Popkin, Gorman, Phillips, and Smith, 1976, p. 787). For example, a mother whose child would go to school in a faraway neighborhood if a school busing program were implemented in her area would presumably be more likely to consider her attitude toward school busing to be personally important than a woman whose children would not be affected by the busing plan. A second possible basis for a policy attitude to become personally important is social identification with reference groups or reference individuals. This circumstance may occur in a number of ways. First, strong identification with a group may lead an attitude to become important to a person if the group's rights or privileges are perceived to be at stake (Key, 1961; Modigliani and Gamson, 1979, p. 14). Thus, a black Wall Street executive who identifies closely with blacks as a group may care deeply about government welfare programs for blacks, even though he or she is unlikely to be affected directly by such programs. Second, strong identification with a group that consensually considers an attitude to be important can serve as an impetus for importance, independent of whether rewards to the group are in question (Sherif and Hovland, 1961). For example, people who strongly identify with Catholics are likely to develop important attitudes toward abortion, an issue on which the Catholic Church has publicly committed itself to a strong stand. In similar fashions, attitude importance may develop as a result of identification with reference individuals whose interests are perceived to be at stake in a policy debate or who are perceived to care deeply about a particular attitude. Third, a policy attitude may become important to an individual if he or she comes to view the policy as relevant to his or her basic social and personal values (Sherif and Cantril, 1947, p. 117; Modigliani and Gamson, 1979, p. 16). A value is "a type of belief, importantly located within one's total belief system, about how [people] ought or ought not to behave, or about some end-state of existence worth or not worth attaining. Values are thus abstract ideals, . . . not tied to any specific attitude object or situation, representing a person's beliefs about ideal modes of conduct and ideal terminal goals" (Rokeach, 1968, p. 124). The closer the perceived linkage between a policy issue and an individual's values, and the more important the values, the more important the attitude is likely to be to him or her (Campbell et al., 1960; Katz, 1960; Rosenberg, 1956; Ostrom and Brock, 1968). According to Rokeach (1973), values are "standardsemployed ... to tell us which beliefs, attitudes, values, and actions of others are worth challenging, protesting, and arguing about, or worth trying to influence or

74

KROSNICK

change (p. 13)." In this sense, values may tell people which attitudes to consider personally important. The issue public hypothesis therefore suggests that unanimity among Americans in terms of which policy attitudes are personally important to them is unlikely to have been the case during much of recent history. At times when the nation was focused on a major event or problem, such as a major war or economic depression, most people were probably concerned about the issue and developed relevant important attitudes. Such widespread importance is likely to result from the prolonged salience of the issue in citizens' information environments (Newcomb, Turner, and Converse, 1965, pp. 59-60; Becker and McLeod, 1976, p. 9), which probably leads many people to recognize connections between the public policy options and their self-interest, identification with Americans as a nation, and basic values. But when such prominent events or problems do not focus national attention, people are unlikely to focus their passions on the same small set of policy debates. Rather, individuals probably come through their own unique personal experiences and on the basis of their own self-interests, social identifications, and values to care deeply about idiosyncratic sets of public policy options. Therefore, a public policy option about which one citizen is passionately concerned is likely to be trivial to most others. Consistent with these assertions, Davis, Hinich, and Ordeshook (1970, p. 440) speculated that voters do not perceive,and especiallydo not have feelings aboutthe Individual as entire spectrumof issues. Instead,voters are characterized being concerned fromvoterto subsetof issueswith the contentof the subsetvarying with a narrow with and voter.Thus,farmers. . . careaboutfarmpricesupports those associated are the petroleumindustry ... concernedwith oil importquotaswhile the rest of us hardlyeven knowabout,and are certainlynot concernedwith, these issues. Various bits of incidental evidence support the issue public hypothesis. For example, a national survey conducted a few years ago found that about 50% of Americans either had contributed to, were members of, or were interested in joining one or more of sixteen types of single-issue-focused special-interest political organizations, though only small proportions of Americans expressed concern in this way about any one of the issues (Gallup, 1981). Second, when asked which issues play the most important roles in determining their presidential vote choices, the vast majority of Americans cite a wide array of issues, none of which attracts broad interest (Alderman, 1985, p. 29; Kelley, 1983; Pomper et al., 1985, p. 96). Congressmen see their constituents as structured into special-interest publics (Kingdon, 1981); the policy issues that citizens discuss when they

POLICY CITIZEN AND PASSION GOVERNMENT

75

contact government officials vary greatly across citizens (Verba and Nie, 1972). Work on abortion activists (e.g., Luker, 1984), the civil rights movement (e.g., Marx, 1967), the women's movement (e.g., Klein, 1984), the Vietnam War (e.g., Verba and Brody, 1970), and similar issues illustrates that small groups of individuals in society are consumed with passion about particular policy questions. All this suggests that a large segment of the nation may have personally important attitudes addressing questions of public policy, but in a patchwork quilt fashion. At most, each citizen probably has a small, idiosyncratic handful of such attitudes. And important attitudes on government policy issues are likely to occur equally often in all educational strata (see, e.g., Rosenau, 1961, pp. 37-40). Evaluations of the Two Hypotheses To formally evaluate these hypotheses, analyses were conducted to answer the following questions. First, how is the importance of one policy attitude related to the importance of others? Are the various importances overlapping only when content areas approach one another, as the issue public hypothesis proposes? Or are they powerfully related to one another regardless of content, as the educational stratification hypothesis proposes? Second, how many policy attitudes do respondents indicate are highly important to them? Is the distribution bimodal, with some people indicating many issues and others indicating none, as the educational stratification hypothesis predicts? Or is the distribution unimodal, with almost no one indicating that a great number of attitudes are highly important, as the issue public hypothesis predicts? Third, what are the relations of importance to education and political involvement? Are these relationships strong, as the educational stratification hypothesis predicts, or weak, as the issue public hypothesis predicts? Finally, are measures of perceived self-interest, social identification, and value relevance related to attitude importance, as the issue public hypothesis predicts? Correlations Among Importances The educational stratificationhypothesis proposes that the importances of various policy attitudes should be strongly positively intercorrelated. According to this view, a factor analysis of a matrix of correlations among various importance measures should indicate that a single-factor solution adequately summarizes the correlations. In contrast, the issue public hypothesis proposes that there should be little if any relationship between the importances of various issues except when the content areas of those issues approach one another. According to this perspective, we should have

