Microsoft Word - Volume15.doc Commonwealth: A Journal of Political Science  Volume 15, Number 2  September 2009  2009 PPSA/LORL, PA House of Reps. ISSN 0890-2410 Viewing Representative Behavior through the Lens of Social and Political Identity Robin Lauermann Messiah College Much scholarly attention has been paid to the representative activities of members of the national legislature. Equally important to representative democracy, however, are the perceptions and evaluations of these members by their constituents. Using the 2000 American National Election Studies survey, this article assesses the impact of social and political identity in the formation of constituency perceptions of members of the U.S. House of Representatives. Demographic characteristics and political predispositions affect the relevance of the four measures of legislator responsiveness—policy, casework, district service, and symbolic responsiveness—to each constituent’s evaluation. Ethnicity, age, gender, and education affect the degree of importance of each measure to constituents; women, minorities, and older citizens are more concerned with symbolic actions on the part of their representatives. Furthermore, the descriptive composition of a representative’s constituency has an influence not only with respect to policy choices the member may make, but also with regard to understanding how particular behaviors are successful in establishing a relationship between the representative and those represented. Constituent Evaluations epresentative democracy provides a governmental framework wherein citizens elect officials who will make policy for the nation. Elections legitimate the actions of policymakers. Yet, officials are not given free reign once in office, for citizens will ultimately be able to assess their performance and hold them accountable for their actions at the next election. Elections are central to the representative process because they provide a means for placing new representatives in office and removing those who have not satisfied their constituents. While voting decisions are an important part of the representational R 66 VIEWING REPRESENTATIVE BEHAVIOR THROUGH THE LENS OF SOCIAL AND POLITICAL IDENTITY relationship, less noticed are the factors used to make them, particularly those that shape constituent evaluations of representatives. Past research has focused on what congressional members do in office (Fenno 1977; Fenno 1996; Fiorina 1982; Kingdon 1989; Mayhew 1974; Miller and Stokes 1963). Member behavior is crucial to the formation of constituent perception of the representatives’ behavior that will contribute to the voting decision. Individuals who do not believe that their interests are being represented by a governing official may become discontented with the officeholder and possibly with the political system itself. Citizens who think government is not responsive to their needs may become more likely to seek alternative help outside regular political channels. This shift in political action can be dangerous to the stability of the political system. Consequently, it is important to understand constituent expectations within the representative process because voters have the ultimate authority to revoke the power of the representatives. When individuals cast their vote, however, there is one additional consideration that affects their decision whether or not to extend the service of their member: their retrospective evaluations of the member’s behavior. Constituent evaluations therefore provide a crucial link between a member’s behavior and a constituent’s decision in the voting booth (Lauermann 2001). While the constituent perspective was relatively neglected by scholars in the past, it has received attention in more recent studies, especially in elections literature focusing on the “personal vote” and incumbency advantage (Cain, Ferejohn, and Fiorina 1987; Erikson and Wright 1993; Jacobson 1981; Jacobson 2001; Johannes and MacAdams 1981; Parker 1989; Serra and Cover 1992; Serra and Moon 1994; Yiannakis 1981). Many of these studies, however, center their analysis on the vote itself rather than on the factors affecting the evaluation of member behavior. Voting decisions involve factors relevant not only to the representative but also to the challenger and to election-specific contexts. Constituent evaluations evolve prior to the vote decision, so understanding them is a prerequisite of understanding the vote. More attention should be given to identifying the factors affecting these evaluations. Furthermore, many studies have focused on only a portion of the factors that affect constituents’ evaluations of their representatives. A representative may cast a vote on a piece of legislation, provide particular or general constituency service, or engage in other activities that may inspire the support of constituents. Researchers who fail to Robin Lauermann 67 address all relevant behaviors will achieve a less complete understanding of the representative process. Past research has demonstrated that constituent perceptions of member voting behavior, casework, district service, and symbolic actions contribute significantly to their general evaluation of member behavior. Among these and other personal and national factors, perceptions of symbolic connections are strongest (Lauermann 2001). The relative importance of each factor may vary, though, depending upon the characteristics of an individual constituent. This article will examine the way in which political and social identities shape the relevance of multi- faceted representative behaviors in the construction of constituent evaluations. Attention to multiple components of representation provides a much more accurate reflection of the “complexities of the real world” (Eulau and Karps 1978, 62). Components of Responsiveness A retrospective view of voter evaluation is based on the idea that elections serve as a means for constituents to reward or punish elected officials for their past behavior. V.O. Key (1966) laid the foundation for retrospective voting theory by noting that voters conduct an “appraisal” of past behavior of elected officials. Similarly, Fenno’s “Theory of Accountability” holds that “if members want to be reelected, they know that they will be held accountable at the next election and their behavior will be more responsive” (1978, 23). Fenno (1996, 78) reiterates in his later work on senatorial campaigns that the representative relationship is reciprocal because “candidates want support and they offer responsiveness; citizens want responsiveness and they offer support.” Thus the voting decision is made in light of an evaluation of member