i. , , 575...}. .1. Nh\.n...- w.— .L .2 41?“? Ah... ....; a in“? x 1.9 @fi% m .4. It . LN...“ “w: risk... . 2 ' LIBRARY :00! Michigan State University This is to certify that the dissertation entitled THE RECIPROCAL EFFECTS OF IDEOLOGY AND ISSUE ; ATTITUDES: CONSIDERING A DIRECTIONAL LINK FROM 3 ISSUES TO IDEOLOGY presented by JEREMY FRANKLIN DUFF has been accepted towards fulfillment of the requirements for the Ph.D degree in Political Science \\ §\\&\\\\R\ MM Major Profe or Signatur \\\\m\A 30025 ( Date MSU is an affinnative-action, equal-opportunity employer PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE jAN 3 O 2013 H13. 3* , LO LA .AN 5/08 K lProj/Acc&Pres/ClRC/DaleDue indd THE RECIPROCAL EFFECTS OF IDEOLOGY AND ISSUE ATTITUDES: CONSIDERING A DIRECTIONAL LINK FROM ISSUES TO IDEOLOGY By Jeremy Franklin Duff A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 2008 ABSTRACT THE RECIPROCAL EFFECTS OF IDEOLOGY AND ISSUE ATTITUDES: CONSIDERING A DIRECTIONAL LINK FROM ISSUES TO IDEOLOGY By Jeremy Franklin Duff In this dissertation, I argue that issues play an important role in determining many individuals’ ideological identification. A wealth of research has been written on how ideological identification helps individuals form opinions about political issues. I argue that for some people the causal relationship works in reverse, fueled by a focus on cues from political parties, which shape how they position themselves on issues and label themselves ideologically. The results of the analysis show that when taking into account a reciprocal relationship between self-reported ideology and positions taken on a set of political issues, the link from issue positions to ideology is consistently significant for a number of different model specifications and in a model looking at respondents in 2004, the link from ideology to issue positions disappears. The data suggest that individuals can use issue positions to structure their ideological orientations, and in some cases ideology does not help individuals structure their own attitudes. Copyright by JEREMY FRANKLIN. DUFF 2008 This dissertation is dedicated to all of those who supported me throughout this process, from my professors at Grayson County College, to those at Texas A&M, and those at Michigan State, as well as, my family and friends. All of you pointed me in this direction, and I finally made it. iv ACKNOWLEDGEMENTS I would like to think William G. .Iacoby for his patience, guidance, and advice during this project. Without his help and encouragement, this would not have been possible. He has been a better adviser than I could have ever imagined. I would also like to thank Saundra K. Schneider for her assistance throughout this process. Her guidance as the chair of the graduate department and as a committee member on this dissertation was invaluable. Additionally, I appreciate the assistance of Charles L. Ballard and Darren W. Davis, whose suggestions and assistance as committee members helped make this possible. I would also like to thank Frederick Wood, Daniel Lewis. and Gregory Robinson for listening to my complaints. sitting in on my presentations, asking questions, making suggestions, letting me bounce ideas offof them, and generally just being there when I needed to talk. Along those lines, I would like to thank the individuals who took the time to read parts of this paper and provide comments and suggestions based on the conference presentations I have given. Finally, and perhaps most importantly, I would like to thank my wife and my family for putting up with all of this the last several years. Their support and motivation was key in my seeing this through. TABLE OF CONTENTS LIST OF TABLES .......................................................................................................... viii LIST OF FIGURES .......................................................................................................... ix Chapter 1 ............................................................................................................................ 1 Introduction .................................................................................................................... 1 Chapter 2 .......................................................................................................................... 13 Ideology ....................................................................................................................... 13 The Origin of Issue Positions ...................................................................................... 15 Ideology as a Predictor of Issue Positions ................................................................... 20 Do Issues Affect Ideology? ......................................................................................... 26 Chapter 3 .......................................................................................................................... 32 The Model .................................................................................................................... 32 Specification, Data, and Methods ................................................................................ 35 Results .......................................................................................................................... 39 1984 Respondents ........................................................................................................ 43 1996 Respondents ........................................................................................................ 45 2000 Respondents ........................................................................................................ 46 Conclusions .................................................................................................................. 47 Chapter 4 .......................................................................................................................... 64 Differences in 2004 ...................................................................................................... 64 Specification, Data, and Methods ................................................................................ 67 Results .......................................................................................................................... 68 Conceptualization and Education ................................................................................ 71 Conclusions .................................................................................................................. 76 Chapter 5 .......................................................................................................................... 88 Same-sex Marriage ...................................................................................................... 88 Data, Specification, and Methods ............................................................................... 92 Results .......................................................................................................................... 93 Attitudes on the Rights of Homosexuals ..................................................................... 94 Discussion .................................................................................................................... 95 Jobs and Standard of Living ........................................................................................ 96 Defense Spending ........................................................................................................ 97 Conclusions .................................................................................................................. 98 Chapter 6 ........................................................................................................................ 107 Discussion and Conclusions ...................................................................................... 107 vi The Questions ............................................................................................................ 1 10 The Conclusions ........ . ................................................................................................ 1 14 Appendix A .................................................................................................................... 1 17 Appendix B .................................................................................................................... 135 Appendix C .................................................................................................................... 145 References ...................................................................................................................... 1 5 1 vii LIST OF TABLES Chapter 1 Table 1.1 — Perceived Meaning of Ideological Labels (1994) ............................... 12 Chapter 3 Table 3.1 — Summary of Issue Positions (1976) ............................................... 49 Table 3.2 — Descriptive Statistics (1972 and 1976) ........................................... 49 Table 3.3 — Correlations between Ideology and Issue Variables (1976) ................... 49 Table 3.4 — Summary of Issue Positions (1984) ............................................... 50 Table 3.5 — Descriptive Statistics (1984) ....................................................... 50 Table 3.6 — Correlations between Ideology and Issue Variables (1984) ................... 50 Table 3.7 — Summary of Issue Positions (1996) ............................................... 51 Table 3.8 — Descriptive Statistics (1996) ....................................................... 51 Table 3.9 — Correlations between Ideology and Issue Variables (1996) .................... 51 Table 3.10 — Summary of Issue Positions (2000) ............................................. 52 Table 3.11 — Descriptive Statistics (2000) ...................................................... 52 Table 3.12 — Correlations between Ideology and Issue Variables (2000) .................. 52 Chapter 4 Table 4.1 — Summary of Issue Positions ........................................................ 79 Table 4.2 — Descriptive Statistics ................................................................ 79 Table 4.3 —- Correlations between Ideology and Issue Variables ............................ 79 Table 4.4 -— Correlations between Ideology and Issue Variables for Different Groups..80 Chapter 5 Table 5.1 — Summary of Attitudes on Same-sex Marriage ................................. 100 viii LIST OF FIGURES Chapter 3 Figure 3.1 — Modeling the Relationship between Ideology and Issue Positions .......... 53 Figure 3.2 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (1976 Respondents) .................................. 54 Figure 3.3 — Modeling the Relationship between Ideology and Issue Positions (Lagged Variables and Party Placements as Instruments ............................................... 55 , Figure 3.4 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions, 1972-1976 Respondents (Lagged Variables and Party Placements as Instruments) ............................................................... 56 Figure 3.5 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions, 1972-1976 Respondents (Individuals Who Shifted Ideological Identification at least Two Positions) ............................................. 57 Figure 3.6 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions, 1972-1976 Respondents (Individuals Who Shifted Ideological Identification One Position or Less) .............................................. 58 Figure 3.7 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions, 1972-1976 Respondents (Individuals Whose Issue Index Changed at least .2 Points) ................................................................ 59 Figure 3.8 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions, 1972-1976 Respondents (Individuals Whose Issue Index Shifted less than .2 Points) ............................................................... 60 Figure 3.9 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (1984 Respondents) ................................. 61 Figure 3.10 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (1996 Respondents) ................................. 62 Figure 3.11 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (2000 Respondents) ................................ 63 Chapter 4 Figure 4.1 — Modeling the Relationship between Ideology and Issue Positions. . . . . ....81 ix Figure 4.2 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (All Respondents) ................................. 82 Figure 4.3 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (Issue-Focused Respondents) .................... 83 Figure 4.4 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (Ideology-Focused Respondents) ............... 84 Figure 4.5 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (Respondents with No High School Diploma) ......................................................................................... 85 Figure 4.6 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (Respondents with High School Diploma but No College) ........................................................................................... 86 Figure 4.7 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (Respondents with at least Some College Education)... . . . ............................................................................................................. 87 Chapter 5 Figure 5.1 — Modeling the Relationship between Ideology and Opinions on Same-sex Marriage ......................................................................................... 101 Figure 5.2 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Opinions on Same-sex Marriage .................................. 102 Figure 5.3 - Modeling the Relationship between Ideology and Opinions on Rights for Homosexuals .................................................................................... 103 Figure 5.4 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Opinions on Rights for Homosexuals ............................ 104 Figure 5.5 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Opinions on Govermnent’s Role in Providing Jobs and a Good Standard of Living ............................................................................................. 105 Figure 5.6 — Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Opinions on Defense Spending .................................... 106 Chapter 1 Introduction In this dissertation, I argue that issues play an important role in determining many individuals" ideological identification. A wealth of research has been written on how ideological identification helps individuals form opinions about political issues. I argue that for some people the causal relationship works in reverse, fueled by a focus on cues from political parties, which shape how people position themselves on issues and label themselves ideologically. To test this theory, 1 model a reciprocal relationship between ideological identification and issue attitudes. The results of the analysis are three-fold. First, analyses performed on data from 1972-2000 indicate that, at the very least, there exists a true reciprocal relationship between ideological self-placements and the positions individuals took on political issues. Second, data analyses on the 2004 American National Election Study show that when taking into. account a reciprocal relationship between self-reported ideology and issue positions, the link from issue positions to ideology is consistently significant for a number of different model specifications and the link from ideology to issue positions disappears. The data suggest that in 2004, individuals were using issue positions to structure their ideological orientations, and not the reverse. Finally, from analyzing opinions on the rights of homosexuals to marry, results indicate a reciprocal relationship between opinions on same-sex marriage and individual self-reported ideology. The main question driving this project is whether issue positions can impact how a person chooses an ideological label. If people define ideology as a set of issues, then perhaps it is the issue positions that are formed first, and then the terms “liberal” or “conservative” simply become labels to identify one’s set of issue attitudes. If so, then what is ideology? Is it a combination of issue attitudes, or is it an abstract belief about how the world should be, or some combination of both? Political science literature has varied in its definition of ideology. For instance, Converse (1964) defines ideology as a belief system that includes a wide range of opinions, high attitude consistency (constraint), and abstract conceptualizations (such as viewing the world in terms of liberalism or conservatism). Additionally, ideology has been defined as a symbolic orientation. In this sense, it is considered a standing predisposition, formed early in life, which affects issue attitudes through its relationship with the “manifest symbolic content” inherent in issue alternatives (Sears, Huddie, and Schaffer 1986). It has also been categorized as a group attachment, similar to party identification (Conover and F eldman 1981; Levitin and Miller 1977), where people develop attachments to certain groups and adopt the positions of those groups. In this case, the groups would be liberals or conservatives and the individual would conform his/her attitudes with the norms of either group. Finally, ideology has been considered a form of schematic information processing (Hamill and Lodge 1986; Hamill, Lodge, and Blake 1985; Sharp and Lodge 1985). According to this interpretation, “ideology provides individuals with a set of organized categories that are used to process and integrate incoming information from the external environment” (J acoby 1991, 180).1 ' Also see Hymes (1986). [O Regardless of how ideology has been defined, one point has remained constant. Ideology helps individuals take positions on current political issues.2 As a symbolic orientation, individuals are able to identify the symbolic content in issue alternatives that they have a positive reaction to. Ideology helps dictate whether they will have a positive or negative reaction to the issue stimuli. As a group attachment, individuals simply take the positions that their preferred groups take. If they have a positive attachment to liberals then they will typically conform their attitudes to fit with other liberals. As a schematic information processing technique, individuals use the terms “liberal” and “conservative” as reference points to sort information coming from the political world. If the issue stimulus is similar to the individual’s own position along the liberal- conservative continuum, it will evoke a positive attitude, and vice versa. (J acoby 1991 ). But, is this the case for most Americans? Is it true that ideology helps individuals forrn opinions on political issues? Or do issues help individuals label themselves ideologically? There is some evidence to suggest the latter. Table 1.1, taken from Erikson and Tedin (2005), displays the distribution of responses to the question: “What sort of things do you have in mind when you say someone’s political views are liberal/conservative?” The results show that issue positions influence respondents’ views on ideology. Thoughts about abortion, gay rights, welfare/social programs, and government spending characterize how individuals define ideology, perhaps even more so than the classical conceptions of acceptance or resistance to change, presence or 2 See Bennett and Bennett 1990; Carmines