76

KROSNICK

little success at predicting the importance of a person's attitude toward legalized abortion knowing the importance of his or her attitude toward economic aid to Mexico. Our success at prediction, however, would be markedly improved if the two attitudes both addressed aspects of social welfare policy. A first cut at testing these two competing predictions is to examine raw correlations among measures of the importance of various policy attitudes. However, these correlations are attenuated by random measurement error, and they are enhanced by correlated method variance (Alwin, 1974). Therefore, multitrait-multimethod confirmatory factor analysis methods must be applied to these correlations in order to assess the overlap between the importances of various policy attitudes more accurately. Corrected correlations among the importance measures included in the 1980 ad 1984 National Election Studies, as estimated by this method, are shown in Tables 8 and 9. These correlations are not at all consistent with a single-factor model. They are generally small except when policy issues are in the same general domain. For example, the correlation is large in the case of government services and government-guaranteed full employment, both of which fall into the category of federal social welfare programs. In cases where policy issues fall into very different domains, the correlations are close to zero, such as in the case of defense spending (which is a foreign policy and military question) and government aid to minorities (which is a domestic social welfare issue) in 1980. Among the 1980 measures, there seem to be two clusters, one of items related to domestic social welfare programs, and the other of items related to international strength. Among the 1984 measures, the one foreign relations item, regarding American involvement in Central America, stands alone. The items related to social welfare programs and guaranteed full employment are tightly linked, and the item on aid to women is moderately associated with the social welfare measures. Generally, though, only one of these
TABLE 8. NES 1980: Pearson Product-Moment Correlations Among Importance Measures Corrected for Unreliability and Method Covariance Issue Issue 1. 2. 3. 4. 5. 6. Unemployment Defense Services Russia Job Minorities 1 1.00 .06 .35 .09 .31 .06 2 1.00 .15 .24 .20 .05 3 4 5 6

1.00 .04 .43 .22

1.00 .25 .12

1.00 .41

1.00

Note:N = 1100.

AND PASSION POLICY CITIZEN GOVERNMENT

77

TABLE 9. NES 1984: Pearson Product-Moment Correlations Among Importance Measures Corrected for Random Error and Method Covariance

Issue Services CentralAmerica Women Job

Services 1.00 .21 .40 .83

Issue Central America 1.00 .19 .15

Women

Job

1.00 .45

1.00

correlations is as large as the educational stratification hypothesis predicts. These data therefore challenge that hypothesis and lend support to the issue public hypothesis. Distributions of High-Importance Group Membership As a second approach to evaluating the issue public and educational stratification hypotheses, we can examine the number of issues each respondent in the 1980 and 1984 NESs indicated is of the greatest possible importance to him or her. If the educational stratification hypothesis is correct, we would expect a bimodal distribution for this variable: There should be concentrations of individuals who consider none of the issues highly important and of individuals who consider all of the issues highly important. In contrast, the issue public hypothesis predicts that very few people should consider all of the issues highly important, so the distribution should be unimodal, with most people considering none, one, or two issues highly important. The results here again support the issue public idea. Table 10 displays the distributions for the 1980 and 1984 NES respondents. Instead of being bimodally distributed, the number of high-importance groups into which each respondent fell is unimodally distributed. Very few respondents fell into three or more high-importance groups, which constitutes further evidence that the importances of these attitudes are not determined by a single "general interest in politics" factor (for details, see Krosnick, 1986). Correlations of Importance with Education and Political Involvement A third evaluation of the educational stratification hypothesis was conducted by examining the correspondence between the individual

78

KROSNICK

TABLE 10. NES 1980and 1984:Membership HighestImportanceGroups in Numberof Highest Importance Groups
0 1 2 3 4 5 6

Percentof Respondents (1980)

Percentof Respondents (1984)

7 Total

0.4% 100%
(N = 3587)

100%
(N = 2127)

importance measures and measures of education and general interest in politics. If the importance of any particular policy attitude is simply a function of an individual's general interest in politics, there should be strong, positive correlations between measures of policy attitude importance and measures of general interest in politics. Furthermore, if general interest in politics is determined by educational attainment, correlations between years of education and the importance measures should be strong and positive. On the other hand, the issue public hypothesis predicts the correlations between importance measures and measures of general political interest and education should be relatively small. As Table 11 shows, Pearson product-moment correlations between the importance measures in the 1968, 1980, and 1984 NESs and measures of education and interest in politics are quite small. In the case of education, the correlations in 1968 and 1984 are generally positive, those in 1980 are negative, and all are small. The political interest indices are composed of questions about attention to political affairs, talking to others about politics, trying to convince others for whom to vote, and the like. The correlations involving these indices are somewhat larger than the education correlations but are hardly powerful. The correlations in 1968 and 1984 mostly fall between .10 and .25. In 1980, they are smaller, most being less than .10. Even acknowledging random measurement error in the measures, these data indicate that policy attitude importance is generally uncorrelated with education and with general interest in politics. Effects of Self-Interest, Social Identification, and Values We have tested the hypotheses that attitude importance is caused by

