Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

Social Groups and Political Judgments Author(s): Christopher Wlezien and Arthur H. Miller Source: Social Science Quarterly, Vol. 78, No. 3 (September 1997), pp. 625-640 Published by: University of Texas Press Stable URL: http://www.jstor.org/stable/42863557 Accessed: 05-04-2016 19:00 UTC Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://about.jstor.org/terms 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. University of Texas Press, Wiley are collaborating with JSTOR to digitize, preserve and extend access to Social Science Quarterly

Social Groups and Political Judgments* Christopher Wlezien, University of Houston Arthur H. Miller, University of Iowa Objective. Much research addresses the influence of issues on political judgments, such as individuals' assessments of parties and candidates. Other, more recent, research suggests that these judgments largely reflect perceptions of social groups in society. We address the relative influence of issue attitudes and group evaluations on party and candidate preferences in the United States. Methods. Building on Abramowitz's (1994) models, we use regression analysis to assess whether and how group evaluations influence party identification and presidential candidate evaluations, relying on data from the National Election Studies in 1988 and 1992. Results. Issue attitudes do have an important influence on individuals' political judgments, but people's evaluations of groups associated with the political parties also influence their political judgments independently of issue attitudes. The politically relevant groups do not merely represent traditional interests, such as labor and business, or even race - indeed, they primarily represent cultural differences. Conclusions. It appears that evaluations of groups and, thus, judgments of the parties and candidates themselves reflect a broad ideology of intergroup relations. Many writers have argued that American politics - including not only partisan and electoral politics, but also basic beliefs and ideology - continues to reflect deep divisions about race-related issues (e.g., Sears, Hensler, and Speer, 1979; Kinder and Sears, 1981; Sniderman, Piazza, Tetlock, and Kendrick, 1991; Tate, 1993). Indeed, during the past few years, racial issues have become widely accepted as the single best explanation of partisan differences in America (see Carmines and Stimson, 1989; Pomper et al., 1989; McWilliams, 1989; Edsall and Edsall, 1991). This explanation, however, runs counter to the more traditional understanding of post- * Direct all correspondence to Christopher Wlezien, Department of Political Science, University of Houston, Houston, TX 77204-3474. An earlier version of this article was presented at the Annual Meeting of the American Political Science Association, New York, August 31-September 4, 1994. We thank Andy Peebler for research assistance, Cristina Adams, Pamela Conover, Gary Dworkin, Robert Erikson, Richard Lau, Jon Lorence, Mark Lundgren, Kathleen McGraw, and this journal's anonymous referees for comments, and Peggy Swails for secretarial assistance. All of the data used herein are available through the Inter-university Consortium for Political and Social Research at the University of Michigan. Editor's Note: Reviewers were Herbert B. Asher, David Sears, and Frank P. Zinni, Jr. SOCIAL SCIENCE QUARTERLY, Volume 78, Number 3, September 1997 1997 by the University of Texas Press, P.O. Box 7819, Austin, TX 78713-7819

626 Social Science Quarterly New Deal party differences in terms of social welfare issues (Campbell, Converse, Miller, and Stokes, 1960; Asher, 1976) and to the findings of those who contend that, when issue attitudes are the focus, social welfare issues still provide a better explanation of the transformation of party politics than do racial issues (see, e.g., Miller, 1987; Stone, 1987; Abramowitz, 1994). Other recent research reflects an alternative approach, one that stresses reference groups rather than issue attitudes as important indicators of (and possible explanations for) political divisions in America (Miller and Wlezien, 1993; Miller, Wlezien, and Hildreth, 1991). 1 This approach is consistent with research demonstrating that a substantial proportion of the citizenry organizes its thinking about politics in terms of groups rather than issues (Hamill, Lodge, and Blake, 1985; Lau, 1986) and that evaluations and cognitions involving social groups influence a variety of political attitudes, including political ideology and policy preferences (Sears, Lau, Tyler, and Allen, 1980; Brady and Sniderman, 1985), and evaluations of government economic performance (Conover, 1987). In general, all of this work reveals an important influence of the perceptions of various social groups - not just blacks - on political attitudes and behavior. This article assesses the influence of group evaluations on political judgments in the United States. To begin with, we briefly outline the "issues" and "reference group" models, focusing on the differences and interrelationships between the two approaches. Then, building on Abramowitz's (1994) recent research, we address the relationships between social groups and both party identification and candidate evaluations. We focus particularly on whether the effects of social group evaluations are independent of ideology, issues, and other factors that influence political judgments. The implications of the findings are considered in the concluding section. Issues and Political Judgments The importance of individual opinions about issues has a long history in political science. Whether conceived to be largely exogenous or not, issue positions are considered among the most potent predictors of both party and candidate evaluations (see, e.g., Campbell, Converse, Miller, and Stokes, 1960; Markus and Converse, 1979; Asher, 1992; Abramowitz, 1994). Traditionally, scholars have emphasized the importance of domestic 1This approach differs from research that focuses on individuals' membership in demographic categories (Axelrod, 1970; Stanley, Bianco, and Niemi, 1986; Erikson, Lancaster, and Romero, 1988). While membership in a group surely influences affect toward the particular group, there is considerable (and important) variation in affect within particular demographic groups (see Miller, Wlezien, and Hildreth, 1991). Perhaps more important, affect toward various "other" groups of which individuals are not members may structure their political judgments. The reference group approach takes this structuring into account.