activities. Moreover, citizens care not about the means by which politicians make policy but about the results that come from policy (Fiorina 1981, 8; Wahlke 1978). Constituent perceptions of these factors require retrospective evaluations of the specific acts as well as the overall performance of a representative. Constituent evaluations stem from a number of sources, including evaluations of responsiveness, party identification, and national factors (Lauermann 2001). Although evaluations of responsiveness are relevant, the relative importance of different member behaviors may vary across 68 VIEWING REPRESENTATIVE BEHAVIOR THROUGH THE LENS OF SOCIAL AND POLITICAL IDENTITY individuals. Eulau and Karps (1978) note that many studies focus on a more limited understanding of member behavior as issue representation. They propose a four-part conception that includes policy, service, allocative responsiveness, and symbolic responsiveness. These components of representation can be distinguished by their degree of partisanship and tangibility (specific or concrete) in outcome. While Eulau and Karps offer a valuable conceptualization, attempts to operationalize this expanded perspective of representation have been limited (Lauermann 2001). Not surprisingly, many of these efforts focus on the voting behavior of representatives and the impact of their voting decisions on their electoral fortunes. Policy responsiveness represents the “presence of a meaningful connection between constituent (public) policy preferences or demands and the representative’s official behavior” (Eulau and Karps 1978, 63). At the individual level, policy responsiveness is indicated by closeness in issue positions.1 People who share their representative’s ideology are more likely to be satisfied with activities in this venue (Johannes and MacAdams 1981). While Congress is first and foremost a lawmaking body, policy is not the only relevant criterion for evaluation. There are several nonpartisan means of responsiveness, i.e., actions that can draw support from individuals who are not members of the representative’s party since these actions do not involve a firm policy commitment.2 One such method includes the “advantages which the representative is able to obtain for particular constituents” (Eulau and Karps 1978, 64). More commonly known as casework, this activity involves member actions to cut through bureaucratic red tape. Representatives may help an individual having difficulty obtaining benefits from an administrative agency, e.g., Social Security disability benefits or remuneration from the Black Lung program for former miners. Fiorina (1989) argues that members of Congress deliberately perpetuate the bureaucratic maze in order to create opportunities for themselves to build electoral credit. Constituents evaluate more favorably members who provide such services (Cover 1977; Rivers and Fiorina 1989; Serra 1991; Serra and Cover 1992; Serra and Moon 1994). These activities draw support from constituents of all partisan stripes since they are not linked to issues. A second nonpartisan means of responsiveness, allocative, involves “advantages and benefits presumably accruing to a representative’s Robin Lauermann 69 district as a whole” (Eulau and Karps 1978, 65). Known also as district assistance or “pork barrel,” allocative responsiveness includes a broader category of tangible awards than does casework. Legislators can bring public works projects or other federal programs to the district. Witness the budget appropriations process in 2004 when members of the Transportation and Infrastructure Committee generated $12 billion in district funding. A Democrat, Tim Holden, represents Pennsylvania’s primarily Republican 17th District. Yet, he managed to win reelection in 2002 by a 51% to 49% margin in a district reapportioned to benefit his Republican opponent, incumbent George Gekas. Holden was reelected even more handily in 2004 (59% to 39%) against challenger Scott Paterno, which may not have been surprising but for the challenger’s name recognition as the son of the celebrated Penn State football coach, Joe Paterno. While some portion of Holden’s win may be explained by his relatively more conservative voting record as a Blue Dog Democrat, he was certainly aided by his success in obtaining funds for the district, as he invariably advertises in his frequent newsletters to constituents. Between 2004 and 2009, for example, the 17th District received over $15 million in federal funding under the Safe, Accountable, Flexible, and Efficient Transportation Equity Act. Constituent perceptions of allocative responsiveness can therefore be a boon to representatives at the voting booth (Cain, Ferejohn, and Fiorina 1987; Jacobson 2001; Stein and Bickers 1994). Accordingly, they are a crucial component of any model of constituent evaluations. Lastly, constituents may evaluate their representatives based on a more nebulous but no less important criterion: symbolic responsiveness. Symbolic responsiveness involves a relationship “built on trust and confidence expressed in the support that the represented gives to the representative and to which he responds by symbolic, significant gestures in order to, in turn, generate trust and support” (Eulau and Karps 1978, 66). These gestures appear on the surface to be important but often have little substance, e.g., appearing at a ribbon-cutting ceremony. This concept is akin to Easton’s (1965) diffuse support of “general or positive sentiments” that allows representatives to weather periods of nonresponsiveness in other areas. In his seminal study of the impact of trust-generating activities, Fenno (1978) notes that constituency service, in the form of “home style,” can provide a representative with greater freedom in the policymaking realm in Washington. One of the more significant aspects of home style is “presentation of self.” Building trust 70 VIEWING REPRESENTATIVE BEHAVIOR THROUGH THE LENS OF SOCIAL AND POLITICAL IDENTITY takes time, so how does a member do it? According to Fenno, members must show that they are qualified, experienced, competent, honest, at one with their district, and able to empathize with its people. In fact, evidence suggests that symbolic responsiveness is more important to voters than the other components of responsiveness (Bianco 1994; Sigelman, Sigelman and Walkrosz 1994; Wahlke 1978). Yet, few studies have incorporated this component within a more comprehensive framework (Lauermann 2001). While each of these components of responsiveness has some level of importance, individual preferences and evaluations are likely to be shaped by personal experiences and factors. The social and