The Role of Identity-Based Perspectives in Making Sense of Politics
The Role of Identity-Based Perspectives in Making Sense of Politics
Abstract and Keywords
This chapter builds on existing research in political understanding, particularly studies of framing, to develop a model of group-level political interpretation. Prevailing studies of framing effects assume that members of the public are already familiar with the frames used by political professionals, but they do not examine the processes by which these perspectives are acquired. The chapter argues that the concept of identity-based perspectives helps us to understand and appreciate the bottom-up components of political interpretation.
Leading models of political understanding posit that people interpret politics through the frames used by elites. The concept of frames is ubiquitous because it acknowledges a simple yet powerful fact of communication: in a world with seemingly infinite amounts of information, all messages are packaged in ways that emphasize particular aspects of the issues.1 They are “interpretive packages” that give meaning to an issue because they suggest which information should be used to think about it (Gamson and Modigliani 1989, 3). The same basic concept has been called many things: “central organizing ideas” (Gamson and Modigliani 1989); “frames of reference” (Zaller 1992, 13; Lau 1986, 112; Katz and Lazarsfeld 1955, 79); “pictures in the head” (Lippmann 1947 ); “attention frames” (Lasswell and Kaplan 1950, 26–38); or “interpretive structures” for understanding (Kinder and Sanders 1996, 164).
It is widely acknowledged that the way elites frame an issue influences how it is understood.2 What is not recognized, however, is that ordinary folks interpret political information, including elite-driven frames, through perspectives that are shaped by their social identities. Elite-provided frames are successful in getting people to view political issues in certain ways because these frames resonate with the perspectives people have used to think and communicate about politics in their own lives.
Understanding is about Categorization
Models of understanding generally agree that interpretation is fundamentally about categorization.3 Simply put, to make sense of the world, people carve it up into manageable parts. The organization of these parts in memory is often referred to as schema-based cognition. A schema is a knowledge (p.19) structure or a framework of thoughts related to a given topic.4 It consists of a category label, and attributes, examples, and affect related to that category, all organized in a hierarchical fashion. Schemas operate like folders in a file drawer—those that are available in memory serve as the categories under which new information is stored. When thinking about a topic, people are more likely to retrieve information from schemata that are clearly relevant to that topic. To continue the file drawer analogy, the more people think about a particular domain, the better organized their filing system will become. That is, they have more folders, more hierarchical organization of these folders, more associations among these folders, and a greater likelihood that they will notice inconsistencies within and across these folders.5
When people try to make sense of an event, they type it as an instance of a given category, thereby calling up information that has been stored in the schema relevant to that category.6 A key influence on interpretation occurs when the schemas that have been called up provide information that fills in the blanks for information not provided by the message or survey question. For example, in a mid-1970s study of attitudes toward presidential candidates Jimmy Carter and Gerald Ford, people who had well-developed schemas about specific issues, groups, parties, or personalities were more likely to use information related to these knowledge structures when asked about their attitudes toward Carter or Ford (Lau 1986). Likewise, when people are provided with information about the party affiliation of hypothetical candidates, they infer information about the candidates' issue stances, even when that specific policy information has not been provided (Rahn 1993).
Are Categories Elite-Driven?
Although both the dominant work on framing and the model advanced in this book build upon this general theory of information processing, they emphasize two different answers to the question of where these categories come from. There is general agreement that people pick up the schemata they use for making evaluations from their information environments. Richard Lau explains, “People are not born with any particular political schema, however; these schemata develop through experience with the political world. If most political information involves party, issues, groups, or individual candidate personalities (and it is my impression that it does), then individual cognitive structures must mirror this information environment” (1986, 114). However, prevailing models of opinion assume that these information environments are a function of elite discourse. “They [ordinary citizens] have little control (p.20) over this environment; it is set by politicians, world events, and the media” (Lau 1986, 96).
Indeed, scholars have been able to explain the main shifts in American mass opinion on the basis of elite rhetoric. For this reason, the origins of the categories people use to interpret politics are typically attributed to elites and the mass media.7 This elite-driven view is most famously outlined in Zaller's The Nature and Origins of Mass Opinion (1992), as Lee (2002, chap. 1) asserts. Zaller posits that the way elites talk about issues affects how members of the public both evaluate and understand political topics. Although Zaller does not investigate understanding directly, his model suggests that elites influence individuals' interpretations of politics in the following manner. Elites provide suggestions, or “cueing messages,” about how a given issue connects with individuals' prior predispositions. People are expected to resist messages that run counter to their priors or “predispositions,” but only as a function of these cueing messages (44). Under Zaller's model, these messages or frames can be derived from interpersonal conversation as well as elite discourse.8 These messages suggest to the public how information fits with their prior preferences (in his model, political ideology serves as a proxy for preferences.) In other words, these frames are expected to operate like interpretive packages. Frames focus attention on chosen portions of an issue, which calls up particular schemata and therefore affects which information gets processed (Entmann 1989).
Like many other contemporary public opinion scholars, Zaller argues that the way political information is organized in individuals' minds depends on the frames provided by elite rhetoric. The people who are most “aware” of elite rhetoric are best equipped with these frames.9 In other words, this tradition posits that unless people are given a road map by elites, they are ill equipped to make sense of the world of politics.