Social Groups and Political Judgments 627 and foreign policy attitudes for understanding political evaluations at particular points in time and over time. Carmines and Stimson (1986, 1988) challenged this traditional characterization. They argued that racial issues, not social welfare and national security, have driven the most recent developments in partisan and candidate evaluations. Carmines and Stimson's (1989) theory of racial issue evolution is first and foremost an examination of how public opinion changes over time. According to Carmines and Stimson, the differential positions on racial issues taken by Democratic and Republican political leaders after the mid-1960s eventually led to a polarization of party supporters along the same lines. The process of partisan transformation occurred gradually. As newly entering cohorts of younger voters realized that the Democratic party was more liberal on racial issues and the Republican party was more conservative, they came to identify with the party that best reflected their preferred position on racial issues. Over time, the new partisan cohorts, which were divided sharply on racial issues, became an increasingly larger proportion of all identifiers, thereby pushing the parties farther and farther apart on the racial dimension. By the 1980s, party leaders and candidates not only had become visibly differentiated on the basis of racial issues, but the partisan alignments, candidate choices, and broader political ideology of the citizens had come to strongly reflect this division. There is considerable agreement on certain evidence supporting Carmines and Stimson's theory, but considerable disagreement about the explanations for the evident patterns. No one questions the following three trends: (1) since the mid-1960s, the Democratic Party has been more supportive of policies aimed at improving the circumstances of blacks while the Republican Party has been less supportive; (2) the public has become increasingly aware of party differences on racial issues; and (3) in recent years, Democratic identifiers, on average, have been relatively more liberal on racial issues than have Republican identifiers. Beyond this basic evidence, however, there is considerable disagreement. As Abramowitz (1994: 4) argues, "The fact that Democratic and Republican partisans have become more polarized on racial issues since 1964 does not necessarily mean that voters were choosing parties on the basis of their racial attitudes." Abramowitz shows that partisan change since 1964 is largely unrelated to racial attitudes and is best accounted for by more traditional dimensions, namely, social welfare and national security issues. Although they draw contrasting conclusions, both Carmines and Stimson (1986, 1988) and Abramowitz (1994) stress the importance of issues in structuring political evaluations. Our position is quite different. We contend that political judgments (perhaps issue positions and even general ideological orientations) are at least partially the by-product of social group conflict. The approach actually has deep roots in political science research (see, e.g., Campbell, Converse, Miller, and Stokes, 1960).

628 Social Science Quarterly Groups As Political Cues The early treatment of reference groups placed a great deal of emphasis on face-to-face interactions and group cohesion (see Lau, 1983, for an excellent review of the relevant literature). More recent work focuses on information processing and treats social groups as categories that provide a source of identification for group members or information cues for nonmembers (see, among others, Tajfel, 1972, 1978; Conover and Feldman, 1981; Fiske and Taylor, 1984; Lau and Sears, 1986). Within this newer theoretical framework, there is no longer the need for the very restrictive assumption of face-to-face interaction for the group to influence individual behavior.2 Merely the perception that one shares an interest with a group is sufficient to differentiate how people will act toward in-group and outgroup members (Brewer, 1979; Jackson and Sullivan, 1987; Tajfel, Flament, Billig, and Bundy, 1971; Lau, 1989). The major components of the theory can be summarized as follows. The complex world of politics must be simplified and organized if it is to have relevance and meaning. In dealing with this complexity and everyday flow of information, individuals sort objects, people, and events into broad categories that are then used as shortcuts for (efficient) processing of information in a coherent rather than a piecemeal fashion. Social groups regularly serve as important and relevant categories that influence one's self-concept as well as one's understanding of social and political relations (Gurin, Miller, and Gurin, 1980). Social groups are visible actors in the political arena, and power struggles between competing groups are salient to the public. The public also has a sense of the status differences between groups and whether these differences are, for them, legitimate or illegitimate (Parkin, 1971; Miller, Gurin, Gurin, and Malanchuk, 1981; Jackman and Muha, 1984; Bobo, 1988; Sidanius, 1993). Some groups are more visibly involved in and salient to politics than others. Over time, citizens develop a sense of those groups with which they 2 The literature (see Lau, 1983; Deaux, 1993) suggests that reference groups serve three major functions for individuals: a source of membership, an information source, and a source of self-identification. People may claim membership in a variety of groups when describing themselves with respect to ascribed categories such as gender, race, or ethnicity. They also can be members of associations or organizations that give political representation or clout to members, such as the Gray Panthers. For those who do not think of themselves as members of particular groups, however, the groups may still act as a source of information. The actions of various groups, and a comparison of groups along various dimensions relevant to politics, may be an important source of information for individuals, regardless of membership. Similarly, an individual can have a psychological feeling of identification with a group without exhibiting any visible categorical or associational membership characteristics, such as partisan identification. It is not the purpose of this article to explore the functions of reference groups. Rather, what is important is that each of these variations on reference group theory predicts that people will come to think of social relations in terms of in-groups and out-groups and that they will be more positive toward those groups with which they have a lot in common and more negative toward those groups with which they have little in common. Also see note 1.