and Stimson 1989; Fleishman 1986; Gamson and Modigliani 1989; Jacoby I994, I991, I986; Kuklinski, Metlay and Kay 1982; Lau, Brown and Sears 1978; Luttbeg and Gant 1985; Sears 1993; Sears and Citrin 1985; Sears, Hensler and Speer 1979; Sears, Huddy and Schaffer 1986; Sears, Lau, Tyler and Allen 1980; Sears, Tyler, Citrin and Kinder 1978; Sniderrnan, Hagen, Tetlock and Brady 1991. absence of moral standards, or even class distinctions. It is clear, that in many respects, when people think about ideology, they are thinking about issues. This could suggest that issue positions influence how individuals label themselves ideologically. Empirical research on individual level self-reported ideology consistently finds a strong relationship between the ideological orientations of the public and the positions they take on policy.3 Conservatives take conservative positions because they value less government intervention in society and the status quo. Liberals take liberal positions because they believe government can solve the problems faced by society and are willing to accept progressive changes in policy to accomplish this. But, few studies have seriously looked at the potential for issue positions to directly influence ideological orientation, nor the possibility for a reciprocal relationship between the two. This dissertation takes that step. Why should we question the conventional belief about the uni-directional relationship between ideology and issue positions? Empirical findings have consistently shown that ideology affects issue positions. However, one important empirical finding questions just how important this relationship is. Most people do not think ideologically. Campbell, Converse, Miller and Stokes (1960) were one of the first to draw this conclusion, and since then research has generally supported their finding. The idea that people are unsophisticated when thinking about politics has been coined “minimalism” and minimalism has been a dominant theme in public opinion 3 See Bennett and Bennett 1990; Carmines and Stimson I989; Fleishman I986; Gamson and Modigliani 1989; Jacoby 1994, l99l, I986; Kuklinski, Metlay and Kay I982; Lau. Brown and Sears 1978; Luttbeg and Gant 1985; Sears 1993; Sears and Citrin 1985; Sears, Hensler and Speer 1979; Sears, Huddy and Schaffer 1986; Sears, Lau, Tyler and Allen 1980; Sears, Tyler, C itrin and Kinder I978: Sniderman, Hagen, Tetlock and Brady I991. research since the 19603. “Mass publics, it was contended, were distinguished by (1) minimal levels of political attention and information; (2) minimal mastery of abstract political concepts such as liberalism-conservatism; (3) minimal stability of political preferences; (4) and quintessentially, minimal levels of attitude constraint” (Sniderman 1993, 219). A focus on minimalism leads to the conclusion that because most people do not think about politics in an ideological way, the relationship between ideology and policy positions only matters for the small subset of the population who do. This limits the extent to which the idea that ideology informs or helps organize policy positions can be generalized. As a consequence of the Campbell, Converse, Miller and Stokes (1960) conclusion, researchers were driven away from using ideology in their research. “Why complicate an empirical model with a variable that was already known to have no effect” (Jacoby 2002, 99). However, ideology became a useable concept once again, in the late 19603, when the Comparative State Elections Project asked respondents about their ideological self—placements. This data was used in an influential study by Kovenock and Beardsley ( 1970) that provided the “push” for the National Election Studies to begin using the ideological self-placement scale in their data set, and has continued to be used since 1972 (Jacoby 2002, 100). Sniderman (1993) argues the focus on minimalism changed, as well. Researchers finally accepted the premise that mass publics did not think ideologically. After which, the predominant question in the literature became how do people still develop issue positions and place themselves on an ideological scale? Whether or not individuals use some sort of abstract ideological concept to sort and organize their issue positions, many will still take non-neutral positions. So, what is behind these positions? In their attempts to answer this question, researchers ran into a road-block. Attempting to orchestrate an overarching theory, which would explain how mass publics organize their worldviews, overlooked the possibility that different individuals may operate in different ways. Trying to develop a theory that would account for how most or all individuals organize their policy positions left researchers scratching their heads after the evidence convinced them that most individuals are not sophisticated political thinkers. The need for a theoretical approach that allowed individuals to form their belief systems in different ways became evident. It was no longer feasible to focus on developing a theoretical model that could be generalized to the population as a whole. Research needed to formulate a model that could account for the different ways in which individuals develop and maintain their positions on policy. As Sniderman (1993) asked, Why suppose that everyone adopts a position, or makes a choice, for the same reason? And more specifically, why suppose, given the striking differences in political information and sophistication within the mass public, that the citizen who is politically aware and attentive makes up her mind in the same way as the political ignoramus? (224). This dissertation attempts to construct a model that accounts for two important ways that the relationship between ideology and policy positions varies. The model not only accounts for the conventional wisdom that ideology helps people determine their policy positions, but it also allows for positions on policy to help form ideological positions. The goal is to show that we should question the assumption that the relationship between ideology and issue positions only works one way. Instead, political scientists should consider the possibility that for many individuals. it is their issue positions that help them label themselves ideologically. For them, it is more difficult to arrange an entire belief system than to take individual positions on a few (or several) policy issues. Consequently, they take these positions, and using cues from their preferred political parties. are able to determine where people who hold positions such as themselves stand ideologically. Thus they are able to identify themselves under an ideological label without being ideological. Evidence for this argument is found in Levitin and Miller (1979), who conclude there is a substantial portion of the American public who use ideological labels with only a limited understanding of what those labels mean. But, these labels are still important for these citizens and they may create links that help citizens make sense out of politics, even though they attach little systematic content to those labels. These individuals are using ideology in a fundamentally different way than would be assumed of the more ideologically sophisticated. Therefore, we should believe that how their ideology interacts with other political attitudes is also different, and we should account for this when modeling their behavior. Therefore, it is important to consider a group of individuals who use policy positions to develop their ideological self-identification. Similarly, Converse (1964) makes a distinction between those who actively use ideological labels to evaluate political objects, and those who merely recognize ideological labels when they are presented. He finds that almost 40% of his sample could recognize the labels and place them with the corresponding political party. Out of this top 40%, only about 16% were considered active users of ideological terms. These individuals could not only recognize the terms “liberal” and “conservative,” but they could also match the label with the proper political party and provide a broad description of the two terms. That leaves about 24% who could recognize the labels and match them with the appropriate parties. but could not provide a broad understanding of the labels. The rest either could not recognize the terms or were unable to correctly the match the terms with the corresponding parties. Clearly, there is a subset of the public that can recognize ideological labels, can accurately apply them in describing political groups, but do not do so actively. In fact, they only give ideological references to parties or candidates when prompted, and even then they show only a limited understanding of what the labels mean. It would be a mistake to assume that this group would use ideological concepts to structure their issue positions, when they do not actively use these concepts to organize their thoughts regarding other political objects. Instead, this group is characterized by its dependence on party cues to label its ideologies. They know basically what sides they take on a few particular issues, but their understanding of ideological labels stems mainly from cues they receive from partisan elites. This greatly limits their understanding of what these concepts mean, but provides them with a label to describe their broader set of issue positions. This could explain why people are able to place themselves on the ideological continuum, but have a limited understanding of what that continuum means. Or, as Levitin and Miller (1979) describe them, “These individuals describe themselves with ideological labels, often hold attitudes similar to those of ‘real’ ideologues, and may vote similarly; but unlike ‘real’ ideologues they do not have coherent, well-articulated views of the political world” (769). They pay attention to politics, but not to a level where they can develop a sophisticated ideological orientation on their own. They need help from elites to tell them which side they are on, based upon how they personally feel about particular issues. It is important ”to identify these individuals for several reasons. First, arguing that there is a directional link from issue positions to ideology challenges the conventional wisdom that correlations (or empirical relationships) between ideology and issues demonstrate a one-way causality from ideology to issue positions. It highlights the possibility that people are organizing their views about politics in previously unconsidered ways. Second, it adds to the growing literature that focuses on developing ways to sort out how different types of people orient their political attitudes. It clearly shows that while at times groups of people may come to the same political conclusions, they can come to those conclusions in fundamentally different ways (and while those conclusions may appear similar, they may be structurally different, as well). We need to model the different ways that people come to political conclusions, and this is a step in that direction. Third, it emphasizes the role that elites play in helping individuals place themselves ideologically. Not only is there an interaction between elites and the positions people take on issues, elites also provide individuals with the proper labels to associate with their collection of issue positions. The empirical model presented accounts for these interactions, and shows to what extent elites play in influencing the attitudes of the American public. To do so, this dissertation has four main objectives. First, it reviews the literature on how individual level political ideology helps people form opinions on political issues, while also developing a theory as to why we should believe that for many individuals it is issue positions that help them label themselves ideologically. Second, it shows that there is clearly a directional link from issue positions to ideology among many American citizens and that we must consider the possibility issues are influencing ideology. In other words, in contrast to previous literature, there is also a causal link from policy positions to ideology that has previously been unexamined. Third, it shows that even when looking at particular subgroups, who should be more likely to have their ideological orientations influence their issue positions, the data suggest that in some cases their ideological identification is indeed influenced by their issue attitudes. These findings enforce the notion that ideology play a weaker role in attitude formation. Finally, this dissertation makes the case that the relationship between ideology and issue positions may be changing significantly. Data analyses in Chapter 3 strongly indicate that a reciprocal link from ideology to attitudes existed as far back as 1976. From 1976 to 2000, we still see that ideology is involved in forming political attitudes, especially for the individuals whose ideological orientation shifted between 1972 and 1976. However, the Chapter 3 analyses also show that issue attitudes played a role in helping individuals label themselves ideologically, as the results continue to demonstrate a reciprocal relationship between ideology and policy positions. The analyses in Chapter 4, show that during 2004, attitude formation had changed considerably. Ideology had lost any importance it once had in forming political attitudes, and instead the effect of attitudes on ideological identification had strengthened and become dominant. In 2004, ideology no longer played a significant 10 role in attitude formation. Instead, the liberal-conservative continuum only served as a way for individuals to label their issue positions. Looking forward, Chapter 2 considers the current state of the literature on the relationship between ideology and issue positions. It first looks at the factors that influence the policy positions that individuals take, and then specifically looks at the literature tying ideology to political attitudes. Finally. it addresses whether there is reason to consider the possibility of a reciprocal relationship between ideology and issue attitudes. Chapter 3 develops and tests a model of attitude formation that accounts for a reciprocal link between ideology and political attitudes. It considers data from 1972 to 2000 taken from the American National Elections Studies (ANES). Specifically, the data are used to test the model on the 1972-1976 ANES Series File and the ANES cross- sectional data from 1984, 1996, and 2000. Chapter 4 looks specifically at the 2004 ANES cross-sectional respondents, applying the same model employed in Chapter 3. Here the model is not only tested on all relevant respondents, but differences in political conceptualization and education are considered as factors that might foster differences in the relationship between ideology and issue attitudes. Chapter 5 delves more deeply into the 2004 data, looking specifically at attitudes toward the right of homosexual couples to marry. Additionally, this chapter looks at attitudes towards the rights of homosexuals in general and compares those findings to attitudes on non-lifestyle issues such as defense spending and whether the government should seek to provide jobs and a good standard of living. 11 Table 1.1 Perceived Meaning of Ideological Labels (1994) Type of Mention Example Percentage Mentioning Issue Mention Liberals are pro-choice. Abortion . . Conservatives are pro-life Liberals favor gay rights. Ga RiUhts . . y ‘3 Conservatives oppose gay rights. Liberals are weak on defense/national security. Conservatives are strong on defense/national security Defense Liberals are for welfare state/give-away programs. Fiscal Conservatives are for 24 capitalism/oppose social prggLams. Liberals are free spenders/favor government spending. Spend/Save Conservatives are 24 thrifty/economize on government spending. Liberals support upholding of Bill of Rights/human rights. Conservatives want to limit Bill ofRights/human riLhts. Civil Liberties Other Mention Liberals accept change/new 'deas/inn v 've. ., Change I 0. an . 2.) Conservatives re51st change/protect status quo/ri id. Liberals are open-minded/not concerned with consequences. Personality Conservatives are 14 moralists/concemed with consequences. Liberals are not interested in setting moral standards/not Morality religious. l3 Conservatives have definite moral standards/reflgfius Liberals are for little people/working people/unions. 12 Class Conservatives are for big business/the rich. Liberals identify label with rominent national f1 ures. People p g 5 Conservatives identify label with prominent national figures. N=595. Questions allowed for multiple responses. Source: Erikson, Robert S. and Kent L. Tedin. 2005. American Public Opinion: Its Origins, Content, and Impact. New York: Pearson Education, Inc. Chapter 2 Ideology Voting research in the 19605 and 19705 dismissed the notion that the American public could use ideological labels to structure their political beliefs. For example, the authors of The American Voter wrote, “Our failure to locate more than a trace of ‘ideological’ thinking in the protocols of our surveys emphasizes the general impoverishment of political thought in a large proportion of the electorate” (Campbell, et al. 1960, 543). These early findings also indicated that the public was unable to use ideological labels to evaluate candidates or issues. “The correlation(s) between them [ideology and attitudes on policy issues] tend. . .to be fairly low, suggesting that for the present, at least, many Americans divide in their party preferences, their support of candidates, their economic views, their stands on public issues, or their political self- identification without reference to their beliefs in liberalism or conservatism” (McClosky I958, 44). The consensus was that Americans did not conceive of the political world in ideological terms, and therefore, it was unnecessary to consider ideology when analyzing how individuals evaluated political phenomena. One question to be addressed then, is whether or not this failure to use ideology has continued. To measure ideology, political scientists have often employed a seven- point scale, ranging from extremely liberal to extremely conservative with moderate being the middle category. Typically, individuals are willing to locate themselves on this scale.4 However, what this measure tells us has been riddled with uncertainty. Are the placements accurate? Does placing yourself on a seven-point scale give a proper 4 Usually about three-fourths of NES respondents can place themselves on the liberal-conservative scale. About one-third select the moderate or midpoint position (Jacoby 2002, 100). 13 assessment of where you stand politically on items such as issue positions or candidate preferences? Free and Cantril (1967) find considerable inconsistency between ideological identifications and policy preferences. Others have shown skepticism regarding the level to which individuals use their self-placements to facilitate issue-based voting 9 (Levitin and Miller 1979; Conover and Feldman 1981). Also, many individuals have a hard time placing candidates and parties along the same ideological dimension (Levitin and Miller 1979; Jacoby 1988; 1995).5 However, Nie, Verba, and Petrocik (1976) found that from 1956 to 1973, there was a considerable increase in issue consistency within the American public. Self- identified conservatives (and liberals) were more likely to take conservative (liberal) stances on traditional issues and new issues than they were during the American Voter ’3 period of analysis. As Jacoby (2002) states, individual ideology and policy attitudes were considered synonymous. In other words, they were two ways of looking at the same thing (102). Holm and Robinson (1978) challenged this notion arguing that liberal- conservative identification “helps define a person’s own political identity and therefore functions in a manner similar to party identification as an influence on subsequent behavior” (Jacoby 2002, 102). In this way, personal ideological identifications have an affect on voting behavior independent of issues and partisanship. This conceptualization of ideology is similar to symbolic politics theory developed by David Sears and colleagues in a series of pieces (Lau, Brown and Sears 1978; Sears 1993; Sears and 5 For a very good summation oftltis literature, see Jacoby (2002). 