PASSION AND POLICY CITIZEN GOVERNMENT

79

TABLE 11. NES 1968, 1980, and 1984: Pearson Product-Moment Correlations Between Importance and Education and General Interest in Politics Issue 1968 Urban unrest Vietnam 1980 Unemployment Defense Services Russia Abortion Job Minorities 1984 Services Central America Women Job
* p < .05; ** p < .01.

Education .04*
(N = 1323)

Interest in Politics
.15** (N = 1318) .18**

.06*
(N = 1322)

(N = 1317) .02
(N = 1704)

- .07**
(N = 1825) -.04 (N = 2484) - .04* (N = 2450) -.10** (N = 2239)

.14** (N = 2294)
.09** (N = 2273) .09** (N = 2164)

- .08**

.04
(N = 2369)

(N = 2576)

- .07**

.06**
(N = 2109) .06**

(N = 2282) - .08**

(N = 2472) .06*
(N = 1840) .09** (N = 1662)

(N = 2235) .21*
(N = 1843)

.20** (N = 1665)
.10** (N = 1836)

.04
(N = 1832) -.01

(N = 1878)

.10* (N= 1884)

self-interest, social identification, and value relevance using two different methods: verbal protocols and correlational analysis (see Krosnick et al., 1988). In the protocol studies, subjects were asked to report the importance of a series of attitudes and were asked to list (1) the thoughts that went through their minds as they made those ratings, and (2) the reasons why they rated themselves as they did. Across a range of political attitudes, nearly all of the explanatory statements addressed self-interest, social identification, or value relevance, and no other causes of attitude importance were mentioned. Self-interest reasons were mentioned more frequently than were social identification or values reasons.

80

KROSNICK

In the correlational studies, respondents reported the personal importance of a series of attitudes, the degree to which their self-interests were involved in each attitude issue, the degree to which reference groups and reference individuals were affected by and concerned about each issue, and the degree to which their most important values were relevant to each issue. Multiple regressions predicting attitude importance revealed that all three predictors had substantial effects as expected across a range of issues, and self-interest had greater effects than social identification and value relevance (see Table 12). Taken together, this evidence provides initial support for the hypotheses that attitude importance is caused by self-interest, social identification, and value relevance. However, the validity of the protocol evidence depends on the validity of people's introspections, which has been called into question (Nisbett and Wilson, 1977). And the cross-sectional correlational evidence is inherently inconclusive regarding the directions of causal relations. Therefore, it is especially important to test these hypotheses via experiments. So far, only experiments testing the self-interest hypothesis have been conducted. As expected, increasing the degree to which an attitude object involved subjects' self-interest increased their concern about it (Apsler and Sears, 1968; Brickner, Harkins, and Ostrom, 1986; Cialdini et al., 1976; Madsen, 1978). Future experimentation will be needed to evaluate the causal impact of social identification and value relevance.

TABLE 12. Standardized Regression Coefficients Estimating the Effects of Self-Interest, Social Identification, and Value Relevance on Attitude Importance Predictor

SelfIssue Study One Capital punishment Abortion Study Two Racial integration Nuclear weapons Study Three Gun control Nuclear weapons Interest .49 .32 .48 .56 .68 .43

Social
Identification .24 .36 .23 -.10 + - .12 + .10 +

Value
Relevance .16 .17 .22 .27 .04 + .17 R2 .49 .44 .51 .43 .43 .37 N 147 141 76 64 110 106

+ All regression coefficients in this table are statistically significant (p < .05) except for those marked with a plus sign, which are nonsignificant.

POLICY CITIZEN AND PASSION GOVERNMENT

81

AND IMPLICATIONS SUMMARY Each day political events occur, and a small selection of these events are conveyed to the American public through the news media. Data are brought to the nation in convenient and discrete packages, in the morning paper and on the evening news. Between these doses of information, Americans have personal experiences that touch in one way or another on the world of politics. From this stream of data, experienced personally and received indirectly through reporters, each citizen must select what to attend to, what to think about, what to store in memory, and what to act on in the future. This selection is especially true during presidential election campaigns, when the volume of political information to which one has access is even greater than usual. In order to understand the forces that drive Americans' political behavior, particularly during elections, we must understand the processes by which information is gathered and integrated. According to my evidence, the nation may be conceived of as an amalgamation of issue publics, groups of people with highly important attitudes toward specific policy options. Individuals tend to belong to only a few issue publics, and it seems that the majority of Americans probably fall into at least one. Issue public members share a set of common characteristics. Their attitudes toward a policy option tend to be accessible, linked to other attitudes and values, bolstered by large stores of knowledge, and resistant to change. During presidential campaigns, important policy attitudes lead individuals to attend closely to relevant information that candidates provide. They prompt people to make inferences on the basis of this information in order to perceive candidates' positions on relevant issues. And when differences between competing candidates are perceived or inferred, they lead individuals to exaggerate the magnitude of these differences. When the candidates are evaluated, these attitudes serve as powerful bases for citizens' preferences. These findings have a number of implications for an understanding of American politics and presidential elections. Below is a discussion of implications regarding general models of voting, the meaning of election outcomes, the dynamics of support for incumbent presidents, and candidate campaign strategies. I then discuss the potential for application of the theory of importance described here to classes of political attitudes other than policy attitudes. General Models of Voting The finding that importance regulates policy voting is consistent with