Social Groups and Political Judgments 629 share concerns and those with which they have little in common. People are generally more positive in their evaluations of groups with which they share concerns and relatively more negative toward those with which they have less in common. Perhaps more important, groups are perceived to be connected, with varying degrees of intensity, to different political parties. These perceived connections reflect historical alignments between groups and parties and the more recent behavior of groups and political party leaders and candidates themselves (see Miller and Wlezien, 1993). In theory, if people like certain groups and they perceive those groups as aligned with a particular candidate or party, they should evaluate the politician and party more positively. Similarly, people who dislike the groups should evaluate the particular party and candidate more negatively. Within this theory, issue or policy positions are also (at least partially) a reflection of group orientations. Change in partisan strength and alignment, according to the theory, may occur in three ways. First, the groups connected with each party could change, either through the rise of new groups or because of shifts in the salience of already-existing groups. Recall that such shifts in salience may reflect the behavior of groups or party leaders and candidates. Second, public evaluations of the groups connected with each party could change; assessments of some groups could become more positive as others became more negative. Finally, the number of people who identify with a group could change, thus influencing the party's fortunes despite stability in the perceived connection of the group to the parties or evaluations of the group. In one sense, the reference group model is not inconsistent with Carmines and Stimson's (1986, 1988) explanation of partisan change. Note, however, that a major limitation of their approach, as well as other work invoking "new racism" theory (see, for example, Kinder and Sears, 1981; McConahay, 1986; Sears, 1988; Edsall and Edsall, 1991), is the singular focus on race as the major social cleavage in American politics. This is not to deny the importance of that cleavage; rather, our argument is that the cleavage should be placed within a broader context, one that reflects attempts at empowerment by a wide range of groups, that is, more than just blacks. Within our theory, the particular social cleavages that are driving American politics at a given point in time are an empirical concern. Thus, we need to determine which social groups the public associates with political parties and the extent to which identification with these groups structures political attitudes and behavior. The Structure of Party-Group Connections Previous research demonstrates that the public does associate certain clusters of social groups with each political party, and that as the political

630 Social Science Quarterly context changes, the associations shift over time (Miller, Wlezien, and Hildreth, 1991; Miller and Wlezien, 1993). Factor analyses of the American National Election Studies (NES) group thermometers for 1972, 1984, and 1988 indicate that the structure of party-group connections changed over that period, particularly for the groups the public associated with the Democratic Party (see Miller and Wlezien, 1993). In the early 1970s, Democrats were defined in terms of the poor, blacks, unions, the middle class, Catholics, and liberals. By 1984, however, the set of groups the public associated with the Democratic Party had narrowed and included such "fringe" groups as black militants, the women's movement, people on welfare, gays, and lesbians. These activist groups displaced, at least in the public's perception, certain more moderate groups of the traditional Democratic coalition, such as Catholics and the middle class (notice that these developments are consistent with Polsby's [1983] analysis of Democratic Party reform). The groups that were perceived to be newly associated with the Democratic Party were less positively evaluated by the public than were the more traditional groups (on average, by 20 degrees on a 100-point rating scale), which may help explain the decline in identification with the Democratic Party since the mid-1970s. During the same period, a shift also occurred in public perceptions of the groups associated with the Republican Party. Unlike the Democrats, however, the change in the groups that the public associates with the Republican Party did not involve displacement of some groups by others. Rather, an increased connection developed between traditional Republican groups - big business, wealthy people, and conservatives - and "fringe" groups, on the Right, specifically anti-abortionists and evangelical groups active in politics. These Republican fringe groups were also less positively evaluated than were traditional Republican groups. Through 1988, however, the correlation between evaluations of Republican fringe groups and feelings about the Republican Party were substantially weaker than the connection between affect toward Democratic fringe groups and Democratic Party assessments (e.g., in 1988 the correlation between assessments of fringe groups associated with the parties and the parties themselves was.13 for the Republican Party compared with.34 for the Democratic Party). But, as will become evident shortly, the Republican fringe became more strongly associated with the Republican Party during the 1992 campaign, consistent with interpretations of that particular election (see, e.g., Pomper et al., 1993). There is a certain structure to the public's evaluations of social groups and political parties, to be sure (also see Zinni, Mattei, and Rhodebeck, 1994), and the structure has evolved over time. The groups that are now associated with the political parties do not merely represent traditional interests, such as labor and business, or even race; indeed, they primarily represent cultural differences. Still, the question remains: Do evaluations of