14 Citrin 1985; Sears, Hensler and Speer 1979; Sears, Huddy and Schaffer 1986; Sears, Lau, Tyler and Allen 1980; Sears, Tyler, Citrin and Kinder 1978). Conover and F eldman (1981) agree that ideology is wrapped in political symbolism. Arguing that positive and negative evaluations of conservatives and liberals as groups drive ideological identification. They conclude that the symbolic content inherent in such group identifications help individuals label themselves ideologically based on how they positively or negatively evaluate those groups. Any effect that issue positions have on ideological identification is mediated by the evaluations of liberals and conservatives as political groups. In other words, where individuals perceive other liberals and conservatives to stand on issues may influence whether or not they evaluate these groups positively or negatively, but the specific issue positions do not have a direct effect on an individual’s own ideological placement. More will be said about this study later. In order to better understand the relationship between issues attitudes and ideological identification, it is necessary to understand the origins of issue positions. In this first section, I look at several factors that have been shown to affect the issue attitudes of individuals. I then consider the role that ideology plays in this development and question whether the relationship is uni-directional. I conclude by arguing that issues attitudes can and do influence ideological identification, and that it is important to model this reciprocal relationship when analyzing any interaction between the two The Origin of Issue Positions There are a number of influences on political attitudes. These factors can include family, peers, political elites, partisan attachments. self-interest, and symbolic 15 predispositions. For instance, looking at opinions on European integration, Ray (2003) finds that the position taken by a political party on an issue can cue supporters of that party to take a similar stance. However, this effect varies according to characteristics of the party, the national context, and the individual (990). Issue importance to the party and the strength of the individual‘s partisan attachment increase the power of the partisan effect on the individual’s issue position. Alt, Sarlvik and Crewe (1976) find that in the British case, both the strength and direction of partisanship affects citizens’ attitudes on policy-related issues. They also find that issue positions have a reciprocal effect on partisanship. That is, partisanship not only influences the opinions citizens have on issues, partisanship is also shaped, reinforced, or changed by those same issue positions. This finding corresponds well with previous research by Campbell, Converse, Miller and Stokes (1960), “Party loyalty plays no small role in the formation of attitudes on specific policy matters. . .On the other hand, if an individual has intense feelings about an issue before partisan alignments form and his party’s subsequent policy conflicts with such belief, they may act as important forces toward partisan change” (169). The possibility of a reciprocal relationship between partisanship and issue positions is interesting. In some ways partisanship works similarly to ideology. For instance both have been considered by some academics as group attachments (Campbell et. al. 1960; Conover and Feldman 1981; Miller and Levitin 1977). Originally, it was argued that partisanship is developed early in life, reliant on parental partisanship and relatively stable over time (Campbell et. al. 1960). However, as noted above and later on, recent research has questioned the exogeneity of partisanship. suggesting that it is 16 malleable and can fluctuate due to changes in issue attitudes and perceptions of where preferred political parties stand 011 issues. Perhaps, in the same sense, ideology and issues are endogenous, able to produce changes in one another. Martinez and Gant (1990) argue that changes in partisanship are determined, at least in part, by partisan issue changes, defined as a shift from being closer to one party on an issue to being either closer to the other party on that issue or indifferent between the parties or, as a shift from indifference to being closer to one of the parties (249). If a person discovers that their preferred party is no longer close to them on an issue they find important, they may change their partisanship to another party that is closer to them on that issue. Additionally, the literature on political socialization has held that issue positions can influence partisanship. Several studies have argued that partisanship is endogenous to political preferences, and can change in response to the current attitudes an individual holds (Markus 1979; F iorina 1981; Franklin 1984; MacKuen, Erikson, and Stimson 1989). For instance, Niemi and Jennings (1991) show that shifts in partisanship during adulthood as well as between generations are influenced by policy preferences. “The number of issues that impinge on partisanship, and the magnitude of their effect, (grows) rapidly as adults move into their twenties and thirties” (986). In fact, it appears that issues play at least some role in developing partisanship before individuals embark on aduhhood. The traditional view is that party identification influences issue attitudes. However, as noted by several of the studies above, the possibility of feedback between the two has been considered even as early as The American Voter. Not surprisingly then, 17 Franklin (1984) considers a reciprocal link between partisanship and issue preferences and finds that once simultaneity is taken into account. party identification appears to play little role in determining issue preferences. “Of much more importance are the measures of education, region, ethnicity, social position, and unemployment” (472). Page and Jones (1979) find similar results. however Jackson’s (1975) conclusion indicates an influence of party identification on issue preferences. He notes, “The most important role played by party identification. aside from the effect on the voting decisions of people. . .is its influence on the issue positions taken by voters” (176). To complicate matters further, Jacoby (1988) finds convincing evidence that partisanship influences individual positions on issues, but that individuals do not project their own positions onto their preferred parties. Testing once again for a reciprocal relationship between issue positions and partisanship, Jacoby (1988) shows that when employing a fuller model, which takes into account how partisanship actually influences issue positions by including measures of where individuals perceive the parties to stand on particular issues, the impact of issue positions on partisanship appears to diminish. The difference between this work and others, is that Jacoby argues partisanship (Republican vs. Democrat) cannot have a direct effect on issue attitudes. Instead. it must work indirectly through cues about where the parties stand on particular issues. The direction of partisanship (Republican or Democrat) determines which cues a person will use. . . Simultaneously, the strength of party attachment (leaning, weak, or strong) affects the importance of the cues—that is, their salience to the individual—in the attitude formation process. Thus party identification affects the sources of the issue attitudes (i.e., the perceived party issue positions); it does not exert a direct influence on the issue attitudes themselves (646). 18 Thus, taken together, it would appear that the empirical evidence on the relationship between issues and partisanship is mixed. However, it seems clear that at least to some extent party identification influenCes individual issues positions. Family and peers can also shape political attitudes. Tedin (1980) looks at agent influence on individual political attitudes and argues that parents and peers can have divergent viewpoints on issues, and yet influence an adolescent in the same direction. He finds that when issues are important to parents or peers and accurately perceived by adolescents, the correspondence between attitudes is high. Generally, parent influence is more prevalent, simply because parents are more political. However, on issues where peers have interest, such as legalization of marijuana or the 18-year old vote (Sebert, Jennings, and Niemi 1974), parental and peer influence can be of similar magnitude. Additionally, opinion often reflects self-interest. For instance, Tedin, Matland, and Weiher (2001) find that self-interest played a considerable role in attitudes towards voting in a school bond election. Symbolic values such, as ideology and party identification did not appear to have an effect on the white voters in the sample, but had a large impact on minority opinions. Racial resentment did have an effect on white voters, but contrary to past findings (ie. Lau, Brown, and Sears 1978; Sears, Lau, Tyler and Allen 1980), was not predominant. This strong support for self-interest corresponds with earlier arguments made by Downs (1957), Campbell et al. (1960), and Lipset (1960), who contend that citizens choose policy positions that further their own self-interest and vote for political candidates who support those policy stands. Similarly, preferences on policy have been analyzed as a function of personal utility (Page 1977). However, a line of work by Sears 19 and others has brought into question the actual dominance of self-interest on issue attitudes. These authors find that symbolic attitudes, such as racism and ideology play a much stronger role in predicting issue attitudes than self-interest. Ideology as a Predictor of Issue Positions While we know that ideological thinking by the public is limited (Campbell, Converse, Miller and Stokes 1960; Converse 1964; Knight 1985; Luttbeg and Gant 1985; Jacoby 1986), there have been several studies that have shown ideology to be one of the strongest predictors of issue attitudes. For instance, Fleishman (1986) uses cluster analysis to test whether individuals can be grouped into the traditional categories of liberal, conservative, or moderate based on their positions on several important issues. He finds that respondents can be clustered into six different groups ranging from liberals (group 1) to economic moderates (group 6). Based on these findings, we can conclude that there is a level of consistency among issue positions for most respondents, and that this consistency is manifested within several different ideological groups. These groupings may not conform to the traditional liberal-moderate-conservative continuum, but there are significant and important policy differences between the groups that Fleishman analyzes, which suggests that respondents are organizing their opinions along some ideological dimension. Soss, Langbein, and Metelko (2003) analyze White America’s support for the death penalty. Their findings suggest that support for the death penalty among whites is based on a number of attitudes. These include social and governmental trust and individualist and authoritarian values. Congruently, they find that racial attitudes influence white support of the death penalty. Racial prejudice increases the likelihood of 20 being in favor of the death penalty, and as the number of blacks in a respondent’s county increases the effect of racial prejudice increases substantially. However, even when controlling for these types of attitudes, ideology is still a significant predictor of attitudes toward the death penalty. Conservatives are more likely to favor the death penalty and liberals are not. Kuklinski, Metlay, and Kay (1982) tackle the question of what happens when a new issue comes along. When individuals are faced with a policy question they have not been able to think about before, what helps determine their attitudes on that policy (if they actually form an attitude on it)? Looking at the issue of nuclear energy, they examine the decision calculus citizens use to come up with a position on a policy that is relatively new to them. To ascertain what goes into these decisions, the authors look at three different modes of citizen decision making. First, individuals may decide instrumentally, i.e. they weight the consequences and the likelihood of those consequences against the potential benefits of said policy. Second, citizens may adopt the preferences of their preferred reference groups. If the position on the policy of the preferred group is known to the individual, he or she may defer to their preferred group (which will share the values and norms of the individual) and choose the side of the group. “In this instance, the reference process consists of the individual seeking a group’s ‘advice’ about an issue that interests him but which he does not fully understand” (618). Obviously, in this situation the group would have to be one that is considered to have knowledge or expertise 011 the policy at hand. Third, citizens may rely on core values that help them order their political world. In the case of nuclear 21 energy, the authors argue two values are important, views on technological advancement and political ideologies. The paper is most concerned with the interaction between the level of knowledge the individual holds concerning the issue at hand and the above three decision-making processes, and how this interaction affects vote choice. They find that knowledgeable citizens rely heavily on ideology, which influences their cost-benefit calculation about the pros and cons of the policy. In other words, ideology has a direct and indirect effect (through the cost-benefit calculation) on attitudes towards the policy (as expressed through vote choice). So, when it comes to new issues facing the public, it appears that ideology plays a key role in determining attitudes for those who have the most information about the issue. Others, notably Sears, Hensler and Speer 1979; Sears, Lau, Tyler and Allen 1980; Sears and Citrin 1985) have found a similar relationship between ideology and issue positions. But, these findings do not mesh with the already extant literature that argues that individuals do not think ideologically, and therefore ideology has little influence on their political beliefs. If the public does not think ideologically, then how can ideological identification impinge so heavily on the attitudes of the public? Jacoby (1991) argues that these contradictory conclusions stem from, “theoretical ambiguities about the nature of ideological identification itself” (179). In fact, “Many people are willing to locate themselves along the liberal-conservative continuum, but it remains unclear how this impinges on other beliefs and attitudes” (179). J acoby (199]) goes on to layout three ways in which ideological identification has been described in the literature. First, ideological identification has been categorized as a symbolic politics orientation. Ideology is viewed as a standing predisposition, formed early in life, which affects issue attitudes through its relationship with the “manifest symbolic content” inherent in issue alternatives (Sears, Huddie, and Schaffer 1986). In other words, “individuals react positively to issue positions that match their own symbolic predispositions and vice versa” (J acoby 1991). Ideology and issue positions contain symbolic content and this helps individuals match their issue positions with their ideology. Sears, Hensler and Speer (1979) found that when operationalizing ideology as a symbolic attitude (along with racial intolerance), ideology was a much stronger predictor of whites’ opposition to integrated busing policy than an individual’s own self-interest in the busing issue. Those who identified themselves as conservatives, were more likely to be opposed to busing than those who identified themselves as liberals, regardless of whether the individual had any type of personal stake in the debate on busing to achieve integration. In an extension of this work, Sears, Lau, Tyler and Allen (1980) looked at not only busing, but also attitudes on unemployment, national health insurance, and law and order. Again, they found that symbolic attitudes, such as ideology, party identification, and racial prejudice, were much more important in predicting positions on these issues than an individual’s self-interest. They also found that sophistication considerably increased the power of symbolic attitudes. Second, another line of literature (Conover and Feldman 1981; Miller and Levitin 1977) has argued that ideological identification involves group attachments. Ix.) b) Brady and Sniderman (1985) found that citizens can estimate what political groups stand for on major issues. People also develop a positive sense of attachment to certain groups . and then adjust their attitudes to conform with group norms. Since, “virtually every issue has identifiable liberal and conservative positions. . .ideological reference groups should have a particularly broad effect, across a variety of issue attitudes” (Jacoby 1991, 180). Finally, J acoby (1991) argues that ideology has been viewed as a form of schematic information processing (Hamill and Lodge 1986; Hammil, Lodge and Blake 1985; Sharp and Lodge 1985). Ideology allows individuals to integrate new information into a set of organized categories, called schemas. These categories are defined as “liberal” or “conservative,” and act as reference points, allowing people to compare information to their own positions along the liberal-conservative continuum, evoking favorable responses to those attitudes that are closer to the individual ideologically. However, Jacoby (1991) finds little support for either the group attachment argument or the symbolic politics argument. If either of these hold, empirical results should show that most people take a non-neutral ideological stance. This is not the case. In any given NES survey, many respondents refuse to place themselves on the ideological continuum and many others place themselves as either “moderate” or “middle of the road.”6 At the same time, J acoby (1991) finds that ideological identifications are influenced by education and conceptualization, which directly contradicts Sears et al. (1980) finding that political sophistication does not change the relationship between symbolic attitudes and issue positions. Thus J acoby (1991) 6 In the 1972-1976 ANES Panel Study about 39% of respondents in 1972 did not place themselves ideologically. In 1976, about 28% failed to do so. In the 2004 ANES, 24% did not place themselves on the ideological scale. 32% chose the moderate or middle of the road position. 