The claim that citizens do not make sense of politics on their own, independent of the maps elites provide, is normatively unappealing, but it is hard to ignore the mass of empirical evidence in its favor (Zaller 1992, 45). Most prominently, there is Converse's evidence, published in 1964, which annihilated support for the belief that ordinary citizens think about politics on the basis of a coherent ideological belief system.10
Subsequent work has argued that people reason on the basis of core beliefs, rather than liberal-conservative ideology. Even this work, however, argues for a strong role for elites. The content of beliefs, such as support for equality of opportunity, economic individualism, and free enterprise, is assumed to trickle down from elites, transmitted directly and indirectly by (p.21) “the political rhetoric and politics of the society” and “maintained over time by the persistence of institutions and policies” (Feldman 1988, 418).
Scholars generally agree that forming political judgments on the basis of core beliefs happens because elites provide the road maps to make it happen. The tension between many beliefs in American political culture, such as that between individualism and egalitarianism, causes policy ambivalence (Feldman and Zaller 1992). The presumption is that elites provide the clarity that causes a particular value to win out.
Various studies demonstrate the difficulty people have with connecting their preexisting beliefs and views with political concerns in the absence of elite guidance. For example, in a study of individuals' interpretations of four policy issues in the Pittsburgh area, Lau et al. (1991) found that only when knowledge structures that individuals commonly and readily use clearly match an interpretation did their prior beliefs matter for evaluation. When this match does not occur, or when only one interpretation was made available, the individuals in their experimental settings made choices that did not correspond to their prior political beliefs.
More evidence of the relative inability of people to evaluate politics by using beliefs to aid interpretation comes from work by Shah and colleagues. They have shown that individuals' values influence their interpretations of media messages (Shah, Domke, and Wackman 1996, 1997; Domke, Shah, and Wackman 1998). They argue that values, ethics, and morals function like heuristics or cues and help people tie political issues to their sense of selves.11 However, their work suggests that these prior beliefs are more likely to matter for their interpretations of politics (and hence their political evaluations) when elite-driven messages point out the relevance of these beliefs. In an experimental test of the effects of different media frames on undergraduates' and also evangelical Christians' interpretations of candidates' stands on health care, they found that evangelical Christians were especially likely to interpret candidates' positions through ethical as opposed to materialist frames (Shah et al. 1996). In addition, when people were encouraged by the frame of the message to interpret health care stances in ethical terms, the individuals' own ethical beliefs had a larger influence on their choice of candidate (Domke et al. 1998).
The research by Shah and his colleagues underscores the undeniable importance of elite rhetoric. However, it also implies that assuming people are unable to think about politics in the absence of elite-provided guidance overlooks an important fact. Elite frames activate particular ways of understanding a political issue because ordinary citizens already have these frames (p.22) in their store of tools for making sense of the world. Frames are cultural resources. We teach these things to each other. Political actors are able to use frames in strategic ways because people give them meaning through the course of everyday interaction.
The success of frames in persuading people to think about an issue in a particular way depends on whether they resonate with perspectives that the audience can readily recognize. As Nelson explains, “[F]raming theory argues that social movement leaders and other political elites attempt to mobilize support by relating their claim or cause to a familiar socio-political frame of reference. Thus, opponents of affirmative action will frame that policy as ‘reverse discrimination’ in an attempt to link a purportedly benign policy to a despised social practice” (1999, 5, emphasis added). Likewise, the frames in which elites commonly convey social welfare information continue to prevail because they resonate with common themes in the political culture (Gamson and Lasch 1983). Frames work when they are familiar. And this familiarity is not acquired by individuals watching the news in isolation. Instead, it is transmitted in part through social interaction. People are immersed in social contexts in which they askeach other if they saw a certain story, wonder aloud about the implications of an event, and, more fundamentally, suggest to each other and reinforce certain ways of interpreting current events. Frames are ways of making sense of issues that people pass on to, and hash out with, one another.
This is an aspect of framing effects that researchers using a cognitive perspective only rarely acknowledge. Sniderman and Theriault (1999) argue that the term “frame” has evolved to suggest that interpretations of issues exist “out there” anddoes not acknowledge the source of these interpretations and the manner in which they are constructed.12 The result, they suggest, is that political scientists' research on framing effects is predisposed to support an elite-driven model.13
Of course, part of the familiarity of frames stems from exposure to the mass media and other sources of elite discourse. The mass media are responsible for the outlines of many categories that we use to think about politics and other topics. If they were not, why specifically would race be an important signifier even to people in homogeneously white American towns? Why would people's evaluations of politics be influenced by perceptions of the conditions and opinions of people they have never met (Mutz 1998)? Moreover, some of the content of these categories can be traced to mass media portrayals as well. There are notable central tendencies in whites' negative stereotypes of African Americans that could not have arisen spontaneously from a multitude of isolated “realities.”14 Finally, a top-down model can also partially (p.23) explain how these categories are connected to specific policies. Gilens (1999) argues that the tendency of the mass media to typify poor people as African Americans explains the relevance of race to attitudes toward welfare, and white Americans' opposition to it.