Social Groups and Political Judgments 631 these groups actually influence evaluations of parties and candidates? Moreover, does the influence represent something that is not otherwise captured by issue positions or the other factors that contribute to people's political judgments? Issue Attitudes, Group Evaluations, and Political Judgments Abramowitz's (1994) models of party identification and vote choice in 1988 represent a useful starting point for our analysis. Apart from demographic variables and parental party identification, his model of party identification contains various measures of issues positions, namely, social welfare, national security, lifestyle, race, the death penalty, and school prayer. His model of vote choice includes party identification, demographic variables, the battery of issue attitudes, and personal and national economic perceptions.3 Abramowitz does not include a measure of liberalconservative identification in his models - perhaps understandably, because his research addresses the relative influence of various issue attitudes, which, when taken together, may effectively constitute (or be summarized by) general ideological dispositions. Since we are interested in whether group evaluations affect political judgments independently of other factors, however, we do incorporate a standard, 7-point measure of liberalconservative identification in our models. After all, it may be that group evaluations reflect ideological orientations. Of course, as is implied by the reference group model, ideological orientations (and issue attitudes) may themselves be at least partially structured by assessments of social groups (see Brady and Sniderman, 1985). Whether this is the case is a subject for other research. The group evaluation measures utilized here represent the mean evaluations of those groups associated with the Democratic and Republican parties, as determined by Miller and Wlezien's (1993) earlier analysis. Note that the evaluations of the political parties as well as "liberals" and "conservatives" are not included in the measures: thus, Democratic group evaluations represent the average thermometer rating for unions, blacks, civil rights leaders, people on welfare, feminists, and gays and lesbians; Republican group evaluations represent the average thermometer rating for big business, evangelicals active in politics, and anti-abortionists.4 Recall 3 Our measures are identical to those used by Abramowitz, as he kindly supplied us with his computer recodes. His variable for the size of the local black population is not included in our model because it may at least partially determine and, thus, conceal the effects of racial attitudes. The variable for the South, which has a sizable effect on party identification in Abramowitz's analysis, also is not included in our model. Excluding the variable makes little difference in the estimated effects of the other variables (or the explanatory power of the model), because the effect of region is substantially reduced by the inclusion of other variables, especially liberal-conservative identification. We have combined arfect toward all these groups into an overall measure rather than keeping each group distinct because previous research (see Miller, Wlezien, and Hildreth,

632 Social Science Quarterly that we are interested in whether these measures are associated with assessments of the political parties and presidential candidates, controlling for issue attitudes, ideological identification, and other factors. Following Abramowitz (1994), the analysis addresses political judgments among whites.5 While Abramowitz models actual vote choice, however, we focus on candidate evaluations - the difference between the thermometer ratings for Bush and Dukakis. The results using the thermometers do not differ meaningfully from the results using vote choice, but they are more directly interpretable. The results from regression analyses of both party identification and candidate evaluations in 1988 are reported in Table 1. Not surprisingly, parental party identification and liberal-conservative orientation are the most powerful predictors of individuals' party identification (see the first estimated model in Table 1). Social welfare and national-security issue attitudes also are strong predictors, whereas, consistent with Abramowitz, racial attitudes are not.6 More important from our perspective, evaluations of groups associated with the political parties, particularly the Democratic Party, predict party identification as well. The sum of the absolute value of the standardized coefficients for the two "groups" variables - one indication of their net effect - suggests that the effect of group evaluations on party identification actually exceeds the effect of social welfare attitudes. This result is quite startling, given the set of controls incorporated in the model, especially parents' party identification and the respondent's ideological orientation. We do stop short of concluding that group evaluations actually "cause" party identification; after all, it may be that those evaluations summarize people's partisan dispositions, as implied by research on issue attitudes (see, e.g., Campbell, Converse, Miller, and Stokes, 1960; Markus and Converse, 1979; Page and Jones, 1979; but also see Erikson, 1982, for a different characterization). Still, it is quite clear from these results that people do associate the two political parties with particular social groups, and this association reflects more than issue positions or liberal-conservative identification.7 1991; Miller and Wlezien, 1993) indicates that individuals do not respond uniquely to each group; they respond to clusters of groups instead. This finding suggests that the actions of a particular group in the cluster can influence how one feels about the other groups in the cluster. In order to assess the effects of positivity bias in group ratings, we adjusted the measures by subtracting out the mean evaluation of the Democratic and Republican groups and included these new measures in separate analyses. Results of these analyses differ only slightly from those reported here. 5 Including blacks makes almost no difference in any of the analyses. 6 Note that, in contrast with Abramowitz s findings, the measure or lifestyle issues, which captures attitudes toward abortion, gay rights, and women's role in society, does not constitute a significant, independent determinant of party identification (or candidate evaluations). These interests still may matter for political judgments, but it may be that the effects are directly captured by assessments of the groups themselves. 7 Disentangling the endogeneity between party identification and group evaluations, which is not straightforward, is well beyond the scope of this analysis. It is worth noting that analysis