24 concludes that, “ideological identifications do not really function like the other symbolic politics orientations examined by Sears and his associates” (198). Similarly, Jacoby (1991) finds little evidence to suggest that ideology is a type of reference group phenomena. Once again, if this is the case, we should find a limited number of individuals who take a neutral ideological stance. This is clearly not the case. Secondly, Jacoby (1991) finds that those who fall into the group benefits level of conceptualization show very little consistency between their issue positions and ideological placements, showing that group attachments do very little to help respondents organize their political attitudes. J acoby (1991) does find strong support for the schematic approach, arguing that the effects of ideology on issue attitudes are confined to only those individuals who actually use liberal-conservative thinking to structure their belief systems. The presence of this ideological schema is determined by level of conceptualization and political sophistication. Those who conceptualize the world as ideologues (Campbell et. al. 1960) and those with higher levels of education are more likely to use ideology to structure their attitudes than those with lower levels of conceptualization and education. This would indicate that what is really driving the empirical relationship between ideology and issue attitudes is a combination of how individuals conceive of the political world and how educated they are. Based on this evidence, we can assume that ideology does have an effect on the issue attitudes of the public, and this effect is mediated by factors such as sophistication and conceptualization. But what role, if any do issue attitudes play in influencing ideological identification? None of the above studies have considered the possibility that there may be a reciprocal relationship between ideology and issue attitudes. What evidence is there that suggests we should consider one? Do Issues Affect Ideology? Conover and F eldman (1981) question whether issues affect ideological g orientations. They argue that issues do not affect ideological self-placement directly. Instead, the relationship is mediated by how an individual evaluates liberals and conservatives as groups. To measure these evaluations they consider feeling thermometer scores for liberals and conservatives. They also find that issues are not nearly as important in this relationship as symbolic meanings for ideological labels. They contend that the meaning individuals may give to ideological labels can be two fold. (1) It can be cognitive, which means people are considering objective information or substantive content when evaluating the label. So, they might be thinking about particular policy positions that are associated with the label. Or, (2) it can be evaluative. That is, they are evaluating the symbol based on the affect elicited by the symbol. This would be manifested in positive or negative feelings towards the symbol. This is the reason the authors use feeling thermometers as a measure of affect. To capture this, their model consists of a link from issues and symbolic content to the evaluations an individual has of the two ideological labels. Then a link is drawn from the two labels to an individual’s ideological self-placement. To measure symbolic content, they use factor analysis to create six additive scales from the feeling- thermometer ratings of 27 different groups in society. They argue that these groups should make up the symbolic content associated with ideological labels. In other words, a particular group serves as a symbol of an ideological orientation. In sum, the model is 26 constructed such that, evaluations of particular groups and a summary of an individual’s positions on the economy, racial issues, and social issues are tested to see if they have an effect on the evaluations of liberals and conservatives (also based on feeling thermometer scores), which is then tested to see if there is an effect on ideological self- placement from the ideological evaluations. They find that evaluations of liberals and conservatives are the most immediate determinants of ideological self-identification, and that these evaluations mediate the effects of the symbolic variables and issue variables. When considered together in a model, the evaluations drown out any effect the symbolic groups or issue positions (save economic issues) have on ideological self-placement. What then, determines how an individual evaluates the ideological labels? To test this, the authors regress the ideological evaluations on the symbolic group scales and the issue scales and find that feelings about the radical left, capitalism, and the reformist left have a significant impact on evaluations of liberals. Feelings towards the status quo, capitalism, and social control have a significant impact on the evaluations of conservatives. Conversely, when looking at issues, only one issue scale per evaluation has a significant effect. Economic positions seem to be influencing evaluations of liberals, and positions on racial policy are influencing evaluations of conservatives. Based on these findings, the authors conclude that it is the symbolic variables that are driving how individuals evaluate conservatives and liberals and not the positions they take on issues. By using feeling thermometers for groups and arguing that ideology is in part driven by affect towards these groups. the authors are making an assumption that ideology is a type of group attachment. But, from the above discussion we know that J acoby (1991) shows ideology is not a group attachment. If it is, we should find some level of consistency between issue attitudes, showing group attachments help organize issue positions. We do not. So, it is of little surprise to find that when conceptualizing ideology as a reference group phenomenon the authors find a weak connection between issue attitudes and ideology. We would not expect to find a strong correlation because those who use ideology as a group attachment are less likely to have consistent issue attitudes. However, an additional study by Levitin and Miller (1979) finds little evidence to suggest that issue positions influence ideological identifications. They contend that individuals view parties and candidates to have more association with ideological sentiments than issue questions, and that this suggests, “policy positions alone may be of relatively limited importance in developing and maintaining an ideological stance” (767). Given the findings from these two lines of research, why should we assume that issue positions have an affect on ideological self-placement? First, Levitin and Miller (1979) do find that ideological identification and issue positions are correlated. They do not correlate very highly, and their relationship is not as strong as that of ideology and candidate or party placements, but they are correlated nonetheless. Determining how strong the relationship between ideology and issue positions is, may depend on whether or not it holds when controlling for perceptions of where the parties and candidates stand ideologically. 28 Second, research consistently finds a link between ideology and issue positions, whether it is correlations between the two or regressions that include ideology as a predictor of issue positions. However, to my knowledge, no study has considered a reciprocal link between the two in the same model. In the cases where the relationship is analyzed simply through bivariate correlations, directionality cannot be assumed to run either way, and in other cases where ideology is used to predict issue positions, directionality is assumed to be unidirectional, but without a test to verify this assumption. Third, there is work that would suggest individuals can conceive of their ideological identification in two different ways, operational or symbolic (Stimson 2004). The concept of an operational ideology is of chief interest here, in that it is made up of issue positions. That is, by measuring the consistency and similarity between an individual’s issue attitudes, one can get a decent picture of what their political ideology is. Stimson (2004) shows that many times this creates a disconnect between the issue attitudes an individual holds and where they place themselves on the ideological continuum. For instance, he discusses conflicted conservatives (about 20% of the US. population) who take liberal positions on issues and yet identify themselves as conservative. This would imply that an individual’s “real” ideology is represented by the issue attitudes they hold, regardless of where they place themselves on the ideological spectrum. It would then not be a stretch to consider how issue attitudes help develop ideological orientations. Fourth, consider party identification. 1 have discussed at length the possibility that party identification and issue attitudes have a reciprocal relationship. This relationship operates in a similar fashion to what I am arguing about ideology. Several studies have shown that party identification helps individuals take positions on issues, and we have also seen that issue attitudes can drive changes in partisanship. Is there any reason to believe that ideology might work in the same way? Both ideology and partisanship have been considered reference group attachments (Conover and Feldman I981; Jacoby 1988; Miller and Levitin 1977) and symbolic orientations (Sears, Lau, Tyler, and Allen 1980), they correlate highly,7 and they tend to be relatively stable over time (N ie, Verba, and Petrocik 1976). Perhaps they operate similarly in their relationship with issue positions. If we conceive of ideology as an orientation similar to party identification, developed early in life and permeating through the decisions individuals make throughout their lifetimes, then clearly we can believe that ideology is developed before many (if not all) issue attitudes. It would then be easy to understand that as new issues arrive, ideology would then help shape attitudes on these issues (Kuklinski, Metlay and Kay 1982). However, perhaps this is not a one-way process. While ideology helps people take positions on new issues they are faced with, these issues positions can also help mold one’s ideology. In other words, ideology and issue positions feed off of each other, similar to how partisanship and issue positions operate. Lau, Brown, and Sears (1978) describe precisely this type of relationship when analyzing opinions about the Vietnam War. Both anti-communism and liberal-conservatism were attitude dimensions clearly antecedent to Vietnam. Attitudes toward Vietnam War protestors and toward ‘the military’ were equally obviously much affected by Vietnam. But, to some unknown degree, they also no doubt were based 7 For example, in the 2004 NES the correlation for party identification and ideological self-placement is .60. in earlier socialization. . .Thus, these symbolic beliefs form a continuum going from those which are clearly antecedent to the war to those which are both determinants of and determined by attitudes toward the war (477). Clearly, Vietnam represents a case where ideological orientations are formed prior to an issue arising, and where they help shape attitudes toward that issue as it moves onto the political agenda. However, it also represents a situation where an issue can help shape ideological orientations through its pervasiveness and importance. In other words, as opinions towards the Vietnam War developed, changed, or solidified ideological orientations were also affected in some way. Therefore, it should, at the very least, be prudent to model the relationship between ideology and issue positions as reciprocal and test whether or not there is an empirical effect of issue attitudes on ideological orientations. The next chapter outlines a model that incorporates a reciprocal link between ideology and issue positions, controlling for demographic variables, where individuals perceive their preferred parties to stand on issues and ideology, and their past issue positions and ideological placements. The model is tested on public opinion data from 1972-2000. If the model accurately captures the relationship between ideology and attitudes, as I have described it, we should find a strong positive link from issues to ideology, indicating that issues indeed play a significant role in how people place themselves ideologically. 31 Chapter 3 The Model The theory posited in the previous chapters, argues that many people use their issue positions to label themselves ideologically. When modeling this relationship, there should be a clear link from issues to ideology, showing that for many individuals, their issue positions help structure their ideological self-placement. They do this by relying on cues from political elites, which not only influence their issue attitudes, but also provide them with information on where they stand ideologically. Individuals have opinions on issues that are influenced by where their party stands on those issues. They also are able to label themselves ideologically by referring to where their party stands ideologically and what the ideology is commonly labeled. If an individual perceives their preferred party to be conservative, they are more likely to label themselves conservative. We should also find that as they place their preferred party more in one direction across issues, they should place themselves in that same direction. In other words, there should be a considerable level of congruence between where individuals place themselves on issues and where they place their preferred party on the same issues. There are several important studies that show perceived party positions impact how citizens place themselves (Campbell, Converse, Miller and Stokes 1960; Conover and Feldman 1981; Hyman and Singer 1968; Jacoby 1988; Miller 1976; Miller and Levitin 1977; Pomper 1972; RePass 1971; Schuman and Pomper 1975). These findings are rooted in reference group theory. As Jacoby puts it, “individuals develop psychological attachments to certain groups in their environment. These groups then C) to provide cues for structuring attitudes and behavior on matters relevant to the group” (Jacoby 1988, 644). Jacoby (1988) clearly shows a relationship between where individuals perceive the parties to stand on issues and where they place themselves on issues. He also finds that there is no reciprocal relationship between party placements and self-placements, so one is not considered in the model presented here. This closely resembles the issue evolution argument developed by Carmines and Stimson (1989). In their study of the evolution of racial attitudes, Carmines and Stimson show that as partisan elites take positions (or change positions) on issues, fellow partisans within the electorate will eventually adopt those same positions. This is mediated through the effect that partisan elites have on party activists. As Democratic elites became more liberal on racial policy, so did Democratic Party activists. This in turn, eventually led to a shift towards more liberal attitudes among Democrats in the electorate. It is important to note, that party identification by itself should not be used as a predictor for issue positions. Implying that one’s party identification has a direct relationship with issue positions, that is as party identification strengthens one’s issue positions move toward the extremes, makes little theoretical sense. One can easily be an extreme partisan without holding extreme issue attitudes. As Jacoby (1988) points out, this relationship is mediated by where the partisan sees their party placed on the issues. A stronger partisan will not necessarily perceive their party to be extreme on the issues, but will presumably be able to better locate where their party stands on the issues. The link from partisanship to issue positions must go through the cues individuals pick up D.) DJ from where they perceive their parties to stand. These cues not only work to influence issues attitudes, but ideological orientations as well. This relationship can be tested with a fairly simple empirical model. Figure 3.1 displays the model incorporating the above-mentioned variables in an explanation for how the relationship between partisanship, ideology, and issue positions works. Within the model, ideology is a function of several demographic variables and the individual’s positions on current political issues, as well as, where they perceive their preferred party to stand ideologically. Their issue positions are a function of the same demographic variables, where they place themselves ideologically, and where they perceive their preferred party to stand on those same issues. The model produces two equations. The first equation models the link from ideology to issue positions: EQ3.1. IssuePositionsl- = ,80 + fllldeologvi + [9’2 ParIyPlacementlssuesl- + 2 ,6 k Demographics ,- + £1 The dependent variable in this first equation, lssuePositions, is an index (or average) of the positions taken by individual i on several political issues. PartyPlaeementlssues is an index of where individual 1' places his or her preferred party on those same issues. Ideology is the individual’s self-placement on a seven-point ideological continuum, and Demographics represents a number of socio-demographic variables that should have an effect on an individual’s policy positions. These include race. gender. income. age, and education. What we are most concerned about here is the coefficient ,8, , which measures the effect that ideology has on an individual’s issue positions. If this coefficient is found to be positive and statistically significant, we can conclude that ideology has an empirical effect on issue positions. The second equation captures the link from issue positions to ideology: EQ 3.2. Ideologyi = a0 + allasuePositionsI- + a2PartyPlacementIdeolog/l- + Z a k Demographics i + 52 where, individual i’s ideology (Ideology) is a function of that individual’s summary of issue positions (IssuePositions), where they place their preferred party ideologically (PartyPlaeementIdeology), and the socio-demographic variables (Demographics). Here we are concerned with a, , the effect of issue positions on ideology. If this coefficient is positive and statistically significant, we can conclude that issue positions have an empirical effect on ideological self-placement. Specification, Data, and Methods The model is non-recursive due to the reciprocal link between ideology and issue positions. The variables that capture the perceptions of where the parties stand ideologically and on issues act as instruments to identify the model. In order to be valid instrumental variables, each must be correlated with its respective dependent variable but uncorrelated with the other dependent variable. In other words, where an individual perceives his/her preferred party to be ideologically is expected to have an effect on the individual’s own ideology, but have no direct effect on their issue positions. Also, where an individual places their preferred party on the selected issues should have an impact on the individual’s own issue positions, but not on the individual’s self-reported ideology. There is evidence to support this assumption. A two-stage least squares model assumes that the errors associated with yl and y2 are uncorrelated, so the instrumental variables (x1 and x2) should predict the error associated with yl and y2 respectively, but not the other. The purpose of instrumental variables is to allow us to distinguish between the direct effect of the two dependent variables upon each other and the errors associated with the dependent variables. That is, what is the effect of yl on y2 and y2 on yl minus whatever error is contained in either