The Bottom-Up Component of Framing Effects
The evidence that elite rhetoric influences which frames people use to understand politics is undeniable. However, several studies suggest a dual top-down and bottom-up process. Neuman, Just, and Crigler (1992), with the aid of surveys, in-depth interviews, content analysis, and experiments, provide evidence that making sense of the news is not entirely a function of elite discourse. Specifically, they looked for the ways in which their in-depth interview respondents conceptualize and interpret public issues and compared these interpretations to those offered by the media. They found that both the mass media and their respondents tended to talk about the issues in one of five frames: “economic themes, divisions of protagonists into ‘us’ and ‘them,’ perceptions of control by powerful others, a sense of the human impact of issues, and the application of moral values” (62). They did find a large degree of overlap between the frames used by the media and the frames invoked by their respondents, and concluded, in accord with the elite-driven model, that the frames provided by the media “helped subjects to determine the personal relevance of the issue, to provide linkages among issues, and to formulate arguments from which opinion could be drawn” (62). However, they found their subjects interpreting the news in ways that suggest a much more active process than a simple top-down model implies.
They conclude that their respondents did not “slavishly follow the framing of issues presented in the mass media. Rather, people frame issues in a more visceral and moralistic (and sometimes racist and xenophobic) style. They actively filter, sort, and reorganize information in personally meaningful ways in the process of constructing an understanding of public issues” (77). This work suggests that although members of the public think through public affairs using the frames provided by the mass media, this is not sufficient evidence that their interpretation of issues is beholden to elites' interpretations. The fact that any given frame encapsulates a variety of stances, as Gamson and Modigliani (1989, 3) point out, implies that within these general narratives, individuals construct their own meanings and make sense of politics on their own terms.
A further challenge to an elite-driven model of understanding is that it cannot adequately explain why two people can interpret the same media (p.24) story in very different ways. Even when exposed to identical stimuli, how individuals think about an event is in large part determined by the way they view the world. Evidence of this is easy to come by. We pick it up in daily conversation when friends or family disagree with our “take” on a movie or our view of the implications of an election. More concrete evidence comes from Gross's work on evaluations of the 1992 Los Angeles riots. Through experiments, Gross (2000) researched judgments about the riots and found that differences in the explanations subjects gave for rioters' behavior could be traced to differences in their levels of racial prejudice.
These differences in interpretation are not a simple matter of differences in predispositions defined as preferences. Instead, the divergence is a function of the use of different categories and considerations in interpretation. To demonstrate, consider the systematic evidence that has been collected on variations in interpretations of the Clarence Thomas hearings (Sapiro and Soss 1999). In 1991, the U.S. Senate held contentious hearings debating the confirmation of Clarence Thomas to the U.S. Supreme Court. A former coworker, Anita Hill, had accused Thomas, the chairman of the Equal Employment Opportunity Commission, of sexually harassing her while serving as her supervisor. The hearings became a major news event. Sapiro and Soss found that members of the public interpreted the event in a variety of ways and that these interpretations varied systematically according to the characteristics of the person evaluating Thomas or Hill. Importantly, these interpretations differed in systematic ways by social category. Among Hill supporters, whites and blacks used different sets of considerations to think about her. This was not true among Thomas supporters. Instead, the most striking differences in their interpretations appeared along the lines of gender. Male Thomas supporters used a more complex structure of considerations than did female Thomas supporters. Strong Hill supporters also displayed gendered differences, as men and women in this group appeared to rely on different reasons for support.
This study suggests that members of different social groups interpret the same political event in different ways. But notice that these patterns are not deterministic.15 Neither all whites nor even all white women looked on this event in the same way. Differences in interpretations within social groups varied according to whether they leaned toward Hill or Thomas. How does this happen? Sapiro and Soss conclude that “meanings vary systematically across social dimensions defined by general faultlines in American politics, by the substance of events themselves, by the stories that journalists and other leaders tell, and by the degree to which people attend to these stories” (308). In their view, though certain social groups maybe more likely to favor (p.25) a particular response to an issue, their interpretations are filtered through elite-driven messages.
However, even without elite suggestions about how to interpret an event, interpretations often diverge along demographic, especially racial, lines.16 Take, for example, Kuklinski and Hurley's (1994) experiment on source cues and persuasion. In their experiment, they randomly assigned 152 blacks and 151 whites to a paper-and-pencil-administered questionnaire. Within the questionnaire, each subject was given the following item:
We would like to get your reaction to a statement that—recently made. He was quoted in the New York Times as saying that “African-Americans must stop making excuses and rely much more on themselves to get ahead in society.” Please indicate how much you agree or disagree with—'s statement.
The name of the speaker was changed across conditions: it was listed as either Jesse Jackson, Clarence Thomas, Ted Kennedy, or George H. W. Bush. They found that when the statement was attributed to Jackson or Thomas, the black respondents were more likely to agree. On a 5-point scale of agreement in which 5 is strong agreement, blacks given the names of Jackson and Thomas averaged scores of 4.11 and 3.79, respectively, while blacks given the names of Kennedy and Bush averaged scores of 3.32 and 2.97.17
The act of making sense of political information is conducted with the tool of social identity. It was the race of the speaker—not the speaker's ideology—that mattered for these interpretations. If it had been ideology, we would have expected approval of the statement to be similar in the Thomas and Bush conditions and the Jackson and Kennedy conditions (conservatives and liberals, respectively). But it was not; instead, responses were clumped by the race of the speaker. Moreover, it was not just the race of the speaker that mattered. The race of the speaker mattered as a function of the race of the respondent. The patterns among whites were discernibly different from those among blacks. Whites were less affected by the name of the speaker.18
Fifteen years earlier, Sapiro (1981–82) demonstrated corroborating results with respect to gender. She randomly assigned undergraduates to read a speech given by one of two hypothetical U.S. senators, John Baker or Joan Baker. The speech was identical, and dealt with poverty, unemployment, and economic growth. But Sapiro asked the subjects to rate the senator's competence on a wide range of issues. Through this simple manipulation, she demonstrated that people make sense of politicians with the aid of social group categories, but, more importantly, they use themselves as reference (p.26) points. Men and women judged John's competence on the environment very similarly, but they disagreed on Joan's competence: 15 percent of men, and 50 percent of women, thought she would be competent on the environment. Like Kuklinksi and Hurley's experiment, the gender of the candidate mattered most notably as a function of the gender of the respondent. The shape of the message—whether the senator is portrayed as a man or a woman—matters for interpretation, but so too does the interpreter's place in the social world.