82

KROSNICK

social psychology's view of individuals as cognitive misers who base judgments on a few salient criteria instead of on complete arrays of relevant knowledge. It might be argued on normative grounds that voters should use all their policy attitudes to generate candidate evaluations by a summing or averaging method, but this cognitive strategy would be costly and demanding. Voters seem to simplify this task by concentrating on only those policy attitudes that they consider important. This process may occur either as the result of deliberate, conscious decisions to focus on these attitudes or simply because these attitudes are especially accessible in memory and come to mind automatically as candidate evaluation standards. The present results do not suggest that people underutilize their attitudes on policy issues when deriving candidate evaluations in what might be called an irrational or counterproductive fashion. Rather, people seem to employ a sensible strategy that minimizes the cognitive costs of deriving candidate evaluations while maximizing subjective expected utility. Previous studies of voting behavior have been concerned largely with documenting associations between candidate preferences and predictors such as policy attitudes, political party identifications, perceptions of candidates' personalities, and retrospective assessments of incumbent performance. Simply documenting these associations does little to further development of a general theory of voting. Instead, the interacting variables that regulate the impact any given criterion has on candidate preferences must be identified. With regard to citizens' attitudes toward government policies, empirical research on regulators has focused primarily on the magnitude and clarity of differences between competing candidates in terms of the stands they take on issues (e.g., Miller et al., 1976; Pomper, 1972). When competing candidates clearly take sharply different positions on an issue, that issue typically has more impact on citizens' vote choices than when candidates are indistinguishable from one another. Thus, the emphasis in this work has been on factors external to the voter. The present research adds to this literature by demonstrating that internal factors regulate policy attitude impact as well. Candidates may regulate the magnitude of policy voting to some degree through their behavior, but it will also be determined in part by the personal importance voters attach to policy attitudes. The Meaning of Elections An important proposition in theories of representative democratic government is that election outcomes communicate messages to political leaders about their constituents' desires regarding government action. The winning candidate is presumed to be given a mandate to carry out the

AND PASSION POLICY CITIZEN GOVERNMENT

83

policies he or she endorsed during the campaign. The presumption here is that "the voters (or at least a majority of them) approve of all or most of the policies presented by the victorious candidate" (Polsby and Wildavsky, 1984, p. 270). The often-heard claim (e.g., Campbell et al., 1960) that most citizens are without many policy preferences suggests that the messages conveyed by election outcomes may instead be largely devoid of policy-relevant content. The research reported here suggests a somewhat different conclusion. Important policy attitudes play potent roles when Americans choose their president. Whereas unimportant attitudes are unrelated to candidate preferences, important attitudes are a useful basis for predicting citizens' votes, even over and above the implications of party affiliation, location in the social structure (as defined by demographic variables), ideological principles, and assessments of the incumbent's performance. Therefore, an election outcome does communicate policy preferences, but it does not convey majority support for the winning candidate's platform. Instead, it communicates a cacophony of desires on the part of many small minorities. In the extreme, important policy attitudes could be the basis of all citizens' vote choices while a majority of voters oppose all of the winner's policy proposals, and all voters oppose a majority of the winner's policy stands (Dahl, 1956; Polsby and Wildavsky, 1984). Clearly, the claim that electoral victory signifies a mandate for the winner's policy platform is most likely to be wrong. Judging from my evidence, then, a great number of citizens seem to be both rational and responsible when electing a president. By responsible, I mean, as Key (1966) did, that citizens' candidate preferences seem to be based on their evaluations of the candidates' policy stands. And by rational, I mean, as Goldberg (1969) did, that people seem to effectively combine their important policy attitudes and their perceptions of candidates' policy attitudes into vote choices in a manner resembling logical deduction. All this lends credibility to the claim that the electorate is not altogether irrational or out of touch when it casts votes in presidential elections. It is interesting to consider these conclusions in light of claims that the vast majority of Americans are uninformed and uninterested in the specifics of government policy and lack crystallized attitudes regarding policy options. My results are consistent with these propositions, because only a small proportion of people are likely to be knowledgeable about and to have potent attitudes regarding any single policy option. However, a large segment of the nation is likely to have informed, strong, and consequential attitudes regarding one policy option or another. It would therefore be wrong to suggest that most Americans have no potent policy attitudes at all. If this argument is valid, and if government is responsive to its citizens,