political composition of a constituency may influence representative behavior as well. Constituent Identity In the field of social psychology, social identity theory conveys the relevance of one’s social identity in the construction of opinions. As Robert Lane (1974, 110) has framed the discussion: “Identity . . . is the complex answer to the simple question ‘Who am I?’” The social categories that construct one’s environment, such as ethnicity and gender, provide a social identity that shapes resulting opinions (Hogg, Terry, and White 1995, 260; Kinder and Sears 1985, 672). Using factor analysis, Hooper (1976) demonstrated that such reference groups as religion, ethnicity, sex, and class provide reliable indicators of social identity. The reference group attachment can create stereotypical perceptions among those with a shared group identity. Social differences can thus produce distinctions in opinion (Kinder and Sears 1985; Lane 1974; Stets and Burke 2000; Turner 1987). Although these distinctions may be based on the self-interest of the group, they may also develop from shared life experiences (Erikson and Tedin 2005, 177). Moreover, the interactive effect of multiple layers of identity produces great variations across individuals. While the idea that political and social identities can affect perceptions of the citizen–government relationship is not new, past studies have focused on attitudes toward the government at large rather than on the particular officials who compose it (Erikson and Tedin 2005; Lane 1974). Among those identity characteristics unchangeable by individual action is ethnic identity. The most significant distinction between the opinions of whites and minorities concerns the appropriate role of Robin Lauermann 71 government in assisting minorities, whether that includes monetary assistance or affirmative action (Erikson and Tedin 2005, 188). This distinction holds even when controlling for socioeconomic differences, which often are mistaken for ethnic differences. African Americans have shown greater concern for district and symbolic (particularly the identification component) factors (Griffin and Flavin 2007; Tate 2003). Minorities may consequently value activities by their member that are designed to help people both individually and as a constituency (Cain, Ferejohn, and Fiorina 1987, 42). Ethnic identity therefore is a significant factor in understanding constituent evaluations of member behavior. A constituent’s age may also affect member evaluations. Younger Americans tend to be less politically involved because they do not perceive most political activity to be relevant to their lives. Youths are less likely to be aware of or concerned about policy matters, which parallels the relationship between age and general political knowledge and interest. Cain, Ferejohn, and Fiorina (1987, 43) found that younger individuals are more concerned with district attentiveness, whereas older individuals are more concerned with casework and policy.3 These results are consistent with findings related to the impact of age on more general political opinions. While the young may not need public assistance themselves, they tend to be more supportive of government programs to help people (Erikson and Tedin 2005, 192–194). By contrast, certainly some policy focus on the part of older Americans is self-interested, such as their concern about proposed changes to Social Security and Medicare. Another important social reference group is gender. Although much has been written about the “gender gap,” the relevance of gender is perhaps less clear with regard to perceptions of representative behavior. Much research has focused on the impact of gender, specifically feminist and nonfeminist perspectives, on policy attitudes (Conover 1988; Kaufmann and Petrocik 1999) or on evaluations of male and female candidates (Dolan 2001; Huddy and Terkildsen 1993) rather than on how gender may shape the expectations of elected officials. Gender differences tend to be greatest with respect to compassion issues, such as increasing the quality and availability of education or protecting the less fortunate (Erikson and Tedin 2005, 209). Cain, Ferejohn, and Fiorina (1987, 41) found that men tend to prefer oversight and policy behavior to constituency-related activities. Support of constituency-related actions certainly is consistent with compassion concerns; yet, policy-based 72 VIEWING REPRESENTATIVE BEHAVIOR THROUGH THE LENS OF SOCIAL AND POLITICAL IDENTITY behavior may offer the most systematic means of eradicating problems in these areas. Among the more relevant reference groups, socioeconomic status has a variable impact on individual perceptions of representatives. Economic cleavages have not been as distinct in America as they have been in other industrialized nations, yet there have been significant differences in preferences on economic issues across social strata (Erikson and Tedin 2005, 178–179; Glynn et al. 1999, 227–229; Kinder and Sears 1985, 673). Individuals from lower classes are more likely to favor programs benefiting poorer people. Distinctions are not as large as expected once ideology is controlled for, as individuals of lower socioeconomic status tend to be less liberal or tolerant on noneconomic issues and thus are perhaps more likely to favor benefits to self and district. Past research has shown that lower class individuals tend to favor constituency service because they are in greater need of the benefits (Cain, Ferejohn, and Fiorina 1987; Yiannakis 1981). In fact, Cain, Ferejohn, and Fiorina (1987, 42) found that working class individuals ranked policy concerns as the third most important criterion for member behavior, whereas middle class individuals ranked it second. Ultimately, expectations of representative behavior will be influenced by this group identity. Education may also affect constituents’ views of what is important. Although often considered a related characteristic of social class, education has its own distinctive impact on the formation of opinions. High levels of education generally provide people with a greater understanding of the political system. It also tends to produce people more likely to be attentive to issues and the extent to which the representative is in accordance with their views. In an early study, Cain, Ferejohn, and Fiorina (1987, 42) found that 40% of college-educated individuals believed policy concerns were most important, as opposed to only 17% of those with less than a high school education. Because of their greater exposure to politics, those with higher levels of education are also more likely to be attentive to the other components of responsiveness. Finally, as