Making Use of the Concept of Perspective
The concept of frame neatly encapsulates the fact that interpretation is done through categorization and that which categories a message invokes influence how it is understood. However, “frame” does not adequately acknowledge that even if two people are given the same message, they may interpret it in two different ways because they see it through different lenses.
It is the concept of perspective that allows us to account for the role of social experience in the act of political understanding. A variety of bodies of research suggest that the perspectives people use to make sense of public affairs are rooted in social identities developed through social interaction. Moreover, they suggest that differences in political interpretation across members of society arise from social interaction among people of specific social locations. By social location, I mean individuals' position in society with respect to characteristics that signify relative status, such as race, gender, and class.
Consider the feminist conception of “standpoint.” The sexual division of labor endows women with a different set of experiences than men, which have consequences for the different ways men and women look at the world (Hartsock 1998). Knowledge therefore depends on where one stands in the social world “because every knower is grounded in his or her own particular identities, including gender, race, and class” (Press and Cole 1999, x).
Work by Kristin Luker (1984) on abortion activists in California demonstrates that this grounding is the result of interaction within specific social contexts. Her work shows that members of pro-life and pro-choice groups understand the issue of abortion through strikingly different “world views.” It is not level of income, education, or religious affiliation that creates these values. Instead, it is the different social contexts that these demographics characterize that generate the activists' worldviews. Luker shows that the entire social context in which these activists live their lives reinforces their divergent interpretations. The women come from different socioeconomic backgrounds, enter into marriages and occupations with different expectations (p.27) and goals, and have established social networks that resonate with and reinforce their pro-choice or pro-life perspectives.
Similar mechanisms account for the distinctiveness of the lenses through which African Americans view the political world. Dawson (2001) argues that African-American ideologies are distinct from mainstream liberalism and include variations ranging from “activist egalitarians” to “black conservatism.”19 He states that African Americans' embrace of the idea of autonomy combined with physical, social, and political separation have caused these subpublic ideologies to develop. Thus, African Americans' perspectives differ from those of whites partly because of social structure: the combination of social location, spatial location (residing and working in segregated and often isolated geographic spaces), and exposure to black “counterpublics” or information networks.
Cultural studies of the mass media and society point out that the act of seeing the world from the standpoint of a particular position in society hinges not just on social structure but on active processes of social identification. Such studies show that people of certain social locations engage in “oppositional processing,” rejecting the dominant perspective and replacing it with their own frame of interpretation (Morley 1980; Liebes and Katz 1990). How this works depends on the viewers' identities. For example, Press and Cole (1999) argue that prime-time television shows about abortion tend to portray the topic in a mainstream, middle-class manner in which the typical abortion seeker is a poor, often nonwhite woman. They find that lower-income women who identify as “working class” rather than “middle class” readily argue that the shows do not represent the reality they have experienced in their own lives.
Additional evidence that variations in perspectives derive from experience in different social locations stems from Gamson's work on political talk. As noted earlier, he did discover a tendency to rely on collective identities, but, importantly, the tendency to do so differed across groups depending on their demographic makeup. Groups in which people had similar backgrounds were more likely to use their shared collective identities. This was especially the case for groups composed entirely of African Americans “in spite of” media discourse (108).20 For example, the media content he analyzed as part of the study did not put the issues of troubled industry and the Arab-Israeli conflict in adversarial terms, but the participants in his study occasionally did.
More recently, Lee (2002) has documented the social group basis of public opinion with respect to the civil rights movement. Whether or not group-based considerations mattered for individuals' stance on civil rights policies depended on whether people were part of the activated mass public (p.28) (African Americans, southern whites, and later on in the 1960s, northern white liberals).21 Thus, Lee suggests that opinion is a function of more than which elites individuals pay attention to. His work also shows that deciding which elites to identify with is an ongoing process. In his massive analysis of letters mailed to the presidents in office during the civil rights struggle, he found various cases of people admitting to a change in party affiliation because of the events of the movement.
Put differently, what causes people to pay attention to different elites? When faced with the possibility of subscribing to The Nation or the National Review, what explains which magazine a person buys? The choice is a function of individuals' self-concepts, which are larger than party identification. Moreover, evidence that two people interpret the same message in divergent ways suggests that individuals' perspectives also influence how the information presented in a given magazine is perceived.
Therefore, part of the explanation for the interpretive lenses people use does lie in the information environments in which they are steeped.22 But people are not blank pages upon which the news of the day is imprinted. They develop and clarify perspectives of understanding in an active, ongoing process in their daily life as social beings. Thus, the concept of perspective enables us to recognize that political understanding is not performed by millions of isolated individuals but by people who have ways of knowing and thinking that they acquire by living within society.