variable? In this case, where individuals place their preferred party on issues should help us predict the error in issue preferences, but we should not expect that it would help us predict the error in their ideological self-placements. At the same time, we should not expect where individuals place the parties ideologically to effect the error associated with their issue positions. but should expect it to help predict the error associated with their own ideological self-placements. In addition, there is little reason to believe theoretically that where an individual perceives his/her party to stand ideologically has an effect on their personal issue positions, without the effect being mediated by his/her personal ideology. An instrumental variable for yl is allowed to have an indirect effect on y2 as long as the effect is mediated by the effect of yl on y2. The same goes for the instrumental variable for y2. In this case, there is also no reason to believe that where an individual perceives his/her party to stand on issues has an effect on his/her personal ideology. outside of an 36 effect that his/her own issue positions might have 011 ideology. In fact, the literature has never argued that issue positions have a direct effect on ideology, so we should not expect perceptions of where the parties stand on issues to have an effect either. While, I am arguing here that issue positions do have a direct effect on ideological self- placement, I find no reason to believe that perceived party placement on issues has a direct effect on ideology. These assumptions are supported by a Hausmann test performed using the data below. For the equations listed above, the Hausmann test shows that the residuals from the IssuePositions equation are not correlated with PartyPlacementIdeology, the instrumental variable in the Ideology equation. Additionally, the residuals from estimating the Ideology equation are not correlated with PartyPlacementIssues, the instrumental variable in the IssuePosition equation. To test the model data are drawn from the 1972-1976 American National Election Studies Panel Study. Respondents were asked a battery of questions in 1972, many of which were asked of them again in 1974 and 1976. Relevant to this study, individuals were asked to place themselves ideologically in both 1972 and 1976 and where also asked their opinions on a number of issue items in both years. These issues included questions on whether government should do more to provide jobs and a good standard of living, whether busing should be used to achieve integration in public schools, healthcare. rights of the accused, aid to minority groups, urban unrest, and the tax rate. Table 3.1 shows the distributions to the answers on these questions. Each was originally measured on a seven—point scale with higher numbers indicating responses 37 typically considered “conservative.” The mean values for each of the items indicate that on average individuals in 1976 were slightly to the right in the positions they took on these issues. Only opinions on how the government should deal with increasing levels of urban unrest leaned slightly left. The distribution of the answers to these same items does not vary significantly between 1972 and 1976. These issues were combined into two indices. for 1972 and 1976 responses. All the issue variables have been re-scaled form zero to one. They were then combined into an additive index and averaged to get an individual’s mean score across all seven issue variables. This index is then re-scaled from zero to one, with higher numbers indicating an individual takes more right leaning issue positions, on average.8 Ideology is measured using the typical seven-point ideological continuum, with an answer of “one” indicating that the respondent identifies as extremely liberal and “seven” indicating the respondent identifies as extremely conservative. Also, several demographic variables measuring a respondent’s race, gender, education level, and income are considered. Table 3.2 shows the mean values and standard. deviations for these measures and for the two complete issue indices. As would be expected, both ideology and the issue indices indicate public sentiment was slightly right-of-center in both 1972 and 1976. Not surprisingly, from Table 3.3 we see that the issue index in 1976 correlates relatively high with ideological self-placement, at .506. Additionally, the issue index from 1972 correlates with the index from 1976 at .53, while ideological identification in 1972 correlates with ideological identification in 1976 at .56. 8 A more detailed explanation ofthis index is provided in Appendix A. 38 Results First, the results from analyzing the data only from 1976 are displayed graphically in Figure 3.2. The coefficients indicate that in 1976, the relationship between ideology and issue attitudes was endogenous. There are significant links from issue attitudes to ideological identification and vice versa, suggesting that these two sets of political beliefs are feeding off of each other. While it appears that the affect of issue positions on ideological self-placement might be stronger, both links are significant at the .05 level. This finding confirms the theory that issue attitudes do have an affect on ideological placements, and supports the argument that a reciprocal relationship between the two should be considered when modeling attitudes and ideology. The model also shows that the positions on which individuals place their preferred parties ideologically and on issues have significant effects on their own attitudes and ideological placements. This too, confirms the hypothesis that perceptions of political party beliefs help structure individual political attitudes. While this may appear as good evidence for the theoretical argument posed in this paper, one might still question the validity of the two instrumental variables. Because the data are drawn from a panel study, lagged variables for issue attitudes and ideological placements can be used instead of the party placements as instruments. Lagged variables can overcome some of the questions regarding instrumental variables. As stated previously, in order to be a proper instrument a variable must be correlated with yl but not y2. A lagged variable, by definition, is expected to be correlated with yl , but is considered temporally independent of y2. Therefore to test the robustness of the original model, a test including the lagged variables as instruments is conducted 39 below. The equations for this test are displayed graphically in Figure 3.3 and are shown here: EQ. 3.3 1.8‘s'uePositi0nsl-t = ,60 + ,BIIdeologyl-t + ,82PartyP/acementlssuesi + t 631.93uePosition.S'it _ l + Z 16k Demographics” + 81, EQ. 3.4 Ideology” = a0 + a11'ssuePositionsl-t + a2PartyPlaeementldeologyl-t + a3 Ideology” _ 1 + Z a k Demographics it + 82, where issue positions are a function of an individual’s ideological placement, where they place their preferred party on the same issues, their previous positions on those issues, and their demographics. Their ideology is a function of their issue attitudes, where they place their preferred party ideologically, their previous ideology, and demographics. A second Hausmann test was performed to analyze the validity of the instruments in this model. As was the case above, the test shows that the error associated with IssuePositions, when estimating EQ 3.3, is not correlated with either PartyPlacementIdeology or Ideology- ,, the instrumental variables in the Ideology equation. Similarly, the error associated with Ideology, when estimating EQ 3.4, is not 40 correlated with either PartyPlacementlssues or Is.S'z.iePo.s'itions,- ,, the instrumental variables in the IssuePosition.s' equation. The results in Figure 3.4, show the same type of relationship found in the model that includes the party placements as instruments. There is a clear endogenous relationship between ideology and issue attitudes, and the instruments work in their predicted ways. Ideological positions in 1972 predict ideological positions in 1976. The same goes for issue attitudes. If we are then convinced that the results are not driven by different model specifications, but perhaps are showing true attitude formation in 1976, what are we to take away from these findings? First, it appears that, at least in 1976, ideology had a discernible effect on the positions individuals took on issues. Second, as far back as 1976, we see that issue positions are also influencing ideological self-placement. There is a give and take between how ideological identification influences issue attitudes and vice versa. This makes it important to ask how changes in ideological identification/issue attitudes affect the other. The model shown in Figure 3.1 is employed to test if the relationship between ideology and issue attitudes is different for those who shifted either of these from 1972 to 1976. Figure 3.5 shows the results of the model when applied to those individuals who shifted their ideological self-placement at least two positions from 1972 to 1976. Clearly, for those individuals who shifted ideologically, the effect of their issue attitudes on their ideological self-placement was much more important than how their ideology influenced their issue attitudes. 41 Note that the coefficient for past ideological identification is negative, yet significant. Remember, these are only individuals who shifted in their ideological identification from 1972-1976. The measure for change is the absolute value of the difference between ideological identification in 1972 and 1976. Consequently, looking at the raw numbers (non-absolute values) shows that changes between 1972 and 1976 were more likely to be in the positive direction, meaning that more people were shifting to the right, rather than to the left. This fits well with the negative coefficient, in that for those who shifted in their identification from 1972 to 1976, they were more likely to shift to the opposite side. For instance, of the 145 individuals who changed, 44 (over 30%) moved from the three categories of liberal, slightly liberal, and moderate, in 1972, to outright conservative by 1976. Comparatively, individuals who shifted only one position, or did not change at all ideologically between 1972 and 1976, still exhibited a reciprocal relationship between ideology and attitudes, similar to the results found in the prior specifications. These results are shown in Figure 3.6. Clearly there is a difference in the relationship between ideology and issue attitudes between those that changed in their ideological identification and those who did not. Figure 3.7 shows the results when considering changes in attitudes. As would be expected, changes in issue attitudes appear to be driven by ideology. The attitudes of those who displayed a change of greater than .2 in their issue scale (equivalent to a 1.4 shift on the ideological scale) were driven by their ideological identification. Consequently, the issue attitudes of these individuals had no discernible effect on their ideological self-placement. Notice here, that the coefficient for attitudes in 1972 is not a 42 predictor of attitudes in 1976. This is reasonable considering that we are only examining those who changed attitudes during this time period. It is not surprising that if attitudes changed significantly between 1972 and 1976 that one does not adequately predict the other. Once again, the relationship between attitudes and ideology for those individuals who showed little or no change in their issue positions (Figure 3.8) fits the reciprocal hypothesis quite well. Taken together, these findings would suggest that changes in ideological self-placement are being driven by issue attitudes and vice versa. 1984 Respondents At least as far back as 1976, attitudes and ideology appear to be endogenous, but it is important to ask if and how this relationship has changed over time. Does data from the 19803, 19903, and 20003 show that the reciprocal relationship between ideology and issue attitudes found here, permeates across time? To address this, the model depicted in Figure 3.1 is tested on data from American National Election Studies cross-sections in 1984, 1996, and 2000. Respondents from 1984 are considered first. The variables used to test the 1984 data are constructed in the same way as the variables for the 1972-1976 data. However, as might be expected, it was not possible to use all the same issue attitudes to construct the issue index. When possible the same exact questions were used, however, a few questions asked in the 1972-1976 data were not available in the 1984 data, but a few new questions were available. The issue index for 1984 was constructed from six different issue attitudes, two of which were the same as those used in 1972-1976 and four new questions. Individuals were asked to identify their positions on governmental spending on services, aid to minorities, spending on 43 defense, the United States’ relationship with the Soviet Union, the role of women in society, and to what extent the government should work to provide jobs and a good standard of living for US citizens.9 Table 3.4 summarizes these issue attitudes. The means are positioned around the center, with attitudes on defense spending, the role of women in society, and spending on services tilting just to the left, and attitudes on aid to minorities, the Soviet Union, and government provided jobs and standards of living leaning just to the right. Table 3.5 provides a summary of the other variables in the model. Ideologically, the 1984 respondents identify between slightly conservative and conservative. The issue index is measured on a scale of 0 to 1. On average, the 1984 respondents appear to take moderate position on issues. Finally, Table 3.5 displays the correlations between ideology, the issue variables and issue index. The correlations for 1984 are a little lower than they were in 1976. This may have an impact on the strength of the relationship between ideology and issue attitudes when tested with the empirical model. The model depicted in Figure 3.1 is tested on the 1984 data using two-stage least squares regression. The results are displayed in Figure 3.9. The 1984 findings mimic those from 1976. Again, the analysis suggests a reciprocal relationship between ideology and issue attitudes. A move from 0 to 1 on the issue index increases ideological self-placement by .46, while a move from 0 to 1 in ideological placement increases an individual’s issue index by .25. The respondents in 1984, just like their 1976 counterparts, were simultaneously using their political attitudes to inform their ideological self-placement, while using their ideological orientations to organize their 9 The two issues that are identical to those used in the l972-l976 analyses are “aid to minorities” and “jobs and standard of living.” 44 political attitudes. However, the strength of the coefficients indicates, that even though both variables are significant, issue attitudes have a stronger impact on ideology. This finding also suggests that the results from 1976 were not isolated. However, it is necessary to continue looking at more recent data. 1996 Respondents As was the case in 1984, the questions used to construct the issue attitudes in 1996 are not all the same as those used in either 1976 or 1984. The issue index for 1996 is constructed from five different questions addressing policy attitudes. Two of these, spending on services and defense spending, were also used to construct the 1984 issue index. The other three ask respondents to provide their attitudes on abortion (four-point scale), the extent to which the government should be willing to create jobs at the expense of the environment, and to what extent government should regulate the environment. Table 3.7 summarizes these issue attitudes. Here again, the attitudes appear to be grouped near the middle. The mean values for attitudes on defense spending and spending on services land slightly to the right, while the mean values for attitudes on abortion, jobs versus the environment, and environmental regulation fall slightly to the left. Table 3.8 summarizes the issue index, the variable on ideology and the demographics. Both the issue index and ideology have been re-scaled from 0 to 1. Respondent ideology leaned slightly to the right, while on average issue attitudes were just to the left. However, table 3.9 shows that ideology and the issue index correlated fairly well in 1996, with a correlation of .54. Two-stage least squares regression is used to estimate the model on the 1996 data. Figure 3.10 displays the results. As was the case in 1976 and 1984, the 45 coefficients for both the link from ideology to issue attitudes and the link from issue attitudes to ideology are significant. A move from 0 to l on the issue index increases ideological placement by .82. A move in ideological placement from 0 to 1 increases an individual’s issue index by .13'. So, even though both coefficients are significant in the model, it would appear that issue attitudes have a much stronger impact on ideology. 2000 Respondents The issue index for the 2000 data is comprised of four questions. Respondents were asked their opinion on defense spending, spending on services, aid to minorities, and the extent to which the government should seek to provide jobs and a good standard of living. Attitudes on defense spending and spending on services were also used to comprise the 1996 issue index, and all four measures used to calculate the 2000 issue index were used to construct the 1984 issue index. Table 3.10 summarizes these attitudes. All four issues are measured on a five-point scale. Attitudes on all except spending on services lean to the right. Table 3.1 1 summarizes where respondents placed themselves ideologically, their calculated issue index, and then how they fell demographically. Ideology and the issue index are measured from 0 to l, and based on the attitudes respondents held on the four issues, it is not surprising that both ideology and the issue index average right of center. Table 3.12 shows the correlations between ideology and the issue variables, as well as, the issue index. In 2000 the correlations are not as strong as they were in 1996, but again when testing the data with the empirical model, we get the same results as those in 1976. 1984, and 1996. 46 Figure 3.11 shows the results from testing the model with two-stage least squares regression. In 2000, the relationship between ideology and issue attitudes is endogenous. Ideology has an affect on how people organize their attitudes, while the positions respondents take on issues also help them identify ideologically. It is important to point out, again, that the coefficient from issue attitudes to ideology is much stronger than the coefficient from ideology to issue attitudes. As an individual moves on the issue index from O to 1, they shift .77 on the ideology scale. As an individual moves ideologically from 0 to 1, they only move .17 on the issue index. Conclusions These results taken from 1976, 1984, 1996, and 2000 suggest that, at the very least, there is an endogenous relationship between ideology and issue attitudes. To this point the hypothesis that issue attitudes help individuals label themselves ideologically appears to be supported by the data. Additionally, while the models show a clear reciprocal relationship between ideology and issue attitudes, it is important to note that in all cases, the affect of issue attitudes on ideology is much stronger than the effect of ideology on issue attitudes. This suggests that where people place themselves on issues plays an important role in how they label themselves ideologically. Looking back at the 1976 data, we also see that changes in attitudes drive changes in ideology. Particularly, those who shifted ideological placement between 1972 and 1976 had changes in their ideological placement driven by their issue attitudes. This suggests that changes in issue attitudes Irave a strong impact on ideological self- placement. 