a function of one’s political identity, ideology also plays a role in the formation of political opinions. Conservatives typically prefer a smaller central government and lower government expenditures. Yet, when using measures of partisanship rather than ideology, Cain, Ferejohn, and Fiorina (1987) found that Republicans (who may be more Robin Lauermann 73 or less conservative) supported greater levels of district spending but less individualized assistance. It is difficult to assess the past evidence for this component since party identification and ideology are not necessarily mutually reinforcing. Indeed, recent work has distinguished between “symbolic conservatives” and those who profess such an ideological leaning yet in reality are often “operationally liberal” because they support more government action and emphasize diffuse support due to pragmatic outcomes rather than programmatic ideals (Stimson 2004, 166–167). Ideological identity thus may have a less focused impact on constituents’ perceptions of their representatives. Despite the strong rationale for examining the impact of these and other factors on constituent evaluations, there has been minimal investigation. The relevance of identity on constituents’ evaluations of their representatives’ behavior (as distinct from their representatives’ votes) has not been sufficiently explored, nor has much effort been made to develop a more robust model of that behavior from the constituent perspective. Such work is needed because evidence of the impact of social and political identity on constituents’ perceptions is important for understanding not only individual opinions but also the implications of constituency composition for member behavior. By incorporating the oft- excluded symbolic concerns, member behavior would likely be encouraged in certain methods, depending upon the portion of individuals in the district possessing a particular identity or combination of identities. Method and Design This article used the 2000 American National Election Studies (ANES) data set to assess the impact of group identities on constituent perceptions (Burns et al. 2002).4 The ANES data includes 1,807 respondents selected through a stratified cluster sampling technique.5 In order to identify the importance of each facet of representation, this article uses an “identity model,” incorporating the group identities discussed above, to estimate the relevance to constituents of different components of responsiveness. Dependent Variables The dependent variables, representing each component of responsiveness, are derived from a similar measurement. Because there 74 VIEWING REPRESENTATIVE BEHAVIOR THROUGH THE LENS OF SOCIAL AND POLITICAL IDENTITY is no set of questions directly assessing the importance of the various components of representation, it was necessary to construct alternative measures to determine their relevance. In the survey, respondents were asked a series of open-ended questions about Democratic and Republican House candidates. Specifically, they were asked what they liked or disliked about the candidates, in this case, their own representatives.6 The constituents had an opportunity to mention up to five positive and five negative aspects of the candidate as reported in variables 1,329–1,351. Open-ended questions allow individuals to communicate their ideas freely without having to conform to pre-selected answers. They also provide an additional benefit in that respondents are able to discuss things important to them. The responses to the open-ended questions are coded with specificity in the ANES code book appendix. These responses were sorted and collapsed into one of four categories based on the fit of the coded open-ended answer with the responsiveness concept. Building on Fenno’s (1978) idea of “presentation of self,” symbolic responsiveness includes those responses that indicate trustworthiness (e.g., honesty, integrity, independence from other political actors, and not acting in self-interest), qualification for the job (i.e., experience), identification, and empathy. District service responses include mentions of actions on behalf of the district as a whole (e.g., keeping constituents informed and doing things to help the district’s economy).7 Service responses provided indications that the representative had helped individual constituents with problems. Specific policy responses were coded as policy, as were all general ideological responses.8 A complete listing of the values coded for each category is provided in the Appendix. An individual’s likelihood of mentioning a particular component of responsiveness provides an indicator of those facets of representation that are most important. While there are alternative ways of assessing the constituent perspective, the model used here is based on the subjective perceptions of the respondents. Although objective actions of House members may contribute to individual constituents’ perceptions of their representative’s behavior, it is ultimately the subjective perceptions that are relevant to constituent evaluations. If the perceptions themselves are skewed by bias or lack of information, the final evaluation itself will be skewed because it will be based more on subjective perceptions of the representative and less on the representative’s objective actions. While these objective actions are relevant to the larger framework of the representational relationship (Box-Steffensmeier et al. Tate 2003), this Robin Lauermann 75 article seeks only to determine what constituents value. It is not concerned with the accuracy of those perceptions. Identity Characteristics: Hypotheses and Measurement9 Ethnicity. Nonwhite individuals are more likely to prefer nonpolicy behavior, such as group or district benefits, because they have a greater need for tangible benefits. To understand how race affects the importance of each component, individuals were placed into one of two variable categories, white and nonwhite, based on the coding of responses to Variable 1,006.10 Age. Individuals at different life stages may have different expectations from their representatives as a result of different needs. Older individuals may be more concerned with policy factors because of their greater exposure to and understanding of the impact of government action on their lives. In accordance with Variable 908, individuals were grouped into one of three categories corresponding to the three age ranges during which political interest and involvement vary: 18–29, 30– 54, and 55 and older. Gender. Women are more likely than men to prefer member behaviors that are targeted at district and constituent assistance. Variable 