84

KROSNICK

we would expect to observe some correspondence between citizens' policy attitudes and government's actions. And furthermore, the resemblance of these two should be greater with regard to policy attitudes that are important to more people. Along these lines, Kingdon (1981) argued that "as [constituents' policy attitudes] become more intense, . .congressmen will weigh their opinions more heavily" (p. 35). Support for the Incumbent My conclusions have important implications regarding how popular support for incumbent presidents may vary between elections (see, e.g., Hibbs, Rivers, and Vasilotos, 1982). If votes are indeed driven by important policy attitudes, a majority of voters (i.e., those who voted for the winner) believe that the newly elected president will advocate in favor of their preferred alternatives (or at least will not advocate passionately for ones they oppose). Therefore, popular satisfaction with an election's outcome by the majority of voters is likely immediately afterward. However, as a president's term progresses, satisfaction presumably depends on which policies are enacted and which are foregone. Any given policy decision is not likely to please or anger a great proportion of citizens, so each individual legislative decision is not particularly consequential for the nation's satisfaction. Among people with important attitudes toward a policy option, the balance of favorable to unfavorable attitudes will determine the gains or losses entailed by a legislative decision regarding that option. The size of an assembly of those dissatisfied with such decisions will partly determine the level of popular support a president enjoys at any particular point in time. Candidate Campaign Strategies This research has highlighted the importance of political context in determining the impact a policy attitude has on candidate evaluations. This impact is regulated by the degree to which candidates distinguish their attitudes regarding the policy from each other. Even voters whose attitudes regarding abortion policy were highly important in 1980 failed to express these preferences through their votes. This failure seems to have occurred because even these voters were generally unable to perceive a difference between the candidates in terms of their policy attitudes on that issue. Attitudes regarding Vietnam-related policy were relatively unexpressed in the 1968 election, apparently for a similar reason. Thus, candidates do have some power to determine an issue's impact on an election outcome by either taking or avoiding a stand regarding relevant policy options. In light of this evidence, we might consider the argument made by

AND POLICY CITIZEN PASSION GOVERNMENT

85

Downs (1957), Page (1976, 1978), and Shepsle (1972) that politicians have incentives to be ambiguous and that they win more votes through vagueness than they would by taking clear policy stands. Though this argument may be generally true, there are some circumstances in which the incentives for ambiguity may be overridden. By clearly endorsing the policy option favored by the majority of people with highly important attitudes on some issue, a candidate may win their votes while losing fewer. Assembling coalitions of supporters in this manner is one way to assure electoral victory. If a candidate disagrees with the majority of citizens on an issue, it may be best to remain silent regarding specific policy proposals. But ambiguity seems unlikely to be the best approach for all issues. My evidence suggests that a candidate is unlikely to succeed in assembling supporters by persuading citizens to adopt his or her policy stands. When a candidate argues in favor of some policy option, his or her audience will be composed primarily of people with highly important attitudes regarding that policy; their opinions are not easily changed. If individuals with less important attitudes do happen to devote attention to such information, and do indeed adopt a candidate's attitude, their newly acquired preferences are unlikely to be expressed at the polls. A losing candidate is therefore likely to be wrong if he or she feels that "if he [or she] continues to educate the public to favor the policies he [or she] prefers, he [or she] will eventually win" (Polsby and Wildavsky, 1984, p. 273). Another strategy available to candidates is to manipulate the importance of voters' policy attitudes. By increasing the importance of attitudes regarding a policy debate on which a candidate's position is favored by a majority of the public, inconsequential attitudes may be called into action. And by reducing the importance of attitudes on an issue that is a candidate's liability, losing votes may be avoided. The evidence reported above suggests that the importance of a policy attitude may increase in response to a variety of events. First, it may increase if an individual comes to recognize that his or her immediate self-interest is at stake in an issue. Policy attitude importance may also increase if identification with a particular reference group or reference individual is enhanced, or if one comes to recognize that the interests of a reference group or individual with which he or she identifies are at stake, or if one comes to recognize that one's reference group or individual cares deeply about a particular issue. Finally, policy attitude importance may increase if the importance of relevant values increases or if the links between one's basic values and a policy issue are strengthened. Reductions in the importance of a policy attitude are probably unusual. Once bonds are established between aspects of the self and some attitude, they are likely to remain as long as the attitude object is at least moderately

86

KROSNICK

salient in one's environment. However, importance reduction could occur as a result of (1) activation of identification with groups or individuals whose self-interests are not at stake and who do not consider the issue an important one, or (2) a change in an individual's basic values. More commonly, importance probably decreases when information about an issue no longer comes to an individual through his or her daily experiences and through the news media. Extensions to Other Constructs This article has focused exclusively on policy attitudes, but the theory of importance presented here may be useful for understanding the impact that other sorts of factors have on candidate preference as well. For example, we might consider Stokes's (1963) valence issues. He argued that candidates are judged in part according to the likelihood that they can achieve desired policy ends, such as peace and prosperity, without regard for the specific policy means that bring them about. Attitudes toward these end-states are likely to be positive for most citizens, but they may vary in terms of importance. Those of greater importance are likely to have more impact on candidate evaluations. Jackson (1973) reported some evidence consistent with this argument. His analyses of 1960 NES data indicated that increased national importance assigned to a policy outcome was associated with a greater likelihood that a citizen will vote for the candidate perceived to be more likely to achieve the outcome. For example, respondents who felt that preventing war was more important than maintaining economic prosperity were more likely to vote for the candidate they felt was more likely to prevent war. This finding is consistent with my theory of importance. Similar extensions of the theory may be made to assessments of the incumbent's performance in office, political party affiliation, and other such consequential variables. CONCLUSION The research summarized here represents the first steps of an effort to further clarify the ways in which Americans think about their government's policies and the impact of Americans' policy preferences on their political behavior. To the extent that this approach is successful, it illustrates the potential for enhancing our understanding of American politics through the application of psychological theory. at This Acknowledgments. paperwas preparedfor presentation the Shambaugh and on Conference Communication, Cognitionand Political Judgments Action,held

PASSION AND CITIZEN POLICY GOVERNMENT

87

November 13-15 at the University of Iowa. This research was supported by a SPSSI Doctoral Dissertation Grant-In-Aid, a Dissertation Research Grant from the Horace H. Rackham School of Graduate Studies, a National Institute of Mental Health Graduate Training Fellowship, a Mershon Center Research Grant, and an Ohio State University Faculty Seed Grant.