Social Groups and Political Judgments 633 TABLE 1 Predicting Party Identification and Candidate Evaluation, 1988 Party Identification Candidate Evaluation Independent Variable b B b B Parents' party.437**.352 - - Identification Party identification - - 10.697**.490 Ideology.399**.258 2.159*.067 Issues Welfare.103**.170 1.597**.152 Defense.068**.138 2.058**.190 Race -.023 -.048.780*.074 Lifestyle -.006 -.008.589.037 Death penalty -.027 -.017 1.253.028 School prayer.017.008 -.231 -.004 Group evaluations Democratic groups -.018** -.119 -.512** -.148 Republican groups.011*.087.134*.067 Demographics Age -.006 -.047 -.172 -.059 Education.022.056.148.005 Gender.042.009 3.064.033 Income.022.056.062.007 Union household.118*.089-2.179* -.077 Economic perceptions Personal finances - - -1.253 -.028 National economy - - -3.769* -.075 Adjusted R2.458.579 Source: American National Election Studies (NES). Note: Party identification is measured using the standard 7-point scale, where high values indicate strong Republican partisanship. Candidate evaluation is measured by subtracting the Dukakis thermometer rating from the Bush rating, so high values indicate pro-bush sentiment. Democratic group evaluations represent the average thermometer rating for blacks, civil rights leaders, feminists, gays and lesbians, people on welfare, and unions. Republican group evaluations represent the average thermometer rating for antiabortionists, big business, and evangelicals active in politics. N = 824, *p <.01 **p <.001 (two-tailed). Both Democratic and Republican group evaluations also affect candidate evaluations independently of party identification itself (see the second estimated model in Table 1). These findings are strong evidence for the reference group theory, that is, unless candidate evaluations somehow using the difference in the party thermometer ratings instead of party identification produces quite similar results (see Miller and Wlezien, 1993), though the estimated effects of group evaluations actually are more robust. This finding implies that the measure of party évalua-

634 Social Science Quarterly Group TABLE 2 Factor Analysis of Group Thermometers, 1992 Factor Gays and lesbians.80 Blacks.67 People on welfare.55.36 Feminists.52.54 Liberals.52.54 Unions -77 Democratic Party.76 Conservatives.77 Evangelicals -75 Big business -62 Republican Party -64 % of variance explained Source: NES. Note: Table entries are varimax rotated factor loadings that are greater than.3. independently determine group evaluations. The influence of Democratic group evaluations is particularly pronounced, consistent with analyses of previous elections (see Miller, Wlezien, and Hildreth, 1991). Social welfare and national security issue attitudes also powerfully predict candidate evaluations in 1988, more than do group evaluations, whereas racial attitudes have only modest influence. These findings support Abramowitz's (1994) analysis. Based on our analysis, however, candidate evaluations in 1988 reflected not only issue attitudes, but also social group evaluations. People's connections between social groups and political parties (and the effect of social group evaluations on political judgments) may change over time, as was the case between 1972 and 1984. A factor analysis of the available 1992 NES group thermometers presented in Table 2 reveals that the structure of party-group connections remained fairly stable between 1988 and 1992, with certain notable exceptions. First, the Democratic factor was less clearly defined by dertain elements of the Democratic fringe, especially "gays and lesbians," than was the case in 1988. Second, as we alluded to earlier, the Republican factor in 1992 was more readily defined by its own fringe, specifically, "evangelicals," than in 1988. 8 The effects of tions is more sensitive to short-run considerations. Understandably, the measure outperforms party identification in the analyses of candidate evaluations that follow, though the effects of the other variables remain essentially the same. 8 The NES altered the designation ot this group in 1W, using Christian fundamentalists in place of "evangelicals." The NES also dropped certain other groups from the battery of thermometers, specifically, "civil rights leaders" and "anti-abortionists."