1,029 was used to control for the gender of constituents. Social class. As a matter of self-interest, individuals from lower social classes are more likely to mention particular and general benefits rather than policy, whereas upper-class individuals may prefer policy considerations. This variable was measured by the self-selected social class response ranging from lower working class to upper middle class as reported by Variable 1,005. Education. Highly educated individuals are the most likely to mention policy concerns rather than other factors because they are best equipped to understand politics. A collapsed scale of education levels was created from Variable 913: no high school degree (individuals with 11 years or less of schooling); high school (individuals who had completed high school or only a few years of college without earning a degree); and college education (individuals with an associate, bachelor, or advanced degree). Ideology. Past research has examined the impact of party identification on assessments of member behavior. Ideology is a more 76 VIEWING REPRESENTATIVE BEHAVIOR THROUGH THE LENS OF SOCIAL AND POLITICAL IDENTITY appropriate measure than is partisanship, however, because it is a more consistent indicator of political opinions. Conservatives are more likely to prefer district benefits than casework. Ideology is measured with Variable 1,370, which provides a seven-point continuum from extremely liberal to extremely conservative. Results and Discussion Before examining the results of the multivariate analysis, it is instructive to review the distribution of responses to the like versus dislike mentions. Table 1 presents the distribution of responses according to the recoded categories. There is a significant disproportionality between the number of responses for likes and dislikes. Positive responses outnumbered negative responses by more than three to one. This response pattern is consistent with the overwhelmingly positive approval that constituents tend to accord their representatives (Parker and Davidson 1979). Across all respondents, individuals were most likely to mention a symbolic consideration as a reason for liking or disliking their representative. This behavior is consistent with past research finding symbolic responsiveness to be of primary importance to constituents (Lauermann 2001), and it confirms the necessity of representatives having to explain convincingly their Washington activity when they are home in their districts (Fenno 1978; Kingdon 1989). More interesting, however, is the relative level of mentions of issue- related considerations. While individuals were more likely to mention policy responses as the next most significant basis of evaluation regardless of the nature of approval, the relative distribution varied. Individuals were more likely to list policy as a consideration of disapproval (33%) than approval (25%). It thus appears that constituents are more likely to consider policy in their evaluation matrix when there is perceived dissonance between themselves and their members. Similarly, they are more likely to punish representatives for perceived policy disagreement than they are to reward those holding similar issue positions. Such behavior is consistent with past studies finding asymmetrical influence of blame relative to credit (Kernell 1977). Robin Lauermann 77 Table 1 Percentage of Like versus Dislike Mentions (Frequencies in Parentheses) Likes Dislikes Symbolic Responsiveness 52% (642) 50% (194) Allocative Responsiveness 12% (146) 3% (11) Service Responsiveness 4% (51) 1% (4) Policy Responsiveness 25% (315) 33% (130) Other 7% (88) 12% (46) Total 100% (1,242) 100% (385) As for tangible nonpartisan behaviors, district service is more relevant to constituents than is casework, perhaps because fewer individuals are likely to seek assistance, whereas pork-barrel benefits are extended to all people within a constituency regardless of their efforts. District service is more likely to be mentioned as a positive attribution than as a negative criticism. Despite public outcries against the allegedly irresponsible budgetary politics of pork, individuals seem willing to make exceptions for their own members. Likewise, few individuals mention service responsiveness as a drawback in evaluating their members. These actions are therefore overwhelmingly beneficial to members in building a relationship with constituents.11 How do social and political identities shape evaluations of members? To answer this question, a more complex analysis is required. Given the dichotomous nature of the dependent variable, the model including the identity characteristics was estimated for each of the four types of responsive behavior mentions using probit estimation techniques. Table 2 summarizes the model estimates.12 The coefficients provide mixed support for the hypotheses. With respect to ethnicity, African Americans were less likely to mention policy concerns. They focused instead on constituency service and symbolic considerations. Older Americans were most likely to address the way in which members presented themselves to the constituency. They were also more likely than younger individuals 78 VIEWING REPRESENTATIVE BEHAVIOR THROUGH THE LENS OF SOCIAL AND POLITICAL IDENTITY to mention both policy and service considerations. Better-educated individuals were also more likely to mention symbolic considerations. This finding is especially interesting given that college graduates are better equipped to process policy concerns. Women were much less likely than men to mention policy concerns, though given the magnitude of the coefficients, they were less likely to respond to all mentions. More conservative individuals were most likely to mention district service consideration yet less likely to cite issues of casework. Finally, those from a higher social class were more likely to raise issue concerns and least concerned with casework, perhaps because such people are less likely to perceive benefits. Still, class and ideology had only modest impact on the relevance of any of these considerations. Indeed, only the influence of social class on policy mentions was statistically significant. Table 2 Influence of Identity Characteristics On Responsiveness Relevance (probit estimates; n=1069 for all models)13 Symbolic Allocative Service Policy African American .214* .007 .549* -.097 Age .450*** .31*** .179 .199*** Female -.139* -.174* -.048 -.275*** Social Class .035 .038 .005 .071** Education .203*** .179* .194 .191*** Ideology .007 .028 -.045 .005 LR Chi2 76.92*** 25.14*** 11.05* 41.34*** ***p<.01, **p<.05, *p<.10 While the initial estimates provide a general assessment of