The Centrality of Social Identity in Perspectives
I submit that social identities occupy a central role in the perspectives people use to think about public affairs. They are not simply another consideration that people take into account when making sense of the world around them. Instead, identities color the lens through which other considerations and factors in opinion—things such as interests, attachments to political parties, and political values—are understood.23
The word “identity” can be used to convey either “individuality” or “sameness.”24 Individuals' identities accordingly comprise both the way they see themselves as individuals (personal identity) and as members of social groups (social identity) (Turner et al. 1987). It is the latter type of identity that is of interest here. In this study, I define social identities as knowledge structures that are developed when people categorize themselves and others as types of people (Tajfel 1969) and compare themselves to others for clues about how they should think and act (Brewer and Miller 1984).25 In more detail, to reassure themselves that their view of the world is a good one, people (p.29) compare themselves with others they think they are like (Festinger 1954). Through social interaction, people evaluate themselves in comparison with individuals, as well as with social norms of behavior (Pettigrew 1967) and reference groups (Kelley 1952). The process of face-to-face interaction with members of a small group is related to the process of identifying with a larger-scale social group.26 As people compare themselves with others in their immediate surroundings, they learn what they perceive to be the appropriate norms of behavior for “someone like me.” At times this identity can be restricted to the group (“the Old Timers”), but it can also take on large-scale connotations (“Ann Arborites,” “Americans”). We can see social identity at work when people think of themselves in terms of “we” and “us” as opposed to “I” and “me” (Turner et al. 1994). When social identities are relevant to a given situation, people respond in ways that are appropriate to that perception of themselves (Tajfel and Turner 1979; Turner et al. 1994). The effect of social identities on attitudes and actions is powerful. Experiments show that people strongly prefer their own group even when their own “group” is a construction of researchers that has only minimal consequence (Tajfel et al. 1971; Brewer 1979).27
A variety of work within political science has argued that social identities guide political thought, despite Lane's claims to the contrary. Whether people identify with women has been shown to influence how they justify their evaluations of candidates and their perceptions of the Democratic and Republican Parties (Conover 1984, 1988). Other work, which focuses on evaluation as opposed to understanding, also poses a role for social identity. The degree to which African Americans perceive a linked fate with other African Americans influences their political choices (Dawson 1994). The more people identify with a social group with which they believe they share interests, the more likely they are to think and behave in the political realm in ways that are distinct from non group members (Campbell et al. 1960). And this use of identities is not just the work of Lippmann's “feeble minds.” For example, among people identifying with Catholics during the 1950s, it was the more politically informed who were likely to support Catholic candidates (Converse and Campbell 1968).
Often, political scientists allude to the centrality of social identity but do not mention the concept explicitly. Take, for example, the following statement by Henry Brady and Paul Sniderman:
Why, then, are many in the mass public able to attribute accurately attitudes to Democrats and Republicans, to blacks and whites, even to liberals and conservatives? Liberals and conservatives (and Democrats (p.30) and Republicans) have, and emphasize, political identities, identities, moreover, that have been developed in contradistinction to one another (liberals vs. conservatives, Democrats vs. Republicans). Because they are competitors, there are incentives—certainly there is permission—for a person who likes liberals to dislike conservatives, and the other way around. What allows citizens to simplify political calculations efficiently is this two-sided, “us vs. them” character of politics; the more attached they are to their side—and the more opposed they are to the other—the more they appreciate the differences between the issue positions of the two sides. What counts, then, is not how people feel toward groups, one by one; rather it is how they feel toward pairs of opposing groups. (1985, 1075)
When the authors state, “What allows citizens to simplify political calculation efficiently is this two-sided, ‘us vs. them’ character of politics…What counts…is how [people] feel toward pairs of opposing groups,” they are pointing out the work of social identity.28
Stated another way, perceptions of groups enter political thinking as pieces of information that are rooted in citizens' self-concepts. Kinder's work on the racial basis of opinion provides evidence of this. His work argues that whites' attitudes toward African-American candidates (Sears and Kinder 1971) and race policies such as fair employment legislation, school desegregation, and affirmative action (Kinder and Sanders 1996)29 are rooted in their attitudes toward blacks acquired as whites in a racist society. Moreover, his work with Thomas Nelson (Nelson and Kinder 1996) demonstrates that individuals' attitudes toward policies depend on their attitudes toward the recipient group and on their position relative to that group. With experiments, they showed that when policy information was framed so that it clearly delimited the recipient group, attitudes toward that group mattered more than when such a frame was not used for the way subjects evaluated the policy.
But again, it is not simply the packaging of the message that matters. These effects depend on the race of the subject. For example, among whites who were exposed to information about spending on the poor that emphasized that such programs “give away money to people who don't really need the help,” their stances were more closely related to their attitudes toward the poor than they were when the information emphasized “given the huge budget deficit, we simply can't afford it.” However, among nonwhites, the relationship was the opposite: attitudes toward the poor were less important for policy stances when “people who don't really need the help” was emphasized. Nelson and Kinder conclude, “While this striking reversal defies easy (p.31) explanation, it does suggest that framing strategies that appeal to negative social stereotypes will not be universally effective” (1065). It also suggests that framing effects hinge on the audience members' own social identity—their attachments to in-groups and out-groups.