Social Groups and Political Judgments 63 5 group evaluations on political judgments also were different in 1992, as the results in Table 3 show. Notice first that the results in Table 3 are structurally quite similar to those in Table 1: party identification and candidate evaluations reflect essentially the same factors in both years. Consistent with this general pattern, Democratic and Republican group evaluations predict both party identification and candidate evaluations in 1992. The effects on candidate TABLE 3 Predicting Party Identification and Candidate Evaluation, 1992 Party Identification Candidate Evaluation Independent Variable b B b B Parents' party identification.466**.398 - - Party identification - - 10.526**.497 Ideology.355**.245 2.943**.113 Issues Welfare.082**.157.791*.073 Defense.078**.090 1.897**.115 Race.013.041.343.049 Lifestyle -.023 -.026 -.810.043 Death penalty -.042 -.027.511.016 School prayer -.121 -.049 2.302.043 Group evaluations Democratic groups -.012** -.084 -.262** -.098 Republican groups.007*.062.337**.146 Demographics Age -.006 -.057 -.118 -.050 Education.107**.089.772.031 Gender.051.011 3.213.046 Income.012.040.165.026 Union household.096*.070-1.016 -.041 Economic perceptions Personal finances - - -1.056 -.027 National economy - - -7.440** -.156 Adjusted R2.478.577 Source: NES Note: Party identification is measured using the standard 7-point scale, where high values indicate strong Republican partisanship. Candidate evaluation is measured by subtracting the Clinton thermometer rating from the Bush rating, so high values indicate pro-bush sentiment. Democratic group evaluations represent the average thermometer rating for blacks, feminists, gays and lesbians, people on welfare, and unions. Republican group evaluations represent the average thermometer rating for big business and Christian fundamentalists. N = 1078 *p <.01 **p <.001 (two-tailed).

636 Social Science Quarterly evaluations are especially strong, though the specific effects of Democratic and Republican group evaluations differ in comparison with 1988. That is, the estimated effect of Democratic group evaluations is smaller and the effect of Republican group evaluations is much larger, about 2.5 times as much. Indeed, a comparison of standardized coefficients in 1992 suggests that the effect of Republican group evaluations on candidate evaluations exceeds the effects of all other variables except party identification and assessments of the national economy. These findings nicely parallel the shifts in party-group structure described earlier; as the party-group structure changed between 1988 and 1992, from emphasizing Democratic fringe groups to Republican fringe groups, the relative impact of group evaluations on candidate evaluations similarly changed. Issue attitudes, particularly those pertaining to national security, were still important for candidate evaluations in 1992, though social welfare attitudes played only a modest role. Racial attitudes, however, did not significantly influence candidate evaluations in 1992. When taken together with the analyses of party identification in both 1988 and 1992, and even the analysis of candidate evaluations in 1988, the finding implies that political judgments are not primarily driven by racial attitudes, confirming what Abramowitz concluded. Separate analyses that include evaluations of "blacks" independently of evaluations of the other Democratic groups indicate that political judgments also are not driven by assessments of racial groups.9 Thus, it appears that racial attitudes, whether defined as issue positions or group evaluations, matter less than much previous research suggests, especially when considering individuals' very broad political judgments.10 Summary and Discussion Clearly, as Abramowitz (1994) shows, issues do have an important influence on individuals' party and candidate evaluations in the United States. 9 This analysis was conducted by excluding the thermometers for blacks and civil rights leaders from the Democratic group affect measure and then including this reduced indicator and the "blacks" thermometer in the models. Consider, for example, candidate evaluations in 1992. In the new equation, the impact of Democratic group evaluations was virtually the same as the measure that includes the race-related thermometer, while the blacks thermometer was insignificant (p =.14). The analysis was performed again using a measure that subtracted ratings of "blacks" from ratings of "whites." This measure also did not have a significant effect (p =.07). It may be that the effects of racial attitudes on political judgments are partially mediated by other variables in the models, but this also may be true for any other group associated with the political parties. Moreover, it is clear from our analyses that evaluations of the clusters of groups associated with the parties do independently predict both party identification and candidate evaluations. 10 This is not to say that racial attitudes do not play an important role in American politics. There are, after all, various "features" of party identification and presidential candidate evaluations (see, e.g., Wilcox and Wlezien, 1996) that may tend to dominate racial considerations. Where the number of features is smaller and more narrowly defined, as for judgments about candidates for state and, especially, local offices, racial attitudes may matter quite a lot, particularly when a black candidate is running against a white.