the impact of group identities on constituent evaluations, a more specific estimate is needed since the relationships in the models are nonlinear. No two individuals are completely alike; each possesses a number of layers Robin Lauermann 79 of identities with potentially cross-cutting influences. To facilitate the interpretation of the probit coefficients, Stata’s SPOST commands were used to estimate probabilities of identity impact providing a sample interaction among group identity characteristics. Table 3 presents data showing the variation of two identities: ethnicity and gender. Among college-educated, middle-class moderates who are middle-aged, African Americans and white females are most likely to consider the symbolic way in which members relate to their constituents. White males, on the other hand, are more likely to weigh policy concerns as a portion of their evaluation. While district and casework considerations are much less likely to be noted, African Americans are more likely than whites to mention casework, while men are slightly more likely than women to mention district considerations. Given the similarity in rankings of probabilities across all categories, however, it is clear that both forms of constituency service are less important to all these individuals. Table 3 Probability of Responsiveness Mention among College-educated, Middle-class, Middle-aged Individuals of Moderate Ideology Symbolic Allocative Service Policy Black male .44 .13 .06 .33 Black female .38 .10 .05 .22 White male .36 .13 .01 .38 White female .29 .09 .01 .25 Since age and education produced some of the strongest coefficients of influence, predicted probabilities were also estimated to assess the relative impact of these reference groups on evaluations of member behavior. Among moderate middle-class whites, young males and females—regardless of education—were actually most likely to mention policy concerns, though college-educated men were much more likely 80 VIEWING REPRESENTATIVE BEHAVIOR THROUGH THE LENS OF SOCIAL AND POLITICAL IDENTITY (.31) to cite policy than were similarly educated women (.22). Middle- aged women at all levels of education were likely to address symbolic concerns and policy concerns equally, though middle-aged men were more likely to provide policy-related comments than symbolic ones. Older citizens, on the other hand, were most likely to mention symbolic concerns. Within this category, education provides only a greater likelihood of responding to representative behavior and not an influence on the relative importance of different responsive behaviors. While other layers of identity could be estimated, the purpose of this study is not to conduct an exhaustive generation of probabilities of potential interactions. Rather, the results can help to direct future analysis and theoretical development. Conclusions The findings of this study concerning the influences of identity on constituents’ perceptions of various member-responsive behaviors are consistent with past research. Given the complexity of assessing the influence of the multiplicity of identities that each individual may possess, further research and analysis is needed. Group identities surely affect individual political perceptions in the representational relationship. In particular, age, gender, education, and ethnicity exhibit significant yet varied effects on the relevance of member behavior for each constituent. These effects result from a combination of self-interest, political knowledge, concern for others, and life experiences. Future research should focus on expanded analysis of the effects of group identities on perceptions of representative behavior with attention devoted to refinement of measurement. Huddy (2001) asserts that while there may be value in viewing identity as a fluid factor determined by its contextual salience, there is also enduring evidence of larger-scale stability in the role that identity plays. Further attempts to examine the relevance of contextual salience for this topic are certainly merited. Most important, this study illustrates the need for research on the constituent perspective of representation to incorporate the nontangible component of symbolic responsiveness, as many individuals are likely to reference this type of activity, regardless of identity. This topic should also interest House members. Much of the literature on Congress focuses on member behavior with a presumption that it matters similarly to all constituents. This study reveals that Robin Lauermann 81 representatives of districts populated by differing social identities—e.g., urban areas with higher proportions of minorities, or districts with large numbers of older citizens—should take heed to know who comprises their districts so that they may best represent their constituents’ concerns. While the relative-ranked importance of each component, represented by the likelihood of mention, does not always vary across groups, these factors may make a difference in the outcome of close elections. Understanding the impact of social identities may help members to recognize when policy and other concerns are most or least relevant to voters, thereby strengthening the relationship between representatives and their constituents. The impact of social and political identities also demonstrates the importance of continued attention to concerns of descriptive representation, for just as identities shape constituent preferences, so too will members’ identities drive their own behaviors. Appendix Coding for Like versus Dislike Mentions Symbolic Mentions: 201, 203, 211–225, 297, 301–320, 334, 335, 397, 401–426 431, 432, 435–442, 447, 449–457, 459–462, 464, 502–505, 603, 604, 609, 610, 613–620, 622, 623, 625, 627, 701–711, 722, 732, 807, 808, 829–838, 841, 842, 845, 846, 1201–1206. Policy Mentions: 327, 328, 500, 501, 508–520, 531–536, 601, 602, 605– 608, 611, 612, 731, 801–828, 847–849, 897, 900–997, 1001–1047, 1101–1199. Service Mentions: 321, 322. Allocative Mentions: 323–326, 329–332. Group Mentions: 207–1236, 1297. Other or Not Applicable: 427–430, 433, 434, 443–445, 446, 448, 495– 498, 506, 507, 543, 551–556, 597, 697, 718–730, 796, 797, 843. 