In a later project, Kinder and Winter (2001) tested this prediction. They modeled racial divides in opinion in two policy domains: issues directly related to race, such as affirmative action and equal employment opportunity, and social welfare issues, such as spending on education and government provision of health insurance. They tested three hypotheses for the origins of this divide: social class, social identity, and adherence to different political principles. They found different results in each domain. For racial issues, the divide in opinion could be explained by differences in political principles and social identity (measured by closeness to in-groups, feelings toward out-groups, and racial resentment). However, for social welfare issues, social identity had little if any affect. Instead, social class and political principles explained the differences.
But if political principles are exerting an effect, does this mean social identities are not? I argue that social identities do not operate separately from principles. Identities function as links between one's social location and one's view of the world. Indeed, Kinder and Winter (2001) leave open this possibility. They argue that blacks and whites adhere to different political principles (e.g., “the idea of limited government appears to be more crystallized and potent for whites than for blacks” [2001, 450]). The relationship between race and principle supports the possibility that social identities and political principles interact with each other.
The Kinder and Winter study did not conceptualize social identity as a central component of citizens' perspectives. Other studies have similarly treated social identity as one consideration, not as part of the interpretive framework people use that constrains other politically relevant considerations. For example, Mutz and Mondak (1997) examined whether—and how—perceptions of the well-being of various social groups (women, blacks, Hispanics, poor people, the well-to-do, working men and women, and the middle class) affected votes for Reagan in 1984. They tested three possible routes: group membership, group identity, and social comparison.30 They found little evidence supporting any of these mechanisms. Instead, they found perceptions of inequality were driving vote choice.
Mutz (1998) summarizes these results this way: “Overall, these results suggest that it is not the direction of change that individuals perceive in any given economic group that matters most; instead, the central issue is whether the groups are perceived to have fared the same, or with some benefiting or (p.32) suffering more than others” (139). But is group identity really not at work here? Consider the way Mutz and Mondak conceptualize the mechanism of social identity. Following Campbell et al. (1960), they theorize that groups operate as proxies for self-interest (1997, 286). In this framework, identity is assumed to influence which information is considered relevant to the vote, which then in turn influences that vote. However, if social identities are central parts of citizens' perspectives, there is reason to think the process is somewhat different. Group identity influences decisions about which information is relevant, but, more fundamentally, it influences how that information is perceived—which then influences the vote.
The evidence that Mutz and Mondak (1997) present with respect to perceptions of inequality is consistent with this view. When people judge the economic well-being of social groups in society, how do they make these calculations? Entirely on the basis of the mass media? Or do they root their evaluations in their sense of where they fit in the social world? Mutz and Mondak argue that social comparison works as if people make calculations of their personal economic status and the well-being of members of a social group and then use the distance between the two to inform their votes. But the theory of social judgments put forth by Brewer and Miller (1984) mentioned earlier holds that social identities are intertwined with the act of social comparison. That is, individuals categorize themselves as given types of people, develop identities and anti-identities, and use the resulting cognitive structures in making judgments (including, presumably, vote choices). A claim that social comparison is driving behavior is necessarily a claim that social identity is also at work.
* * *
Previous research generally conceives the process of understanding politics as an act of categorization. Prevailing models of public opinion posit that the categories or, relatedly, frames that people use to interpret public affairs are bestowed on them by elites, despite constructionist work that suggests that bottom-up processes involving social identities are at work as well. Reviving the concept of perspective allows us to acknowledge that people are influenced by the categories elites provide but that they are continually defining these categories through experience in their own social contexts.
Although elite-driven frames induce some categories to be more accessible than others, what these categories mean differs significantly across people in different social locations. Moreover, the categories that people apply are a function of their perspectives, which are rooted in experience within particular social locations. We can imagine, for example, that the way a white (p.33) man married to a woman in the Army Reserve makes sense of war with Iraq is different from the way an Arab-American restaurant owner in Dearborn, Michigan, interprets it. Knowing how the war is framed by the news media is not enough. To observe how it is that people interpret politics through identity-based perspectives, we need to study interaction within actual social contexts.
(2.) A wide variety of framing effects have been demonstrated. A focus on the horse race as opposed to the issue substance of a campaign influences what people recall from a given news story (Cappella and Jamieson 1997; Valentino, Beckmann, and Buhr 2001). A focus on the potential recipients of policy influences whether affect toward those recipients influences evaluation of the policy (Nelson and Kinder 1996). Variations in the framing of a wide variety of policies, such as affirmative action (Kinder and Sanders 1990), gay rights (Brewer 2000), and partial birth abortion (Freedman 1997), have been shown to affect evaluation. Also, researchers have demonstrated that implicit appeals to race influence candidate and policy evaluation (Mendelberg 2001).
Does framing influence opinion because it alters the accessibility of certain categories or because it changes the importance of various categories? In earlier formulations of framing by political scientists, effects were considered to result from variations in the accessibility of certain schemata or schematic elements across different packages of a message (e.g., Lau et al. 1991). Recently, however, Nelson and colleagues (Nelson and Oxley 1999) have argued that framing works by making certain considerations seem more important; it is not necessarily the case that framing works by changing the accessibility of considerations (Nelson, Clawson, and Oxley 1997). These theories may be addressing two different problems. Both deal with the relevance of information. However, the accessibility argument appears to relate to the cognitive fit of a given piece of information, while the importance argument pertains to the socially driven appropriateness of an item of information to the problem. I am grateful to Dhavan Shah for help in clarifying this point.