Social Groups and Political Judgments 6 37 Social welfare and national security issue attitudes are particularly important, while racial attitudes matter little. Issues do not tell the whole story, however. Groups also matter, perhaps even more than issues themselves. People tend to associate certain groups with the political parties, and their evaluations of these clusters of groups influence their political judgments independently of issue attitudes. It appears that people do use groups to organize their understanding of politics. Certainly, the structuring role of groups varies across individuals, just as the role of issues varies. That groups do play an important role for a number of people, however, is not surprising. Converse (1964) has shown that many more people organize their understanding of politics in terms of groups and group conflict than in terms of issues - indeed, three times as many. Clearly, our research supports this characterization, although it further reveals that people's cognitions of groups are not structured solely by membership or simple self-interest, but by the perceptions of a broad range of politically active groups. The relevant groups do not merely represent traditional interests, such as labor and business, or even race; they now primarily represent cultural differences. The emphasis on cultural groups is consistent with developments in the real world of politics during the last quarter century. Since the late-1960s, a number of social groups have sought their own unique form of empowerment. It is not that racial minorities alone have challenged for greater economic and political equality. Various groups on the Left and, more recently, the Right have demanded greater roles in the political process, to define the American way of life. Our analyses indicate that the public perceives this group competition. Indeed, it appears that evaluations of groups and, thus, judgments of the parties and candidates with which these groups are associated reflect a broad ideology of intergroup relations. REFERENCES Abramowitz, Alan I. 1994. "Issue Evolution Reconsidered: Racial Attitudes and Partisanship in the U.S. Electorate." American Journal of Political Science 38:1-24. Asher, Herbert B. 1976. Presidential Elections and American Politics. Chicago, 111.: Dorsey. Axelrod, Robert. 1970. "Where the Votes Come From: An Analysis of Electoral Coalitions, 1952-1968." American Political Science Review 60: 11-20. Bobo, Lawrence. 1988. "Group Conflict, Prejudice, and the Paradox of Contemporary Racial Attitudes." In P. A. Katz and D. A. Taylor, eds., Eliminating Racism: Profiles in Controversy. New York: Plenum. Brady, Henry E., and Paul M. Sniderman. 1985. "Attitude Attribution: A Group Basis for Political Reasoning." American Political Science Review 79: 1061-1078. Brewer, M. B. 1979. "In-Group Bias in the Minimal Intergroup Situation: A Cognitive Motivational Analysis." Psychological Bulletin 86: 307-324.

638 Social Science Quarterly Campbell, Angus, Phillip E. Converse, Warren E. Miller, Donald E. Stokes. 1960. The American Voter. New York: Wiley. Carmines, Edward G., and James A. Stimson. 1986. "On the Structure and Sequence of Issue Evolution." American Political Science Review 80: 902-921. N.J.: Princeton University Press. Cartwright, D., and A. Zander. 1968. Group Dynamics. New York: Harper and Row. Conover, Pamela J. 1984. "The Influence of Group Identification on Political Perception and Evaluation." Journal of Politics 46: 760-785. Politics Quarterly 33: 139-166. Science 18: 51-76. Conover, Pamela J., and Stanley Feldman. 1981. "The Origins and Meaning of Liberal/Conservative Self-Identifications." American Journal of Political Science 25: 617-645. Converse, Phillip E. 1964. "The Nature of Belief Systems in Mass Publics." In David Apter, ed., Ideology and Discontent. New York: Free Press. Deaux, Kay. 1993. "Reconstructing Social Identity." Personality and Social Psychology Bulletin 19: 4-12. Edsall, Thomas B., and Mary D. Edsall. 1991. "When the Official Subject Is Presidential Politics, Taxes, Welfare, Crime, Rights, or Values... the Real Subject is Race." Atlantic Monthly (May): 53-86. Erikson, Robert S. 1982. "The 'Uncorrelated Errors' Approach to the Problem of Causal Feedback." Journal of Politics 44: 863-881. Erikson, Robert S., Thomas B. Lancaster, and David W. Romero. 1988. "Group Components of the Presidential Vote, 1952-1984." Journal of Politics 51: 337-346. Fiske, Susan T., and Shelley E. Taylor. 1984. Social Cognition. New York: Random House. Green, Donald Philip. 1988. "On the Dimensionality of Public Sentiment toward Partisan and Ideological Groups." American Journal of Political Science 32: 758-780. Gurin, Patricia, Arthur H. Miller, and Gerald Gurin. 1980. "Stratum Identification and Consciousness." Social Psychology Quarterly 43: 30-47. Hammill, Ruth, Milton Lodge, and F. Blake. 1985. "The Breadth, Depth and Utility of Class, Partisan, and Ideological Schemata." American Journal of Political Science 29: 850-870. Jackman, Mary R., and Michael J. Muha. 1984. "Education and Intergroup Attitudes: Moral Enlightenment, Superficial Democratic Commitment, or Ideological Refinement? American Sociological Review 49: 751-769. Jackson, L. A., and L. Sullivan. 1987. "The Ingroup Favorability Bias in the Minimal Group Situation." Journal of Social Psychology 127: 461-472. Kinder, Donald R., and David O. Sears. 1981. "Prejudice and Politics: Symbolic Racism versus Racial Threats to the Good Life." Journal of Personality and Social Psychology 40: 414-431. Lau, Richard. 1983. "A Preliminary Report on Social Identification, Reference Groups, and Political Behavior: The Importance of Social, Political and Psychological Contexts." Paper presented at the Annual Meeting of the American Political Science Association, Chicago.