82 VIEWING REPRESENTATIVE BEHAVIOR THROUGH THE LENS OF SOCIAL AND POLITICAL IDENTITY Table A1 Pearson’s Correlations: Identity Characteristics and Responsiveness Mentions Age Education Female Ideology Ethnicity Social Class Age 1.00 Education -.125*** 1.00 Female .038 -.042 1.00 Ideology .096*** .011 -.064** 1.00 Ethnicity .134*** .142*** -.034 .134*** 1.00 Social Class .089*** .314*** -.011 .099*** .158*** 1.00 Symbolic Mention .029 -.014 -.009 -.012 .029 -.026 Allocative Mention -.033 .062 -.040 -.009 -.018 .030 Service Mention -.009 .057 .029 -.087** .009 .009 Policy Mention -.077 .059 .072 .002 -.098** -.061 ***p<.01, **p<.05, *p<.10 Notes 1While issue voting is problematic at best (see Campbell, Converse, Miller, and Stokes 1960; Hurley and Hill 1981; and Abramson, Aldrich, and Rohde 2001), it must be remembered that evaluations are calculated on the basis of subjective perceptions, accurate or not. Moreover, Bishin (2002) finds that while constituents may not fulfill a sophisticated level of issue voting, there still are constraints on representative behavior. 2Credit claiming is a principal electoral activity (Mayhew 1974). It is defined as “acting so as to generate a belief in a relevant political actor (or actors) that one is personally responsible for causing the government, or some unit thereof, to do something that the actor (or actors) consider desirable” (Mayhew 1974, 53). These actions may include particular benefits to the constituency or groups within the constituency or for the district as a whole. The idea is that if members Robin Lauermann 83 can convince their constituents that they were responsible for some positive government outputs, even in part, then that will work in their favor when they stand for reelection. Compared with policy matters, casework and project assistance provide relatively non-controversial and nonpartisan means for getting constituency approval. Furthermore, it is easier to take credit for smaller projects than for the passage of a bill (Fiorina, 1989, 44). 3Multivariate analysis is needed to assess the extent to which the policy differences are not confounded by educational characteristics. 4This is the most recent year for which all variables were available. 5Only one National Election Study survey provides a measure in which respondents are able to rank various member activities in terms of importance. While the 1978 NES has a question that allows respondents to rank behaviors, it does not correspond exactly to concepts here. The concepts included: “Helping people in the district who have personal problems with the government,” “Making sure the district gets its fair share of government money and projects,” “Keeping track of the way government agencies are carrying out laws passed by Congress,” “Keeping in touch with the people about what government is doing,” and “Working in Congress on bills concerning national policy.” 6The responses were not coded specifically with respect to the incumbent. Those respondents who were coded in the race type variable as “Republican incumbent running” were rated using the House Republican candidate questions and those with a Democratic incumbent running were rated using the House Democratic candidate questions. While this method did not permit the inclusion of those individuals whose incumbent was retiring, there was no measure that would also allow for a similar evaluation. Few cases lacked an incumbent race, so excluding them is not problematic. 7This category did not include the comments for whether the representative conveys the views of the district. These responses were considered to be more policy-oriented in nature and were coded accordingly. 8Individuals also mentioned the importance of the representative acting on behalf of or against specific groups. While this categorization does not expressly fit within the framework established, the significant number of mentions suggests that it is a worthwhile consideration to measure. This distinction differs from that of identification, which suggests explicitly that a member is connected in some social grouping to the constituent. Instead, it provides a measure of relevance of group connections irrespective of the characteristics of the constituent. Only a handful of responses were included in this category, however, so a separate model of estimation was not included in this study. The remaining responses were placed in a residual category because they typically bore no relation to any consideration of responsiveness, such as campaign- specific comments indicating support of candidates because they were the underdog. 9In this study, identity serves as a factor influencing the constituent’s opinion, not as a measure of descriptive congruence, as used for example by Box- 84 VIEWING REPRESENTATIVE BEHAVIOR THROUGH THE LENS OF SOCIAL AND POLITICAL IDENTITY Steffensmeier, Kimball, Meinke, and Tate (2003). Therefore, whether individuals who share important identity characteristics with their representatives have more favorable evaluations is an issue worth exploring, it is beyond the scope of this particular project. 10The decision to place both Hispanic and Asian American individuals in the non-white category was based not on the erroneous assumption that all minorities tend to respond with the same characteristics, but on the small pool of Hispanic and Asian Americans (due to missing values). Analyzing these two groups separately would not produce statistically meaningful results given the limited number of cases. 11While these results are instructive, it is important to place the actual responses within the larger potential framework. As the unit of analysis is the mention and not the respondent, the 1,600 cases are much fewer than if each respondent gave four mentions of like and dislike, which would result in 7,200 cases (based on the 1,807 person sample) for each set of evaluations. Even if all people mentioned only one item of like and dislike, it would produce over 3,600 units for analysis. 12Bivariate correlations presented in Table A1 reveal no significant multi- colinearity concerns. 13The limited number of statistically significant coefficients for the service model is likely due to a small number of responses for this category relative to the other models. References Abramson, Paul, John Aldrich, and David Rohde. 2001. Change and Continuity in the 2000 Elections. Washington, D.C.: CQ Press. Bishin, Benjamin. 2002. “What Constituents Don’t Know Won’t Hurt Them: Representation and Constituent Ignorance.” Presented at the annual meeting of the American Political Science Association, Boston. Bianco, William. 1994. Trust: Representatives and Constituents. Ann Arbor, MI: University of Michigan Press. Box-Steffensmeier, Janet, David Kimball, Scott Meinke, and Katherine Tate. 2003. “The Effects of Political Representations on the Electoral Advantages of House Incumbents.” Political Research Quarterly 56(3): 259–270. Burns, Nancy, Donald Kinder, Stephen Rosenstone, Virginia Sapiro, and the National Election Studies. 2002. AMERICAN NATIONAL ELECTION STUDY, 2000: PRE- AND POST-ELECTION SURVEY [Computer file]. 