(3.) Kuklinski and Hurley (1996), in a study of the effect of source cues on political interpretation, cite Lakoff (1987): “There is nothing more basic than categorization to our thought, perception, action, and speech” (127). Nelson (2000), in a recent framing study, writes, “We have at our disposal an impressive range of cognitive tools for culling, storing, and using social and political information; one of the choicest is categorization: the assignment of novel objects to familiar classes” (5, emphasis in original). Likewise, Kinder and Berinsky (1998), in a study of interpretation, write that “we think that understanding is typically achieved through categorization” (3). Each of these scholars are building from psychological research that shows that the categories people use to process information influence how that information is understood (e.g., Medin and Coley 1998; Chi, Feltovich, and Glaser 1981; Hinsley, Hayes, and Simon 1978).
(4.) For an overview of schema-based information processing, see Lau and Sears (1986), Graber (1988), and also the Lodge and McGraw (1991), Conover and Feldman (1991), and Miller (1991) replies to the Kuklinski, Luskin, and Bolland (1991) critique of schema conceptualizations within political science.
(p.239) (5.) Tetlock's research program on the differences in styles of policy reasoning among elites supports the idea that motivation to think carefully about a policy domain results in complexity of thought. He finds that when elites are in a “low accountability” role, such as serving as a member of the opposition party (the party that does not make policy), they are more likely to exhibit reasoning that displays less complex connections among related considerations (Tetlock 1989).
(6.) This is why schema-based processing is called “theory-driven” rather than “data-driven” processing: storing and retrieving information is likely to be conducted consistent with the generic traits of the category rather than with respect to information specific to the particular instance.
(7.) For example, note the way Simon and Xenos (2000) begin a recent publication on framing effects and public deliberation: “Thanks to recent advances in public opinion research, we now know that the origins of public opinion—the sacred icon of democracy—lay in elite discourse. We also know that the public relies on the mass media for its political information” (363).
(8.) This is implied by his Reception Axiom (44).
(9.) As Lee (2002) argues, Zaller operationalizes awareness in a way that measures attention to elite discourse. Zaller conceptualizes awareness as attention to politics and understanding of political information (21), and measures it with knowledge of “neutral” factual information (333–45). Lee (2002) explains that “two of the three questions Zaller uses measure factual knowledge limited to the sphere of elite political actors and institutional politics. [These questions tap knowledge of the issue positions of elite actors, and the ability to recognize political actors on feeling thermometer items. The third indicator is the interviewers' global assessments of their respondents' political knowledge.] When Zaller controls for the transmission of political messages, he examines only mainstream media sources such as the New York Times, Newsweek, and Time. In this respect, although Zaller may be agnostic about where political information comes from, he is decidedly not agnostic about what constitutes political information” (28).
(10.) For the uninitiated, Converse's article displayed the following: He used survey interviews of a nationally representative sample of people to show that Americans are unfamiliar with the ideological labels “liberal” and “conservative.” Also, he showed that Americans were not reasoning on the basis of liberal-conservative ideology (the most commonly mentioned set of ideological labels), even if not verbalizing them. Typically, a respondent's policy stances were neither generally liberal nor generally conservative. He also showed that citizens' opinions are unstable over time (from 1956 to 1960), again suggesting a lack of any coherent basis of reasoning. “On the average, less than two-thirds of the public came down on the same side of a policy controversy over a two-year period, where one-half would be expected to do so by chance alone” (Kinder 1983, 393). Therefore, Converse concluded that ordinary American citizens do not reason about politics on the basis of coherent belief systems. To top off his argument, he showed that people who do have reason to think about politics, candidates for the House of Representatives, showed a markedly higher degree of correlation among their attitudes (according to the left-right ideological scale), as well as more stability over time.
(p.240) (11.) Relatedly, Sotirovic and McLeod (2001) estimate a structural equation model that suggests that values (postmaterialist vs. materialist [Inglehart 1990]) influence the use of information mediated by the mass media. They argue that postmaterialist values have a direct positive influence on newspaper use that in turn positively affects participation in diverse discussions. Conversely, materialist values have a positive effect on television entertainment use and a direct negative effect on exposure to diverse discussions.
(12.) Kuklinski, Luskin, and Bolland (1991) lodge this complaint against the concept of schema when they write, “A final limitation [in the use of schema theory] lies in the neglect of social context” (1346). The concept of schema does not prohibit recognition of the role of social context. However, researching political interpretation in a laboratory setting does inhibit our ability to investigate and theorize about socially rooted processes.