REFERENCES
Alderman, J. D. (1985). Reagan won on the issues, not just his personality. In The '84 Vote, C. Smith (ed.). New York: ABC, Inc. Almond, G. A. (1950). The American People and Foreign Policy. New York:Praeger. Alwin, D. F. (1974). Approaches to the interpretation of relationships in the multitrait-multimethod matrix. In Sociological Methodology 1973-1974, H. L. Costner (ed.). San Francisco: Jossey-Bass. Apsler, R., and Sears, D. 0. (1968). Warning, personal involvement, and attitude change. Journal of Personality and Social Psychology 9: 162-166. Becker, L. B., and McLeod, J. M. (1976). Political consequences of agenda-setting. Mass Communication Review 3: 8-15. Bradburn, N. M., and Caplovitz, D. (1965). Reports on Happiness: A Pilot Study of Behavior Related to Mental Health. Chicago: Aldine. Brent, E. E., and Granberg, D. (1982). Subjective agreement and the presidential candidates of 1976 and 1980. Journal of Personality and Social Psychology 42: 393-403. Brickner, M. A., Harkins, S. G., and Ostrom, T. M. (1986). Effects of personal involvement: Thought-provoking implications for social loafing. Journal of Personality and Social Psychology 51: 763-769. Byrne, D. (1971). The Attraction Paradigm. New York: Academic Press. Cacioppo, J. T., and Petty, R. E. (1981). Effects of extent of thought on the pleasantness ratings of P-O-X triads: Evidence for three judgmental tendencies in evaluating social situations. Journal of Personality and Social Psychology 40: 1000-1009. Campbell, A., Converse, P. E., Miller, W. A., and Stokes, D. E. (1960). The American Voter. Chicago: University of Chicago Press. Carmines, E., and Stimson, J. (1980). The two faces of issue voting. American Political Science Review 74: 78-91. Cialdini, R. B., Levy, A., Herman, P., Kozlowski, L., and Petty, R. E. (1976). Elastic shifts of opinion: Determinants of direction and durability. Journal of Personality and Social Psychology 34: 663-672. Comstock, G., Chaffee, S., Katzman, N., McCombs, M., and Roberts, D. (1978). Television and Human Behavior. New York: Columbia University Press. Conover, P. J. (1981). Political cues and the perception of candidates. American Politics Quarterly 9: 427-448. Converse, J. M. (1977). Predicting no opinion in the polls. Public Opinion Quarterly 40: 515-530.

88

KROSNICK

Converse, P. E. (1964). The nature of belief systems in the mass public. In Ideology and Discontent, D. E. Apter (ed.). New York: Free Press. Converse, P. E. (1970). Attitudes and non-attitudes: Continuation of a dialogue. In The Quantitative Analysis of Social Problems, E. R. Tufte (ed.). Reading, MA: Addison-Wesley. Crano, W. D. (1983). Assumed consensus of attitudes: The effect of vested interest. Personality and Social Psychology Bulletin 9: 597-608. Dahl, R. A. (1956). A Preface to Democratic Theory. Chicago: University of Chicago Press. Davis, O. A., Hinich, M. J., and Ordeshook, P. C. (1970). An expository development of a mathematical model of the electoral process. American Political Science Review 64: 426-448. Downs, A. (1957). An Economic Theory of Democracy. New York: Harper. Ewing, T. N. (1942). A study of certain factors involved in changes of opinion. Journal of Social Psychology 16: 63-88. Faulkenberry, G. D., and Mason, R. (1978). Characteristics of nonopinion and no opinion response groups. Public Opinion Quarterly 42: 533-543. Feldman, S., and Conover, P. J. (1983). Candidates, issues and voters: The role of inference in political perception. Journal of Politics 45: 810-839. Ferber, R. (1966). Item nonresponse in a consumer survey. Public Opinion Quarterly 30: 399-415. Festinger, L. (1954). A theory of social comparison processes. Human Relations 7: 117-140. Festinger, L. (1957). A Theory of Cognitive Dissonance. Stanford, CA: Stanford University Press. Fine, B. J. (1957). Conclusion-drawing, communicator credibility, and anxiety as factors in opinion change. Journal of Abnormal and Social Psychology 54: 369-374. Francis, J. D., and Busch, L. (1975). What we now know about "I don't knows." Public Opinion Quarterly 39: 207-218. Freedman, J. L. (1964). Involvement, discrepancy, and change. Journal of Abnormal and Social Psychology 69: 290-295. Gallup, G. (1981). The Gallup Report, no. 191. Princeton, NJ: Gallup Organization. Goldberg, A. S. (1969). Social determinism and rationality as bases of party identification. American Political Science Review 63: 5-25. Heider, F. (1958). The Psychology of Interpersonal Relations. New York: Wiley. Hibbs, D. A., Rivers, R. D., and Vasilotos, N. (1982). The dynamics of political support for American presidents among occupational and partisan groups. American Journal of Political Science 26: 312-332. Higgins, E. T., and King, G. (1981). Accessibility of social constructs: Informationprocessing consequences of individuals and contextual variability. In Personality, Cognition and Social Interaction, N. Cantor and J. Kihlstrom (eds.). Hillsdale, NJ: Erlbaum Associates. Hovland, C. I., Campbell, E. H., and Brock, T. C. (1957). The effects of "Commitment" on opinion change following communication. In The Order of Presentation in Persuasion, C. I. Hovland, W. Mandell, E. H. Campbell, T. C. Brock, A. S.