47 It is also necessary to discuss the effectiveness of the control variables. Where people perceive their preferred parties to stand ideologically and on issues has a profound impact on where they place themselves on these two areas. The effects of these instrumental variables on ideology and issue attitudes are consistent and strongly significant throughout the four time periods. These results taken with the consistent endogeneity between ideology and issue attitudes lend support to the story being told. The data indicate that individuals are picking up cues from their preferred political parties and these help them take positions on issues and identify ideologically. And most importantly, their issue attitudes have a direct impact on how they identify ideologically. In the next chapter, I apply the model to data from 2004. I also test whether differences in education and how individuals conceptualize the political world change the relationship between ideology and issue attitudes. Does the extent to which an individual uses ideology as a way to conceptualize politics have an impact on the effect ideology has on their issue attitudes. Does the strength to which ideology impacts issues attitudes vary by level of education? And finally, I find some important differences between respondents in 2004 and those from the earlier years examined here. 48 Table 3.1. Summary of Issue Positions (1976) Variable Mean Standard N Devratron Jobs/Standard of Living 4.49 1.88 1261 Tax Rate 4.20 2.20 1316 Busing 6.08 1.63 1379 Healthcare 4.06 2.26 1235 Rights of the Accused 4.31 2.03 1291 Aid to Minorities 4.38 1.89 1280 Urban Unrest 3.35 1.86 1215 Source: l972-1976 American National Election Study Table 3.2. Descriptive Statistics (1972 and 1976) Variable Mean Standard N Devratron Ideology 72 .53 .21 933 Ideology 76 .55 .22 1092 Issue Index 72 .58 .21 1211 Issue Index 76 .57 .21 - 1508 Age 76 46.33 17.23 1518 Education 76 5.49 2.65 1504 Income 76 1 1.76 5.75 1426 Gender 76 (1=Female) .56 . 50 1523 Source: 1972-1976 American National Election Study Table 3.3. Correlations between Ideology and Issue Variables (1976) Variable Correlation with Ideology Issue Index .506 Job/Standard of Living .401 Tax Rate .129 Busing .343 Healthcare .405 Rights of the Accused .230 Aid to Minorities .349 Urban Unrest .342 Source: 1972-1976 American National Election Study 49 Table 3.4. Summary of Issue Positions (1984) Variable Mean Standard N Devratron Spending on Services 3.96 1.59 1866 Aid to Minorities 4.08 1.64 1944 Spending on Defense 3.99 1.63 1933 Russia 4.11 1.86 1864 Woman’s Role 3.82 1.69 1872 Jobs/Standard of Living 4.13 1.80 1918 Source: 1984 American National Election Study Table 3.5. Descriptive Statistics (1984) Variable Mean Standard N Devratron Ideology .57 .32 1834 Issue Index .50 .19 2182 Age 44.41 18.08 2237 Education 5.78 2.44 2243 Income 7.86 5.97 2025 Gender (1=Female) .56 .50 2257 Source: 1984 American National Election Study Table 3.6. Correlations between Ideology and Issue Variables (1984) Variable Correlation with Ideology Issue Index .335 Spending on Services .261 Spending on Defense .182 Jobs and Standard of Living .264 Aid to Minorities .208 Woman’s Role .226 Russia .229 Source: 1984 American National Election Study 50 Table 3.7. Summary of Issue Positions (1996) Standard Variable Mean . . N Devrat1on Spending on Services 4.1 1 1.51 1466 Abortion (1-4 Scale) 2.1 1 1.64 ' 1679 Spending on Defense 4.02 1.42 1481 J obs/ Environment 3.53 1.56 1461 Environmental Regulation 3.42 1 .63 1 372 Source: 1996 American National Election Study Table 3.8. Descriptive Statistics (1996) Variable Mean Standard N Devratron Ideology .56 .23 1329 Issue Index .45 .18 1709 Age 47.53 17.41 1712 Education 4.10 1.65 171 1 Income 1 1.67 6.66 1632 Gender (1=Female) .55 .50 1714 Source: 1996 American National Election Study Table 3.9. Correlations between Ideologyand Issue Variables (1996) Variable Correlation with Ideology Issue Index .538 Spending on Services .373 Spending on Defense .323 Abortion .357 Jobs/Environment .289 Environmental Regulation .367 Source: 1996 American National Election Study 51 Table 3.10. Summary of Issue Positions (2000) Standard Variable Mean . . N Devratron Spending on Services 2.72 1.10 1502 Aid to Minorities 3.46 1.24 1561 Spending on Defense 3.40 1.06 1421 Jobs/Standard of Living 3.45 1.35 1595 These issues are measured on a 5-point scale. Source: 2000 American National Election Study Table 3.11. Descriptive Statistics (2000) Variable Mean 8””?de N Devratron Ideology .56 .27 1623 Issue Index .56 .22 1774 Age 47.21 16.96 1798 Education 4.29 l .62 1800 Income 11.67 3.14 1595 Gender (1=Female) .56 .50 1807 Source: 2000 American National Election Study Table 3.12. Correlations between Ideology and Issue Variables (2000) Variable Correlation with Ideology Issue Index .342 Spending on Services .31 1 Spending on Defense .290 Aid to Minorities .231 Jobs/Standard of Living .212 Source: 2000 American National Election Study 52 Figure 3.1. Modeling the Relationship between Ideology and Issue Positions Ideology A :61 at V' Issue Positions Z A. 53 Party Placement of Ideology Demographics \ ,82_____ Party Placement of Issues Figure 3. 2. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (1976 Respondents) R2=.47 .080 Party Placement of Ideology (.000) Ideology A .326 .355 Demographics (.000) (.001) i . . .304 Party Placement of Issue Posrtrons (.000) Issues l R2=.38 l ,. Source: 1972-1976 National Election Studies Series Numbers in parentheses are p-values N=933 54 Figure 3.3. Modeling the Relationship between Ideology and Issue Positions (Lagged Variables and Party Placements as Instruments) Ideology 1A :61 a1 t Issue Positions fl3\ 55 Ideology (t-l) Party Placement of Ideology Demographics Party Placement of Issues \ / 2,8. [32/ Issue Positions (t-l) Figure 3.4. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions, l972-1976 Respondents (Lagged Variables and Party Placements as Instruments) R2=.55 Id 1 1 . t- .321 / 60 08M ) Ideology A .066 (.000) \ Party Placement of Ideology 167 354 Demographics (.000) (.000) Party Placement of Issues 1 Issue Positions Issue Positions (t-l) —_—l R2=.53 Source: 1972-1976 National Election Studies Series Numbers in parentheses are p-values N=687 56 Figure 3.5. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions 1972-1976 Respondents (Individuals Who Shifted Ideological Identification at least Two Positions) R2=.42 Ideology (1972) -.347 // Ideology A Party Placement of Ideology .079 .613 Demographics (.368) (.007) Party Placement of Issues 17 Issue Positions Issue Positions (1972) R2=.44 Source: [972-1976 National Election Studies Series Numbers in parentheses are p-values N=128 57 Figure 3.6. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions l972-1976 Respondents (Individuals Who Shifted their Ideological Identification One Position or Less) R2=i78 Id 1 ( 1972) .694 / 6" ogy Ideology ‘ Party Placement of Ideology .137 .146 Demographics (.001) (.003) Party Placement of Issues 1 Issue Positions Issue Positions (1972) R2=.55 Source: 1972-1976 National Election Studies Series Numbers in parentheses are p-values N=559 58 Figure 3.7. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions 1972-1976 Respondents (Individuals Whose Issue Index Changed at least .2 Points) R2=.56 251 / Ideology (1972) Ideology (.000) A .081 (.000) \ Party Placement of Ideology 314 1 6 1 Demographics (.019) (.364) Party Placement of Issues V Issue Positions Issue Positions (1972) R2=.30 Source: 1972-1976 National Election Studies Series Numbers in parentheses are p-values N=135 59 Figure 3.8. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions l972-l976 Respondents (Individuals Whose Issue Index Shifted Less than .2 Points) R2=.56 341 / Ideology (1972) Ideology (.000) A . .064 (000) \ Party Placement of Ideology 073 362 Demographics (032) (000) l L Issue Positions l (.000) \ Party Placement of Issues Issue Positions R2=.75 (1972) Source: 1972-1976 National Election Studies Series Numbers in parentheses are p-values N=552 60 Figure 3.9. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (1984 Respondents) R2=.24 .047 (.000) Party Placement of Ideology Ideologv A .250 .457 Demographic (.000) (.003) I V , , ‘ Party Placement of ssue Posrtrons .313 Issue Positions (.000) R2=.23 Source: 1984 American National Election Study Numbers in parentheses are p-values N=1515 61 Figure 3.10. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (1996 Respondents) Party Placement of Ideoloav R2=.42 .050 (.000) Ideology ‘ A .134 .820 Demographic (.018) (.000) V . . ‘ Issue Posrtlons .394 (.000) R2=.39 Source: 1996 American National Election Study Numbers in parentheses are p-values N=1223 62 Party Placement of Issue Positions Figure 3.11. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Issue Positions (2000 Respondents) RT” .064 ‘ “000) Party Placement of Ideology Ideologv A .168 .769 Demographic (.002) (.000) y _ , ‘ Party Placement of Issue Posrtrons .368 Issue Positions (.000) Source: 2000 American National Election Study Numbers in parentheses are p-values N=1223 R2=.22 63 Chapter 4 Differences in 2004 Jacoby (2006) notes that campaign rhetoric during the 2004 election did not focus primarily on ideological themes. However, people were reasonably able to identify the ideological positions of the presidential candidates. In his study, slightly less than two-thirds of respondents were able to place Bush, Kerry, and the two parties at their appropriate positions along the liberal/conservative continuum. This suggests that people can recognize ideological labels and can effectively use them to describe politics. However, this does not indicate that individuals use ideological labels as more than a label. While Jacoby finds that voters can accurately locate candidates on ideological lines, and perceptions of the candidates seem to be structured around ideological concerns, he shows that ideology was not an important indicator of vote choice in 2004. Jacoby concludes, “. . .people do not take the next step, and make a direct connection between abstract ideological positions and the concrete choice between two candidates” (13). This meshes well with the narrative here. People can accurately use ideological labels to describe candidates and parties, and look to the labels associated with their preferred candidates and parties to inform their own ideological labels. But for some, this use of ideology does not go any further. Ideology did not directly affect vote choice in 2004, suggesting that while voters used ideological labels to characterize candidates and parties, it did not have an impact on their own choices. Therefore, it is important to ask if ideology failed to predict their issue attitudes in 2004, as well. 64 Jacoby (2006) suggests that it did. His results, from an OLS regression analysis, show that ideology had a strong impact (even stronger than partisanship) on issue attitudes, in 2004. This suggests that “people do employ ideological considerations in the political reasoning” (l2). Ideology affects issue attitudes and issue attitudes affect vote choice, indicating ideology indirectly affects vote choice through its influence on issue attitudes. Unfortunately, J acoby failed to test directly whether issue attitudes had a reciprocal effect on ideological placement. It could be thatjust as voters were using ideological labels to characterize candidates and parties, they were using them similarly to characterize their issue attitudes. As ideology had no direct effect on vote choice, perhaps it has no direct effect on issue attitudes, as well. The relationship between ideology and attitudes discovered in Jacoby’s model, might be due to correlation between issue attitudes and ideology, but without a direct test for endogeneity, we cannot tell if it is ideology helping structure issue attitudes or ideology merely serving as a label for attitudes already formed. This can be tested with the basic model discussed in Chapter 3. Figure 4.1 displays the model incorporating the above-mentioned variables in an explanation for how the relationship between partisanship, ideology, and issue positions works. Within the model, ideology is a function of several demographic variables and the individual’s positions on current political issues, as well as where they perceive their preferred party to stand ideologically. Their issue positions are a function of the same demographic variables, where they place themselves ideologically, and where they perceive their preferred party to stand on those same issues. 65 The model produces two equations. The first equation models the link from ideology to issue positions: EQ 4.1. IssuePosit ions ,- = '60 + ,6] Ideology i + ,82 PartyPIace mentIssues i + Z 16k Demographi csl- + £1 The dependent variable in this first equation, IssuePositions; is an index (or average) of the positions taken by individual i on several political issues. PartyPlacementIssues is an index of where individual i places his or her preferred party on those same issues. Ideology is the individual’s self-placement on the seven-point ideological continuum, and Demographics represents a number of socio-demographic variables that should have an effect on an individual’s policy positions. These include race, gender, income, age, and education. What we are most concerned about here is the coefficient ,6, , which measures the effect that ideology has on an individual’s issue positions. If this coefficient is found to be positive and statistically significant, we can conclude that ideology has an empirical effect on issue positions. The second equation captures the link from issue positions to ideology: EQ 4.2. Ideology,- = a0 + a] IssuePositi()ns,- + a2 PartyP/acementldeology,- + X a k Demographics, + 6‘2 66 where, individual i ’s ideology (Ideology) is a function of that individual’s index summary of issue positions (IssuePositions), where they place their preferred party ideologically (Party/Placementldeology), and a number of socio-demographic variables (Demographics) as controls. Here we are concerned with a, , the effect of issue positions on ideology. If this coefficient is positive and statistically significant, we can conclude that issue positions have an empirical effect 011 ideological self-placement. Specification, Data, and Methods The model is non-recursive due to the reciprocal link between ideology and issue positions. The variables that capture the perceptions of where the parties stand ideologically and on issues act as instruments to identify the model. As was the case in the previous chapter, a Hausmann test indicates that the instrumental variables are not correlated with the error terms in the dependent variables. To test this model, data are drawn from the 2004 American National Elections Study. To accurately compare the issue index with its instrument (party placements on issues), it is necessary to use attitudes on issues where the respondent is also asked to locate where they believe their preferred party stands. The 2004 ANES has seven such questions covering the policy areas of (1) spending on government services, (2) defense spending, (3) government assistance in providing jobs and a good standard of living, (4) government assistance to blacks, (5) the role of women in society, (6) abortion, and (7) whether or not the United States should use diplomacy or force to deal with international crises. All of these variables, except abortion, are measured on a seven-point scale, with higher numbers associated with more “conservative” responses. Abortion is measured 67 on a four-point scale, but has been re-scaled to fit with the other issue variables. All the issue variables have been re-scaled form zero to one. They were then combined into an additive index and averaged to get an individual’s mean score across all seven issue variables. This index is then re-scaled from zero to one, with higher numbers indicating an individual takes more conservative issue positions, on average. Results Looking at Table 4.1, we can see the distributions of the responses to the seven issue related questions. The means for most of the items are near the median value, leaning slightly to the conservative side. Attitudes about the role of women are skewed to the left, with the View that the role of men and women in society should be equal, dominating responses to this question. Looking at Table 4.2, the mean of the issue index is .45, which is near the center but slightly to the left“). Ideology is operationalized by the standard seven-point ideological continuum. Table 4.2 shows the distribution of the responses to the ideology question. Clearly it appears more difficult for respondents to locate themselves on an ideological continuum than it is for them to give their attitudes on particular issues. 76% of possible respondents were able to place themselves ideologically. 88% were able to place themselves on at least one of the issues. The mean of ideology is .54, once again near the center, but leaning slightly to the conservative side. Table 4.3 shows the correlations between ideology and the separate issue variables, as well as the overall issue index. The issue index is highly correlated (.61) with ideology, so there should be a strong relationship between ideology and the issue '0 The Chronbach’s Alpha score for the scale is .66, which is moderate, but within the acceptable range >60. 