Social Groups and Political Judgments 6 39 Lau and David O. Sears, eds., Political Cognition. Hillsdale, N.J.: Erlbaum. chology Quarterly 52: 220-231. Lau, Richard, and David O. Sears. 1986. Political Cognition : The 19th Annual Carnegie Symposium on Cognition. Hillsdale, N.J.: Erlbaum. Lewin, K. 1951. Field Theory in Social Science. New York: Harper. McConahay, John B. 1986. "Modern Racism, Ambivalence, and the Modern Racism Scale." In John Dovidio and Samuel L. Gaertner, eds., Prejudice, Discrimination, and Racism. New York: Academic Press. McWilliams, Wilson C. 1989. "The Meaning of the Election." In Gerald M. Pomper et al., eds., The Election of 1988: Reports and Interpretations. Chatham, N.J.: Chatham House. Markus, Gregory, and Phillip Converse. 1979. "A Dynamic Simultaneous Equations Model of Electoral Choice." American Political Science Review 73: 1055-1070. Miller, Arthur H. 1987. "Public Opinion and Regional Political Realignment." In Peter F. Galderisi et al., eds., The Politics of Realignment. Boulder, Colo.: Westview. Miller, Arthur H., Patricia Gurin, Gerald Gurin, and Oksana Malanchuk. 1981. "Group Consciousness and Political Participation." American Journal of Political Science 25: 494-511. Miller, Arthur H., and Christopher Wlezien. 1993. "The Social Group Dynamics of Partisan Evaluations." Electoral Studies 12: 5-22. Miller, Arthur H., Christopher Wlezien, and Anne Hildreth. 1991. "A Reference Group Theory of Partisan Coalitions." Journal of Politics 53: 1134-1149. Page, Benjamin, and Calvin Jones. 1979. "Reciprocal Effects, Policy Preferences, Party Loyalties, and the Vote." American Political Science Review 73: 1071-1090. Parkin, F. 1971. Class Inequality and Political Order: Social Stratification in Capitalist and Communist Societies. New York: Praeger. Polsby, Nelson. 1983. Consequences of Party Reform. New York: Oxford University Press. Pomper, Gerald, et al., eds. 1989. The Election of 1988: Reports and Interpretations. Chatham, N.J.: Chatham House. House. Popkin, Samuel L. 1991. The Reasoning Voter: Communication and Persuasion in Presidential Campaigns. Chicago: University of Chicago Press. Sears, David O. 1988. "Symbolic Racism." In Phyllis A. Katz and Dalmas Taylor, eds., Towards the Elimination of Racism: Profiles in Controversy. New York: Plenum. Sears, David O., C. P. Hensler, and L. K. Speer. 1979. "Whites' Opposition to Busing: Selfinterest or Symbolic Politics?" American Political Science Review 73: 369-384. Sears, David O., Richard R. Lau, Tom R. Tyler, and H. M. Allen, Jr. 1980. "Self-Interest versus Symbolic Politics in Policy Attitudes and Presidential Voting." American Political Science Review 74: 670-684. Sidanius, James. 1993. "The Psychology of Group Conflict and the Dynamics of Oppression: A Social Dominance Perspective." In Shanto Iyengar and William McGuire, eds., Explorations in Political Psychology. Durham, N.C.: Duke University Press.

640 Social Science Quarterly Sniderman, Paul M., and Philip E. Tetlock. 1986. "Symbolic Racism: Problems of Political Motive Attribution." Journal of Social Issues 42: 129-150. Sniderman, Paul M., Thomas Piazza, Philip E. Tetlock, and Ann Kendrick. 1991. "The New Racism." American Journal of Political Science 35: 423-447. Stanley, Harold, William Bianco, and Richard Niemi. 1986. "Partisanship and Group Support over Time: A Multivariate Analysis." American Political Science Review 80: 969-976. Stone, Walter J. 1987. "Regional Variation in Partisan Change: Realignment in the Mountain West." In Peter F. Galderisi et al., eds., The Politics of Realignment. Boulder, Colo.: Westview. Tajfel, Henri. 1972. "Some Developments in European Social Psychology." European Journal of Social Psychology 2: 307-322. Introduction to Social Psychology. Narmondsworth: Penguin Press. Tajfel, Henri, C. Flament, M. G. Billig, and R. F. Bundy. 1971. "Social Categorization and Intergroup Behavior." European Journal of Social Psychology 1: 149-178. Tate, Katherine. 1993. From Protest to Politics. Cambridge, Mass.: Harvard University Press. Wilcox, Nathaniel, and Christopher Wlezien. 1996. "The Contamination of Responses to Survey Items: Economic Perceptions and Political Judgments." Political Analysis 5: 181-213. Zinni, Frank, Franco Mattei, and Laurie Rhodebeck. 1994. "The Structure and Dynamics of Group Politics, 1964-1992." Paper presented at the Annual Meeting of the American Political Science Association, New York.