2nd ICPSR version. Ann Arbor, MI: University of Michigan, Center for Political Studies [producer], 2001. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor]. Cain, Bruce, John Ferejohn, and Morris Fiorina. 1987. The Personal Vote: Constituency Service and Electoral Independence. Cambridge, MA: Harvard University Press. Robin Lauermann 85 Campbell, Angus, Philip Converse, Warren Miller, and Donald Stokes. 1960. The American Voter. New York: John Wiley and Sons. Conover, Pamela Johnston. 1988. “Feminists and the Gender Gap.” Journal of Politics 50(1): 985–1,010. Cover, Albert. 1977. “One Good Term Deserves Another: The Advantage of Incumbency in Congressional Elections.” American Journal of Political Science 21(3): 523–539. Dolan, Kathleen. 2001. “Electoral Context, Issues, and Voting for Women in the 1990s.” Women & Politics 23(1): 21–36. Erikson, Robert, and Kent Tedin. 2005. American Public Opinion. 6th ed. Boston: Pearson Longman. Erikson, Robert, and Gerald Wright. 1993. “Voters, Candidates, and Issues in Congressional Elections.” In Congress Reconsidered, 5th ed., eds. Lawrence Dodd and Bruce Oppenheimer. Washington, D.C.: CQ Press, 91–114. Eulau, Heinz, and Philip Karps. 1978. “The Puzzle of Representation: Specifying Components of Responsiveness.” In The Politics of Representation, eds. Heinz Eulau and John Wahlke. Beverly Hills, CA: Sage, 55–69. Fenno, Richard Jr. 1978. Home Style: House Members in their Districts. Boston: Little Brown. Fenno, Richard Jr. 1996. Senators on the Campaign Trail: The Politics of Representation. Norman, OK: University of Oklahoma Press. Fiorina, Morris. 1981. Retrospective Voting in American National Elections. New Haven, CT: Yale University. Press. Fiorina, Morris. 1982. “Congressmen and their Constituencies: 1958 and 1978.” In Proceedings of the Thomas P. O'Neill. Jr. Symposium on the U.S. Congress, ed. Dennis Hale. Boston: Eusey Press, pp. 33–64. Fiorina, Morris. 1989. Congress: Keystone of the Washington Establishment. New Haven, CT: Yale University Press. Glynn, Carroll, Susan Herbst, Garrett O’Keefe, Robert Shapiro, and Mark Lindeman. 1999. Public Opinion. Boulder, CO: Westview Press. Griffin, Jack, and Patrick Flavin. 2007. “Racial Differences in Information, Expectations, and Accountability.” Journal of Politics 69(10): 220–236. Hogg, Michael, Deborah Terry, and Katherine White. 1995. “A Tale of Two Theories: A Critical Comparison of Identity Theory with Social Identity Theory.” Social Psychology Quarterly 58(4): 255–269. Huddy, Leonie. 2001. “From Social to Political Identity: A Critical Examination of Social Identity Theory.” Political Psychology 22(1): 127-156. Huddy, Leonie, and Nayda Terkildsen. 1993. “Gender Stereotypes and the Perception of Male and Female Candidates.” American Journal of Political Science 37(1): 119– 147. Hurley, Patricia, and Kim Hill. 1980. “The Prospects for Issue Voting in Contemporary Congressional Elections: An Assessment of Citizen Awareness and Representation.” American Politics Quarterly 8(4): 425–458. 86 VIEWING REPRESENTATIVE BEHAVIOR THROUGH THE LENS OF SOCIAL AND POLITICAL IDENTITY Jacobson, Gary. 1981. “Incumbents’ Advantages in the 1978 Congressional Elections.” Legislative Studies Quarterly 6: 183–200. Jacobson, Gary. 2001. The Politics of Congressional Elections. 5th ed. Boston: Longman. Johannes, John, and John MacAdams. 1981. “The Congressional Incumbency Effect: Is it Casework, Policy Compatibility, or Something Else? An Examination of the 1978 Election.” American Journal of Political Science 25(3): 512–541. Key, V.O. 1966. The Responsible Electorate: Rationality in Presidential Voting, 1936– 1960. Cambridge, MA: Belknap Press of Harvard University Press. Kaufmann, Karen, and John Petrocik. 1999. “The Changing Politics of American Men: Understanding the Sources of the Gender Gap.” American Journal of Political Science 43(3): 864–887. Kinder, Donald R., and David O. Sears. 1985. “Public Opinion and Political Action.” In The Handbook of Social Psychology, vol. 2, 3rd ed., eds. Garner Lindzey, and Elliot Aronson. New York: Random House, 659–714. Kingdon, John. 1989. Congressmen’s Voting Decisions. Ann Arbor, MI: University of Michigan Press. Lane, Robert E. 1974. “Patterns of Political Belief.” In The Handbook of Political Psychology, ed. Jeanne Knutson. San Francisco, CA: Jossey-Bass, 83–116. Lauermann, Robin. 2001. Evaluations of Responsiveness: The Constituent Perspective of Representation. Ph.D. diss. State University of New York at Buffalo. Mayhew, David. 1974. Congress: The Electoral Connection. New Haven, CT: Yale University Press. Miller, Warren, and Donald Stokes. 1963. “Constituency Influence in Congress,” The American Political Science Review 57(1): 45-56 Parker, Glenn. 1989. “The Role of Constituent Trust in Congressional Elections.” Public Opinion Quarterly 53(2): 175–196. Parker, Glenn, and Roger Davidson. 1979. “Why Do Americans Love their Congressmen So Much More Than Their Congress?” Legislative Studies Quarterly 4(1): 53–61. Rivers, David, and Morris Fiorina. 1989. “Constituency Service, Reputation, and the Incumbency Advantage.” In Home Style and Washington Work: Studies of Congressional Politics, eds. Morris Fiorina and David Rohde. Ann Arbor, MI: University of Michigan Press, 17–45. Serra, George. 1991. “What’s in it for Me? The Impact of Congressional Casework on Incumbent Evaluations.” American Politics Quarterly 22(4): 403–420. Serra, George, and Albert Cover. (1992). “The electoral consequences of perquisite use: The casework case.” Legislative Studies Quarterly 27: 233–246. Serra, George, and David Moon. 1994. “Casework, Issue Positions, and Voting in Congressional Elections.” Journal of Politics 56(1): 200–213. Sigelman, Lee, Carol Sigelman, and Barbara Walkosz. 1992. “The Public and the Paradox of Leadership: An Experimental Analysis.” American Political Science Review 36(2): 366–385. Robin Lauermann 87 Stein, Robert, and Kenneth Bickers. 1994. “Congressional Elections and the Pork Barrel.” Journal of Politics 56(2): 377–399. Stets, Jan, and Peter Burke. 2000. “Identity Theory and Social Identity Theory.” Social Psychology Quarterly 63(3): 224–237. Tate, Katherine. 2003. Black Faces in the Mirror: African Americans and Their Representatives in the U.S. Congress. Princeton, N.J.: Princeton University Press. Turner, John. 1987. Rediscovering the Social Group: A Self-Categorization Theory. Oxford: Blackwell. Wahlke, John. 1978. “Policy Demands and System Support: The Role of the Represented.” In The Politics of Representation, eds. Heinz Eulau and John Wahlke. Beverly Hills, CA: Sage, 73–90. Yiannakis, Diana. 1981. “The Grateful Electorate: Casework and Congressional Elections.” American Journal of Political Science 25(3): 568–580.