68 index in a multivariate model. This also lends support to the argument that those who are able to place themselves 011 the ideological continuum can do so fairly accurately and are consistent with the positions they take on issues. A Hausman test shows there is reason to believe simultaneity exists within the model. Therefore, it is appropriate to use two-stage least squares regression to estimate the effects of the model. The results are displayed in Figure 4.2I '. Looking at the figure, the R2 for the two equations shows the model is explaining over 40% of the variance in both equations. This suggests that the model is explaining a good bit of information regarding the relationship between issue positions, partisanship, and ideology. What is most important is the issue index side of the model. Here, the coefficient for ideology ( ,B,) is small and insignificant, indicating that when taking into account a reciprocal link between ideology and issues, ideological self-placement does not predict the positions individuals take on issues. However, when looking at the ideology side of the model it appears that issue positions have a large and significant effect on where individuals place themselves ideologically. The instrumental variables also work in the predicted ways. Both show that where individuals place their preferred parties on issues and ideology influences where they locate themselves on those two measures. These findings correspond well with what J acoby (1988) finds, suggesting that partisanship works to influence issue positions and ideology through where individuals perceive their parties to be located ideologically and on issues. Taken together, the results here show that perceptions of where the parties stand on issues help individuals form their own issue preferences. This supports ” Complete tables, including the effects from demographics. for analyses in Chapter 3 can be viewed in Appendix B. 69 the reference group argument that parties serve as reference points to individuals, providing cues as to what the appropriate positions are on particular issues. Once these issues positions are formed, the individual also having picked up cues as to where their party stands ideologically. are able to then label themselves with the appropriate ideological label. These results suggest that in 2004, once the reciprocal nature of the relationship is considered, ideology no longer predicts where people place themselves on issues. Instead, it appears that how people organize their issue positions determines where they place themselves ideologically. At first glance, the empirical results appear to show that for the majority of the public, the conventional wisdom that ideology informs issue positions, is wrong. However, I have argued earlier the possibility of considerable variance between groups in the population. The people who are most likely to use issue positions to form their ideological self-placements should be those who use their issue positions to organize the political world in general. One way to identify these people is to locate those individuals who use issues to evaluate political parties and candidates. It may be that evaluating the entire sample in the original model is overlooking important differences among sub-groups. Finding that this result holds across a ntunber of different specifications for the model, would add considerable weight to the conclusions that can be drawn from it. In the next section, I explore how these results may vary depending on mediating factors such as how individuals conceptualize the political world, whether it be through issues or ideology and their levels of education. 70 Jacoby (1991) finds that there are important differences in how individuals use ideology to organize their belief systems. The two most important factors that define these differences are conceptualization and education. Those with higher levels of education should be more likely to use ideology to structure their political beliefs. Those with lower levels of education should be less likely to use ideology as an organizing tool, and consequently should have less organization across their political attitudes. Additionally, those who conceptualize the political world through ideology should be more likely to have ideology influence their issue attitudes. Those who conceptualize the political world through their issue attitudes should be more likely to have issue attitudes influence their ideological identification. Conceptualization and Education ’ In this context, conceptualization refers to the levels of conceptualization introduced in The American Voter. Individuals were classified into different levels depending on how they appeared to conceptualize the political world, specifically political parties and candidates. They were divided into five groups: ideologues, near ideologues, group benefits, nature of the times, and no issue content. Respondents were asked to give reasons why they liked or disliked a particular candidate or political party and their responses were used to classify them into one of the five groups. The categories are considered ordered according to the degree of abstraction an individual uses to describe political objects implied by each category’s defining characteristics. In other words, an individual who uses more abstract ideological concepts to describe their political world, the higher they are placed in the categories. The lower individuals are 71 placed, the more their responses are confined to narrow and idiosyncratic evaluations of political objects (Jacoby 1991). Important for this analysis, is the group labeled ideologues. Ideologues were those who showed they interpreted politics ideologically. In other words, they not only used ideological terms to describe the candidates or parties, but they could wrap these terms into an appropriate issue based context. In short they could use a specific ideological label and show they knew what it meant. Jacoby (1991) finds that ideologues display a distinctive ideological orientation in their political beliefs. Thus, issues attitudes should be strongly tied to ideological orientations, especially for those who think about politics in ideological terms. However, the concentration of ideologues within the American public has been found to be quite small. The authors of The American Voter were only able to classify 12% of the population as such in 1956, while J acoby (1991) determined that 19% fit within the category. The analysis in the previous section showed that on average ideology did not help individuals in 2004 organize their issue attitudes. But, by considering the entire sample as a whole, the empirical test could be hiding differences in that relationship for a small group such as the ideologues. Based on prior evidence there should be differences in how ideologues organize their issue preferences compared to other groups. Before we can make any definitive conclusions, this hypothesis must be checked. To do this, a second group who frame their likes and dislikes towards the parties and candidates in terms of issues, are included along with the ideologues in the analysis. This coding varies from the original levels of conceptualization coding in a two important ways. First, the original levels were coded using open-ended responses to the likes and dislikes questions. This allowed the researchers to view the actual context and discussion the individual gave regarding their answers. However, the 2004 NES data does not provide the open-ended responses. Instead the answers are placed into a number of different categories based on the reading of the original responses.” Second, the original levels of conceptualization did not allow for an issue- focused group. However, for the purposes of testing the empirical model presented here, it seems appropriate to specify a group who thinks about the political world in terms of issues. The theory posited argues that there is a large segment of the population who thinks of politics in terms of issue positions and use those issue positions to organize and label beliefs such as ideology. We should also expect those individuals to evaluate candidates and parties based on these same issues. Therefore, it is important to distinguish between those individuals who organize their political beliefs around ideological concepts and those who use their issue attitudes instead. For this purpose, those who provided responses to the candidate and party likes and dislikes questions that predominantly focused on particular issues were included in the analysis. The model was tested separately for both groups using two-stage least squares regression. The results are displayed in Figures 4.3 and 4.4.13 Figure 4.3 shows the results for the issue-focused respondents. There is little change between these results and those obtained from the entire sample. The variance explained in the dependent variable for each equation is still around 40%, and similar. to the previous analysis, the '2 A list of these categories can be obtained from the author. '3 Those individuals who placed an ideological label (liberal or conservative) at higher than 80% and the other ideological label at lower than 20%, on their respective feeling thermometers were also included in the ideologically focused category. 73 issue index is a strong predictor of ideology, but ideology does not predict the issue index. The surprising result is in Figure 4.4. For those who give predominantly ideological references for candidates and parties, the ideology equation is the strongest of all specifications of the model. The model explains over 80% of the variance in ideological self-placements and the coefficient for the issue index is positive and significant. However, the coefficient for ideology in the issue index equation fails to reach significance, so we cannot reject the null hypothesis that ideology does not predict issue positions. Even when conceptualization is taken into account, it appears that those who would be considered ideologues are still not organizing their issues around their ideology. In fact, from these results we can infer the relationship works in reverse. Perhaps this finding is due simply to differing levels of education in the sample. J acoby (1991) shows that conceptualization and education are distinct and should be considered as separate in a model explaining issue attitudes. He finds that ideology has a stronger impact on issue attitudes for those at higher levels of education. However, in his 2006 piece, J acoby finds that education had no impact on the degree to which ideology influenced vote choice. To test for educational effects, the sample is divided into three groups, those who never completed high school, those with only a high school education, and those with at least some college education. The same model is tested on all three groups using two-stage least squares regression. The results are shown in Figures 4.5, 4.6 and 4.7. The only important difference between the model with all respondents and these three is that for those individuals with less than a high school education (Figure 4.5), 74 nothing in the model predicts their ideology or issue positions. In fact, the R2 for both equations is .00, indicating that the variables used explain absolutely none of the variance in the dependent variables. This should come as no surprise. Those with low levels of education have a much more difficult time organizing their political world. It should not be a shock to find that constructing any model that might predict consistency in their attitudes would be very difficult. What is most striking about these three figures is the relationship between ideology and issue positions for the most educated (Figure 4.7). The college educated display little difference in how their attitudes are formed from the entire sample. In the ideology equation for the college educated, not only is the coefficient for issue attitudes still positive and significant, but the explained variance for that equation is increased by over 15%, from the test using the entire sample. Looking at the issue equation, ideology is insignificant, showing little reason to assume that ideology helps the most educated organize their issue positions. The results from looking at differing levels of education suggest that even the most educated do not use their ideology to formulate their issue attitudes, but instead, use their issue attitudes to structure an ideology. Looking at Table 4.4, for the most educated, the correlation between ideology and issue attitudes is .72. This suggests that the most educated show a high level of consistency between where they place themselves ideologically, and the positions they take on issues. This should be no surprise, and follows strongly from Jacoby’s work on ideological thinking. However, the findings here suggest that even the most educated do not use their ideological orientation as a tool to help them take positions on issues. Instead, the results imply that 75 they, just like everyone else, use their issue positions to locate themselves ideologically, but they do this more accurately than the less educated. Conclusions The findings in this chapter suggest several important implications for the study of mass political attitudes. First, they clearly demonstrate that for a large portion of the public, in 2004, ideology did not preclude opinions on issues. It is important to consider a causal link from issue positions to ideology, and the link holds even when controlling for levels of conceptualization and education. Individuals take positions on issues and use the ideological labels they learn from political elites, then categorize themselves as “liberal” or “conservative” without having a full understanding of what those labels mean. This is not to say that levels of conceptualization and education do not have an effect, indeed they do. The models for the most educated and those focused on ideology, when thinking about candidates or parties, show quite convincingly that these individuals are much more consistent in matching their ideological labels with their issue positions. The issues attitudes for both groups are highly correlated with their ideological self-placement, more so than any other group in the study. And, the combination of issue attitudes and control variables in the model, explains over 80% of the variance in ideology for the ideologically focused individuals. Second, the results mesh well with Jacoby’s (2006) results from 2004. As he finds that ideology does not directly influence vote choice, I find that ideology does not influence issue attitudes, as well. Additionally, neither analysis shows any impact from education on the ability of an individual to use ideology to make political decisions. The 76 key difference is that Jacoby did not test for endogeneity between ideology and issue attitudes. Had he done this, he might have found the same results as presented here. Third, these findings give some support to work that has used a scale of issue positions as a proxy for what is termed an “operational ideology” (Stimson 2004). Stimson suggests that an individual’s true ideology is their operational ideology, which is made up of their positions on issues that are important to them. The evidence here suggests that this is 30. Most individuals have a limited view of ideology (even though they may be able to locate themselves on the continuum), but this ideology is only understood through the positions they take on issues. This may be where the similarities end, however. Stimson argues that there is a subset of the population, he terms “conflicted conservatives” who take relatively liberal positions on policy, but place themselves on the conservative end of the ideological continuum. A quick correlation between ideology and the issue index shows little difference between the relationship between ideology and the issue index for conservatives (correlation of .39) and for liberals (correlation of .41). Running the full model for conservatives and liberals also produces the same results for both groups as the other model specifications listed above”. Fourth, the results here create a puzzle that perhaps current and future literature can solve. If the relationship 1 describe here holds, and people are truly using their issue positions to develop their ideologies, then what happens when a new issue comes along that most of the public had not taken a stance on? Kuklinski, Metlay and Kay (1982) '4 The reliability ofthese results could be questioned for issues of selecting on the dependent variable. For tables that contain the results of these estimations. please contact the author. 77 argue that knowledgeable citizens draw heavily on ideology to inform their analysis of new issues, while unknowledgeable citizens draw on cues from groups. Here the authors attempt to predict voting decisions, which eliminates the problem of endogeneity. There is no question about which came first, the issue position or the vote. Presumably, the attitude toward the issue must preclude the decision to vote on it. An individual’s personal ideology also precludes the vote, and in their model, the most knowledgeable individuals use their ideology as a cue to take a position on the issue. But, all issues are new to an individual at some point, and this creates a conundrum. If they are using ideology to organize new issues, then why does it appear that as these issues become old hat, they begin to shape ideology? One answer is that they are indeed using groups rather than ideology to organize their issue positions. This would be consistent with the party element in the model presented in this paper. However, Kuklinski, Metlay, and Kay (1982) argue this holds for only the less knowledgeable individuals. So, this does not explain the similar findings for levels of education in this study. At this point, I do not have an answer for this problem, but it should merit further study. Finally, conceptualizing the relationship between ideology and policy positions as reciprocal challenges the traditional view that ideology is a predictor of issue positions and should be considered a standard control variable when modeling attitudes on issues. This chapter shows that considering a reciprocal relationship, when conventional wisdom has held uni-directional causality, brings to light a new empirical connection that has been overlooked. 78 Table 4.1. Summary of Issue Positions Variable Mean Seeded N Devratron Spending on Services 4.52 1.59 1060 Spending on Defense 4.57 1.48 1061 Jobs/Standard of Living 4.21 1.87 1 103 Assistance to Blacks 4.54 1.79 1073 Woman’s Role 1.92 1.47 1 157 Abortion 2.79 1.09 1047 Diplomacy vs. Military Force 3.86 1. 