(13.) A further distinction in traditions underlying political science work on framing is the difference between two common points of departure, Gamson's sociological approach and Tversky and Kahneman's psychological approach. The latter's behavioral decision approach centers on prospect framing: an individual will pick a near-guaranteed option when a choice is framed in terms of gains, but will opt for a risky choice when the question is framed in terms of losses (Tversky and Kahneman 1988). Iyengar (1992, p. 163, n. 19) notes that the psychological and sociological conceptions are quite different. While Tversky and Kahneman approach framing as the act of altering choices by altering problem presentation, Gamson views framing as the act of affecting perceptions by stimulating the use of certain symbols and shared understandings to think about an issue. In addition, while Tversky and Kahneman focus on the mechanics of decision making, Gamson is attuned to the power associated with the ability to persuade populations of people to see an issue an a particular way. Lau et al. (1991) intentionally avoid the use of the term “frame” to “eliminate any confusion with the framing phenomenon documented in the behavioral decision theory literature [e.g., work by Tversky and Kahneman]” (645, n. 1). They distinguish framing effects in the behavioral literature by the process by which they cause preference reversals. They note, “In that literature, framing is the presentation of an identical set of consequences of a policy proposal in different ways. Typically, one frame presents the consequences in terms of gains and the other frame presents the consequences in terms of losses.” The goal of the work of Tversky and Kahneman is to study changes in evaluations that result from alterations in problem presentation. The interest of Lau et al. (1991), instead, is on “framing” as it is commonly understood among political scientists: persuasion of preferences through the “strategic presentations of consequences” or emphasizing some consequences over others, rather than creating different portrayals of the same consequences (645, n. 1).
(15.) Graber (1988) makes a similar claim when drawing conclusions from her study of the way a panel of in-depth interview respondents processed the news. She notes that the slight variations in the way people processed public affairs information depended on “needs created by life-style. Insofar as life-style coincides with (p.241) demographic categories, such as age, sex, and ethnicity, life-style differences take on the appearance of demographic differences” (252).
(17.) The average score among blacks who were not told the name of the speaker was 2.57.
(18.) Average agreement scores across the Jackson, Thomas, Kennedy, and Bush conditions were 3.37, 3.36, 3.29 and 3.15, respectively. The average score among whites who were not given the name of the speaker was 2.71.
(19.) In his definition, an ideology is “a world view readily found in the population, including sets of ideas and values that cohere, that are used publicly to justify political stances, and that shape and are shaped by society…. Cognitively, an ideology serves as a filter of what one sees and responds to in the social world” (4–5).
(20.) The relevance of race to the focus group discussions depended on the issue. For example, with respect to nuclear power, “more typically, the relevance of race was explicitly denied, and overall, there were very few differences between white and black groups in the framing of nuclear power” (Gamson 1992, 106).
(22.) Lau (1986) shows that there is considerable interpersonal variation by age in the schemata people employ to think about politics. People who were socialized in the New Deal era tend to evaluate presidential candidates on the basis of group-based schemata.
(23.) For example, in a study of political sophisticates' and professionals' interpretations of public opinion, Herbst (1998) argues that perceptions of interest are inextricably bound up with perceptions of collective identity. She argues that to understand someone's interpretation of a political event, it is essential to also understand their place in the world, including their roles and duties. Knowing how a given journalist or politician sees themselves in the world makes it easier to understand why they read political information as they do.
(24.) See, for example, definitions in Webster's New Universal Unabridged Dictionary, 2d Edition (1983): “1. the condition or fact of being the same in all qualities under consideration; sameness; oneness. 2. (a) the condition or fact of being some specific person or thing; individuality; (b) the condition of being the same as something or someone assumed, described, or claimed.”
(25.) Tajfel (1969) argued that social identities were developed to provide a positive self-image. However, Huddy (2001, 134–36) explains that subsequent research has made various claims about the motivation behind social identification. She explains that identification with groups appears to fill affiliative needs for members of low-status and high-status group members alike. Also, the motivation might be better described as the need for positive distinctiveness, as her own research on national identity among various subgroups within “Latinos” illustrates (Huddy and Virtanen 1995). Similarly, Brewer's (1991) theory of “optimal distinctiveness” (p.242) views the motivation as a need to identify in a way that optimally provides differentiation and affiliation.
(26.) The process of identifying with a small group and the process of identifying with a large-scale social group are typically considered distinct traditions within the study of group identification. Lau (1989) explains that the “social interdependence model” draws from a definition of groups as entities within which the members interact with one another, while the “social identification” model does not require interaction, but rather a shared perception of membership in a social group (220–21). However, the two theories merge because the social-comparison function served by interaction (the first tradition) feeds identification with larger-scale social groupings (the second tradition).
(27.) A “minimal group effect” in which people show preference for in-group members has been demonstrated using categorizations that were made meaningful by the researcher in experimental settings, such as delimiting group memberships according to whether the subjects preferred a Klee or a Kandinsky painting (Tajfel et al. 1971) or whether the subjects were art or science students (Oakes, Turner, and Haslam 1991).
(28.) One might argue that this is not identity but merely affect. However, Brady and Sniderman argue that affect alone does not drive this “likability heuristic” in which individuals evaluate political issues on the basis of their feelings toward social and political groups and their perception of attitudes among members of these groups. Instead, they theorize that the heuristic operates on the basis of affect rooted in a cognitive framework in which social groups are arrayed as in-groups and out-groups.
(30.) These analyses were conducted in the following manner. They estimated logit models in which, in addition to controls and the respondents' perceptions of the economic well-being of their families and the nation as a whole, they included perceptions of the economic well-being of social groups (women, blacks, Hispanics, poor people, the well-to-do, working men and women, and the middle class). To test whether membership in the groups caused perceptions of their economic well-being to matter, they investigated the significance of interactions between perceptions of each group and membership in it. To test identity, they investigated whether “statistical” interactions of perceptions and feelings of closeness to these groups (an indication of group identification) were statistically significant. To test whether respondents were engaging in social comparison, they included indicators derived from measures of the distance between the respondents' perceptions of their own economic well-being and that of each of the social groups.