PASSION AND POLICY CITIZEN GOVERNMENT

89

Luchins, A. R. Cohen, W. J. McGuire, I. L. Janis, R. L. Feierabend, and N. H. Anderson. New Haven, CT: Yale University Press. Hyman, H. H. and Sheatsley, P. B. (1947). Some reasons why information campaigns fail. Public Opinion Quarterly 11: 412-423. Jackman, M. R. (1977). Prejudice, tolerance, and attitudes toward ethnic groups. Social Science Research 6: 145-169. Jackson, J. E. (1973). The importance of issues and issue importance in presidential elections. Harvard University. Typescript. Judd, C. M., and Johnson, J. T. (1981). Attitudes, polarization, and diagnosticity: Exploring the effects of affect. Journal of Personality and Social Psychology 41: 26-36. Judd, C. M., and Krosnick, J. A. (1989). The structural bases of consistency among political attitudes: The effects of political expertise and attitude importance. In Attitude Structure and Function, A. R. Pratkanis, S. J. Breckler, and A. G. Greenwald (eds.). Hillsdale, NJ: Erlbaum Associates. Katz, D. (1960). The functional approach to the study of attitudes. Public Opinion Quarterly 24: 163-204. Kelley, S. (1983). Interpreting Elections. Princeton: Princeton University Press. Key, V. 0. (1961). Public Opinion and American Democracy. New York: Knopf. Key, V. 0. (1966). The Responsible Electorate. Cambridge, MA: Harvard University Press. Kinder, D. R., and Sears, D. 0. (1985). Public opinion and political action. In Handbook of Social Psychology, Vol. 2, G. Lindzey and E. Aronson (eds.). Hillsdale NJ: Erlbaum Associates, pp. 659-741. Kingdon, J. W. (1981). Congressmen's Voting Decisions. New York:Harper and Row. Klein, E. (1984). Gender Politics: From Consciousness to Mass Politics. Cambridge, MA: Harvard University Press. Knower, F. H. (1936). Experimental studies of changes in attitude-III: Some incidence of attitude changes. Journal of Applied Psychology 20: 114-127. Krech, D., and Crutchfeld, R. S. (1948). Theory and Problems of Social Psychology. New York: McGraw-Hill. Krosnick, J. A. (1986). Policy voting in American presidential elections: An application of psychological theory to American politics. Unpublished doctoral dissertation, University of Michigan, Ann Arbor, Michigan. Krosnick, J. A. (1988a). The role of attitude importance in social evaluation: A study of policy preferences, presidential candidate evaluations, and voting behavior. Journal of Personality and Social Psychology 55: 196-210. Krosnick, J. A. (1988b). Attitude importance and attitude change. Journal of Experimental Social Psychology 24: 240-255. Krosnick,J. A. (1989). Attitude importance and attitude accessibility. Personality and Social Psychology Bulletin, 15: 297-308. Krosnick,J. A., and Schuman, H. (1988). Attitude intensity, importance and certainty and susceptibility to response effects. Journal of Personality and Social Psychology 54: 940-952. Krosnick,J. A., Boninger, D. S., Berent, M. K., and Carnot, C. G. (1988). The origins

90

KROSNICK

of attitude importance. Presented at the annual meeting of the Midwestern Psychological Association, Chicago, IL. Lemon, N. (1973). Attitudes and Their Measurement. London: Batsford. Lewin, K. (1951). Field Theory in Social Science. New York: Harper. Lippmann, W. (1922). Public Opinion. New York: Macmillan. Lippmann, W. (1925). The Phantom Public. New York: Harcourt, Brace. Luker, K. (1984). Abortion and the Politics of Motherhood.Berkeley, CA: University of California Press. Madsen, D. B. (1978). Issue importance and group choice shifts: A persuasive arguments approach.Journal of Personality and Social Psychology 36: 1118-1127. Marx, G. T. (1967). Protest and Prejudice:A Study of Belief in the Black Community. New York: Harper and Row. McGuire, W. J. (1960). A syllogistic analysis of cognitive relationships. In Attitude Organization and Change, M. J. Rosenberg, C. I. Hovland, W. J. McGuire, R. P. Abelson, and J. W. Brehm. New Haven, CT: Yale University Press. Miller, A. H., Miller, W. E., Raine, A. S. and Brown, T. A. (1976). A majority party in disarray: Policy polarization in the 1972 election. American Political Science Review 70: 753-778. Modigliani, A., and Gamson, W. A. (1979). Thinking about politics. Political Behavior 1: 5-30. Neuman, W. R. (1985). The Paradox of Mass Publics: Knowledge and Opinion in the American Electorate. Cambridge, MA: Harvard University Press. Newcomb, T. M. (1956). The prediction of interpersonal attraction. American Psychologist 11: 575-586. Newcomb, T. M. (1961). The Acquaintance Process. New York: Holt, Rinehart, and Winston. Newcomb, T. M., Turner, R. H., and Converse, P. E. (1965). Social Psychology. New York: Holt, Rinehart, and Winston. Nisbett, R. E., and Wilson, T. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review 84: 231-259. Ostrom, T. M., and Brock, T. C. (1968). A cognitive model of attitudinal involvement. In Theories of Cognitive Consistency: A Sourcebook, R. P. Abelson et al. (eds.). Chicago: Rand McNally. Ostrom, T. M., and Brock, T. C. (1969). Cognitive bonding to central values and resistance to a communicating change in policy orientation. Journal of Experimental Research in Personality 4: 42-50. Page, B. I. (1976). The theory of political ambiguity. American Political Science Review 70: 742-752. Page, B. I. (1978). Choices and Echoes in Presidential Elections: Rational Man and Electoral Democracy. Chicago: University of Chicago Press. Pennock, J. R. (1979). Democratic Political Theory. Princeton, NJ: Princeton University Press. Petty, R. E., and Cacioppo, J. T. (1986). Communication and Persuasion: Central and Peripheral Routes to Attitude Change. New York: Springer-Verlag. Polsby, N. W., and Wildavsky, A. (1984). Presidential Elections. New York:Scribner.

AND CITIZEN PASSION POLICY GOVERNMENT

91

Pomper, G. M. (1972). From confusion to clarity: Issues and American voters, 1956-1968. American Political Science Review 66: 415-428. Pomper, G. M. (1975). Voters' Choice. New York: Harper and Row. Pomper, G. M., Baker, R. K., Jacob, C. E., Keeter, S., McWilliams, W. C. and Plotkin, H. A. (1985). The Election of 1984: Reports and Interpretations. Chatham, NJ: Chatham House. Popkin, S., Gorman, J. W., Phillips, C., and Smith, J. A. (1976). Comment: What have you done for me lately? Toward an investment theory of voting. American Political Science Review 70: 779-805. Rhine, R. J., and Severance, L. J. (1970). Ego-involvement, discrepancy, source credibility, and attitude change. Journal of Personality and Social Psychology 16: 175-190. Rokeach, M. (1968). Beliefs, Attitudes, and Values. San Francisco: Jossey-Bass. Rokeach, M. (1973). The Nature of Human Values. New York:The Free Press. Rosenau, J. N. (1961). Public Opinion and Foreign Policy. New York:Random House. Rosenberg, M. J. (1956). Cognitive structure and attitudinal affect. Journal of Abnormal and Social Psychology 53: 367-372. Schuman, H., and Presser, S. (1981). Questions and Answers: Experiments on Question Form, Wording, and Context in Attitude Surveys. New York:Academic Press. Scott, W. A. (1968). Attitude measurement. In The Handbook of Social Psychology, 2nd ed., Vol. 2, G. Lindzey and A. Aronson (eds.), Reading, MA: Addison-Wesley. Shepsle, K. A. (1972). The strategy of ambiguity: Uncertainty and electoral competition. American Political Science Review 66: 555-568. Sherif, C. W. (1980). Social values, attitudes, and the involvement of the self. In Nebraska Symposium on Motivation 1979: Beliefs, Attitudes and Values, M. M. Page (ed.). Lincoln: University of Nebraska Press, pp. 1-64. Sherif, C. W., Sherif, M., and Nebergall, R. E. (1965). Attitude and Attitude Change. Philadelphia: W. B. Saunders. Sherif, M., and Cantril, H. (1947). The Psychology of Ego-involvements. New York: Wiley. Sherif, M., and Hovland, C. I. (1953). Judgmental phenomena and scales of attitude measurement: Placement of items with individual choice of number of categories. Journal of Abnormal and Social Psychology 48: 135-141. Sherif, M., and Hovland, C. I. (1961). Social Judgement: Assimilation and Contrast Effects in Communication and Attitude Change. New Haven, CT: Yale University Press. Singer, J. E. (1968). The bothersomeness of inconsistency. In Theories of Cognitive Consistency: A Sourcebook, R. P. Abelson, E. Aronson, W. J. Mcguire, T. M. Newcomb, M. J. Rosenberg, and P. H. Tannenbaum (eds.). Chicago: Rand McNally, pp. 393-399. Smith, M. B., Bruner, J. S., and White, R. W. (1956). Opinions and Personality. New York: Wiley. Smith, T. W. (1982). Attitude constraint as a function of non-affective dimensions.

92

KROSNICK

General Social Survey Technical Report No. 39. Chicago: National Opinion Research Center. Stimson, J. A. (1975). Belief systems: Constraint, complexity and the 1972 election. American Journal of Political Science 19: 393-418. Stokes, D. E. (1963). Spatial models of party competition. American Political Science Review 57: 368-377. Tedin, K. L. (1980). Assessing peer and parent influence on adolescent political attitudes. American Journal of Political Science 24: 136-154. Vallone, R. P., Ross, L., and Lepper, M. R. (1985). The hostile media phenomenon: Biased perception and perceptions of media bias in coverage of the Beirut massacre. Journal of Personality and Social Psychology 49: 577-585. Verba, S., and Brody, R. A. (1970). Participation, political preferences, and the War in Vietnam. Public Opinion Quarterly 34: 325-332. Verba, S., and Nie, N. H. (1972). Participation in America: Political Democracy and Social Equality. New York: Harper. Wolfinger, R. E., and Rosenstone, S. J. (1980). Who Votes? New Haven, CT: Yale University Press. Wood, W. (1982). Retrieval of attitude-relevant informationfrom memory: Effects on susceptibility to persuasion and on intrinsic motivation. Journal of Personality and Social Psychology 42: 798-810. Wood, W., Kallgren, C. A., and Preisler, R. M. (1985). Access to attitude-relevant information in memory as a determinant of persuasion: The role of message attributes. Journal of Experimental Social Psychology 21: 73-85. Yankelovich, D., Skelly, F., and White, A. (1981). The Mushiness Index: A Refinement in Public Policy Polling Techniques. New York: Yankelovich, Clancey, Shulman.