7 5 999 Source: 2004 American National Election Study Table 4.2. Descriptive Statistics Variable Mean 8’2”}de N Devratron Ideology .54 .25 920 Issue Index .45 .17 1210 Age 47.27 17.14 1212 Education 4.30 1.61 1212 Income 14.94 6.00 1070 Gender (1=Female) .53 .50 1212 Source: 2004 American National Election Study Table 4.3. Correlations between Ideology and Issue Variables Variable Correlation with Ideology Issue Index Spending on Services Spending on Defense Jobs and Standard of Living Assistance to Blacks Woman’s Role Abortion Diplomacy vs. Military Force .610 .417 .377 .433 .372 .306 .348 .420 Source: 2004 American National Election Study 79 Table 4.4. Correlations between Ideology and Issue Positions for Different Groups Group Correlation All Respondents .61 Issue Focused People .59 Ideology Focused People .82 No High School Diploma .32 High School Educated .52 College Educated .72 Conservatives .39 Liberals .41 Source: 2004 American National Elections Study 80 Figure 4.1. Modeling the Relationship between Ideology and Issue Positions Ideology fl. V Issue Positions Party Placement of Ideoloav Demographic 81 Party Placement of Issue Positions Figure 4.2. Modeling the Relationship between Ideology and Issue Positions (All Respondents) R2=.46 .036 ‘ (.000) Party Placement of Ideology Ideoloav .057 .965 Demographic (.536) (.000) v . . ‘ Party Placement of Issue Posrtrons .494 Issue Positions (.000) R2=.41 Source: 2004 American National Election Study Numbers in parentheses are p-values N=819 82 Figure 4.3. Modeling the Relationship between Ideology and Issue Positions (Issue- Focused Respondents) R2=.46 .042 (.000) Party Placement of Ideology K Ideoloav A .003 .846 Demographic (.975) (.000) V Issue Positions + Party Placement Of .432 Issue Posrtrons (.000) R2=.39 Source: 2004 American National Election Study Numbers in parentheses are p-values N=468 83 Figure 4.4. Modeling the Relationship between Ideology and Issue Positions (Ideology-Focused Respondents) R2=.83 .060 (.000) Party Placement of Ideology f Ideoloav A -.073 .991 Demographic (.841) (.000) Y . _ ‘ Party Placement of Issue Posrtrons .716 Issue Positions (.062) R2=.52 Source: 2004 American National Election Study Numbers in parentheses are p-values N=97 84 Figure 4.5. Modeling the Relationship between Ideology and Issue Positions (Respondents with No High School Diploma) Party Placement of Ideology R2=.00 _.050 (.330) Ideology ‘ a -.876 2.262 Demographic (.549) (.254) i . . ‘ Issue Posrtrons .727 (.354) R2=.00 Source: 2004 American National Election Study Numbers in parentheses are p-values N=52 85 Party Placement of Issue Positions Figure 4.6. Modeling the Relationship between Ideology and Issue Positions (Respondents with High School Diploma but No College) R240 .042 ‘ (.000) Party Placement of IdCOIOBY Ideologv a .076 .921 Demographic (.460) (.000) Y . . ‘ Party Placement of Issue Posrtrons .461 Issue positions ( .000) R2=.35 Source: 2004 American National Election Study Numbers in parentheses are p-values N=377 86 Figure 4.7. Modeling the Relationship between Ideology and Issue Positions (Respondents with at Least Some College Education) 2: R ~61 .041 ‘ (.000) Party Placement of Ideology Ideoloav -.045 .892 Demographic (.778) (.000) Y . . ‘ Party Placement of Issue POSltIOHS .580 Issue Positions (.000) R2=.40 Source: 2004 American National Election Study Numbers in parentheses are p-values N=390 87 Chapter 5 Same-sex Marriage The current debate over same-sex marriage began in Hawaii during 1993. The Supreme Court of Hawaii, in Baehr v. Lewin, reversed a lower court’s finding that discrimination of same-sex couples is allowed under the state constitution’s equal protection clause. Additionally, in Baehr v. Miike (1996), a trial court held that there was no compelling state interest in not allowing same-sex couples to marry. However, this was overturned in 1998 by a state constitutional amendment, which reserved the power to grant same-sex marriages to the state legislature. One of the effects of the Hawaii case was that it brought to the forefront the question of marriage as part of the gay rights debate and, in general, the broad move for civil rights. Since the Hawaii decision, several states have ratified constitutional amendments that have banned same-sex marriages. In 2004, voters in Arkansas, Georgia, Kentucky, Michigan, Mississippi, Montana, North Dakota, Ohio, Oklahoma, Oregon and Utah approved amendments outlawing same-sex marriage by double-digit margins. In Mississippi, the amendment passed with 86% of the vote. Between 1993 and 2004, 31 states passed legislation prohibiting same-sex marriage and 33 states passed legislation that prohibited the recognition of same-sex marriages performed outside the state. Many other states had already explicitly defined marriage as being between a man and a woman. The Massachusetts case in 2003 (Goodridge v. Department of Public Health), has been noted as a keystone in the fight for the rights of homosexuals. However, in 1999 the Vermont Supreme Court decided in Baker v. State, that the state legislature should 88 provide equal benefits of marriage to same-sex couples. Following the Massachusetts decision, in April of 2005, the Connecticut legislature created civil unions for same—sex couples, and in December of 2006, New Jersey passed a law creating civil unions. The proximity of the Massachusetts case to the 2004 election created a backlash across the country, especially among religious/social conservatives, who saw state constitutional amendments banning same-sex marriage as a way to circumvent court decisions allowing the practice. This coincided with a move by opponents of same-sex marriage to pass a federal constitutional amendment banning same-sex marriage nationwide. This measure ultimately failed, not obtaining the required two-thirds vote in both houses of Congress. With the publicity surrounding the Massachusetts case, along with the move in several states to pass constitutional amendments, same-sex marriage became an important issue in the 2004 presidential elections. Both candidates were on record opposing same-sex marriage and supporting civil unions. However, Bush supported the federal constitutional amendment, while Kerry opposed it.15 The salience of same-sex marriage during the 2004 Presidential Election is important, because it provides a unique issue to examine in the context of the discussion of this paper. This chapter asks, to what extent does the endogeneity between ideology and issue attitudes, hold up when tested against the issue of same-sex marriage during the 2004 Presidential Election? Stoutenborough, Haider-Markel, and Allen (2006) suggest that ideology does influence opinions on same-sex marriage. While testing the impact of court decisions on public opinion, they find that levels of conservatism and liberalism influence opinions on '5 Although Kerry did not oppose state constitutional bans on same-sex marriage (Roberts and Gibbons 2004). . 89 same-sex marriage. In their model, conservatives were less likely to support same-sex marriage than were liberals, and were less likely to have their opinions affected by a court decision involving same-sex marriage. Support for gay rights, in general, has been associated with liberalism and the Democratic party, while opposition to gay rights has come from conservatives and the Republican party (Haeberle 1999; Lewis and Rogers 1999; Wilcox and Norrander 2002). However, Brewer (2003) finds that shifts in ideology do not affect changes in attitudes towards homosexuals. The public’s attitudes about homosexuals changed considerably during the 19903, becoming more positive. As Brewer notes, when asked whether they believed sex between two people of the same sex is wrong, “In 1992, 71% of the General Social Survey Respondents chose the ‘always wrong’ option; in 1994, 63% did; by 1998, only 54% did” (2003, 1208). His conclusion is that much of this change in opinion was driven by changes in “feelings” towards homosexuals. People became more positive towards homosexuals as a group, which made them more likely to support policies aimed at increasing gay rights. On one hand, it should be intuitive that ideology plays an important role in the attitudes individuals have on same-sex marriage. The movement in opposition to same- sex marriage has been characterized as one of the conservative religious right, and as noted above, research has found that levels of conservatism and liberalism influence attitudes on gay rights. However, it is important to note that Brewer (2003), when controlling for feelings towards homosexuals, finds no impact of ideology on attitudes towards gay rights. This suggests that attitudes towards same-sex marriage are driven more by affect than they are by ideological orientations. 90 This can be tested with a slight adjustment to the model presented in Chapter 3. Figure 5.1 depicts the original model with modifications in the instrumental variables. Instead of employing the party placement variables as controls, the model incorporates feelings towards homosexuals and ideological groups. This will directly test the extent to which ideology has an effect on attitudes towards same-sex marriage versus feelings towards homosexuals as a group, and also test the argument that issues affect ideological self-placement. Below are the relevant equations for the model: EQ 5.1 ,Ideologyi = a0 + alGayMarriageI- + azldeoFeeli + a3Partyidi + Z a k Demographics i + 81 EQ 5.2 GayMarriagei .-= flo + '61 Ideology, + flzGayFeeli + ,83 Partyidi + Z ,Bk Demographicsl- + .92 where Ideology is a function of an individual’s position on same-sex marriage (GayMarriage), an individual’s feelings towards their preferred ideological group (IdeoF eel), their partisan affiliation (Partyid), and their demographic characteristics (race, gender, education and income). An individual’s position on same-sex marriage, is a function of their ideological orientation, their feelings towards homosexuals as a group (GayF eel), their partisanship (Partyid), and their demographical characteristics. The variable that measures feelings towards homosexuals not only allows a test of the effect feelings towards homosexuals have on attitudes towards policies to advance gay rights, but also serves as an effective instrumental variable. There is no reason to 91 believe that how an individual feels about homosexuals influences their ideological orientation. Additionally, the variable IdeoF eel serves the same purpose; how an individual feels about liberals or conservatives as a group,'should not affect their attitudes towards same-sex marriage. In fact, in the 2004 data, the correlation between feelings towards homosexuals and ideological placements is .05, and the correlation between attitudes towards same-sex marriage and feelings towards ideological groups is -.30. Data, Specification, and Methods To test the model, data are again drawn from the 2004 American National Election Studies. Respondents were asked to provide their opinions on same-sex marriage. They chose from three different responses, (1) same-sex couples should be allowed to marry, (2) same-sex couples should not be allowed to marry, but should be allowed to legally form a civil union, and (3) same-sex couples should not be allowed to marry.'6 Responses to this question were re-scaled from 0 to 1. Table 5.1 shows the break down of responses to this answer, in 2004. A strong majority of respondents were against allowing same-sex couples to marry, while a surprisingly small amount of respondents were willing to allow civil unions for same-sex couples as long as marriage was not allowed. Ideology is measured on the typical seven-point scale with extremely liberal at the low end (1) and extremely conservative on the high end (7). This variable was also re-scaled from 0 to 1. Partisanship is measured on a six-point scale with strong democrat on the low end of the scale (1), and strong republican on the high end (6). Each respondent was also asked to rank how favorable he or she felt towards '6 The exact wording ofthis question can be found in Appendix C. 92 homosexuals as a group. This feeling thermometer is scaled from 0 to 100, with favorable ratings at the high end and not favorable ratings at the low. They were also asked to place liberals and conservatives on this same scale. The placements of liberals and conservatives were combined to form a single scale measuring how favorable the respondent rated the ideological group they personally identified with. For example, the IdeoF eel score for an individual who self-identifies as a liberal is the place they ranked liberals on the feeling thermometer. To create this measure for moderates, the positions where they ranked liberals and conservatives on the feeling thermometer were averaged to get a single score. Results The model is estimated using two-stage least squares regression. The results are in Figure 5.2. As was the case in the data from 1972-2000, we find that the relationship between ideology and attitudes on same-sex marriage is endogenous. The coefficients for both the link from ideology to the issue attitude, and from the issue attitude to ideology are strong and significant. Additionally, the instrumental variables act as expected. Feelings towards homosexuals have a significant impact on attitudes towards same-sex marriage, and feelings towards one’s preferred ideological group influences self-placement on the ideological continuum. Finally, party identification appears to have an impact on ideological identification, but not on attitudes toward same-sex marriage. This suggests that attitudes on same-sex marriage are driven primarily by feelings towards homosexuals (Brewer 2003) and ideology. Ideological identification appears to be driven by feelings towards liberals or conservatives, and individual positions on same-sex marriage. 93 This might be considered a “hard case” for the theory that issue attitudes influence ideological self-placement. A single issue, even a particularly salient one, should be less likely to have a strong influence on an individual’s ideological identification than a combination of issues. And yet, the data shows that attitudes on same-sex marriage have a significant effect on ideological placement. However, the findings here do not duplicate the results from the model of all 2004 respondents. Ideology still has an impact on same-sex marriage attitudes, while it failed to predict the summary index of the attitudes used in Chapter 4. Why might this be the case? A simple answer is that same-sex marriage is much more of an ideological issue than the seven issues that make up the issue index in Chapter 4. Opposition to same-sex marriage has been explicitly tied to religious/social conservatism, especially in the media.17 In accordance, there may be differences between liberals and conservatives. It could be that the ideology is driving attitudes on same-sex marriage for those who identify as conservative, because of the ideological nature of the issue, and how it has been used by conservative groups to promote a conservative agenda. The answer to this question is beyond the scope of this project. However, future plans involve attempts at addressing this question directly. Attitudes on the Rights of Homosexuals To get a better grasp on the relationship between ideology and attitudes towards homosexuals, I turn to general attitudes on the rights of homosexuals. The 2004 ANES asked respondents to provide their opinions on three issues dealing with the rights of homosexuals, other than the issue of same-sex marriage. First, individuals were asked to indicate the degree to which they felt the government should pass laws to protect '7 See for instance, Jefferson (2006), Sullivan (2003), and Cooperman (2004). 94 homosexuals against job discrimination. Second, they were asked to indicate whether they thought homosexuals should be allowed to serve in the military. Finally, they were asked if they felt homosexual couples should be allowed to adopt children. Responses to these three questions, along with the question on same-sex marriage, were combined into an additive index measuring attitudes towards the rights of homosexuals.18 This index is scaled from 0 to 1 with right leaning answers at the higher end. The index was then substituted into the model for attitudes on same-sex marriage, so that the model now looks like Figure 5.3. The model was estimated using two-stage least squares regression. The results are displayed in Figure 5.4. The results look very similar to those found when estimating the model on attitudes towards same-sex marriage. Again, the relationship between ideology and attitudes on rights for homosexuals is endogenous. Ideology is impacting attitudes and attitudes have an effect on ideology. The instrumental variables work as expected. Individuals with more positive feelings towards homosexuals are more likely to favor rights for that group. Feelings toward one’s preferred ideological group also impact ideological placement. Finally, partisanship has an impact on ideology and attitudes on rights for homosexuals. Republicans are less likely to favor rights for homosexuals and are more likely to identify as conservative. Discussion The empirical results from looking at same-sex marriage are straightforward. The relationship between ideology and attitudes on same-sex marriage is not unidirectional. For some, their belief about same-sex marriage is dictating, to a degree, '8 The actual question wording for these items, and the method for computing the index measure of attitudes on rights for homosexuals can be found in Appendix C. 95 how they label themselves ideologically. For these individuals, their attitudes on same- sex marriage are influenced by how they personally feel about homosexuals, and then this reflects on their ideological orientation. For others, attitudes on same-sex marriage are driven by ideology. These results do not directly mimic the findings from the 2004 ANES respondents at large, but do appear similar to the results from 1976-2000. When breaking down the issues to the single issue of same-sex marriage, the relationship between ideology and the issue attitudes is endogenous. To some extent, this should be expected. It should be difficult for one single attitude to have a strong influence on an ideological orientation. Attitudes on same-sex marriage should not explicitly define whether someone considers themselves liberal or conservative. It should have an impact, but not a strong one, and that is what the analysis shows. Instead, the theory argues that individuals look at a collection of attitudes and this helps them select an ideological label. This is what we find when considering attitudes across a number of different issues, at the same time. However, this still does not explain why ideology has a strong impact on attitudes towards same-sex marriage, especially since when analyzing attitudes on other “single” issues, the results are very similar to the unidirectional relationship found when looking at ideology and the issue index from in Chapter 4. Jobs and Standard of Living For example, Figure 5.5 models the relationship between ideology and attitudes on the extent to which government should seek to provide jobs and a good standard of living. An individual’s ideological placement is, in part, a function of their position on the standard of living question, where they place their preferred party ideologically, and 96 demographic characteristics. Attitudes on the standard of living question are a function of ideological self-placement, placement of preferred party on the standard of living question, and demographic characteristics. Continuing to use the 2004 ANES, the model is estimated using two-stage least squares regression. The results are shown in Figure 5.6. Ideology and the jobs/standard of living question are re-scaled from 0 to 1, with right/conservative leaning answers on the higher end of the scale. The results correspond well with those from estimating the model with the complete 2004 issue index. Opinions on the standard of living question influence where an individual places ideologically. However, ideology has no impact on opinions toward the standard of living question. On this socioeconomic issue, attitudes appear to be driving ideological orientation, while ideology has no reciprocal impact on attitudes toward the issue. Contrast this with the results from analyzing attitudes on same-sex marriage, and we see that ideology has a stronger impact on a life-style issue than it does on a socioeconomic issue. What about an issue dealing with government spending/foreign policy? Defense Spending Figure 5.7 models the relationship between attitudes on defense spending and ideological placement. Similar to the jobs and standard of living question, attitudes on defense spending are a function of an individual’s ideological placement, where they place their preferred party on defense spending and demographics. Their ideological placement is a function of their attitudes on defense spending, where they place their preferred party ideologically, and demographics. 97 Again, using the 2004 ANES, the model is estimated using two-stage least squares regression. The results are in Figure 5.8. Ideology and defense spending are re- scaled from 0 to l, with more right/conservative leaning answers on the higher end. As we saw with attitudes on jobs/standard of living, how much an individual thinks the government should spend on national defense influences ideological self-placement, and yet on attitudes towards defense spending, ideology has no discernible impact. Similar to the socioeconomic issue above, opinions on defense spending help individuals develop an ideological label. Ideology has a stronger impact on a life-style issue, such as homosexuality, than it does on attitudes towards defense spending. Conclusions The analyses presented in this chapter suggest two important points and raise some venues for futures research. First, it is clear that ideology still has an important impact on attitudes towards homosexuals. In both the analysis on same-sex marriage and the analysis on rights for homosexuals, the relationship between ideology and issue attitudes is endogenous. This is dissimilar to the results in Chapter 4, which show that when pooling attitudes together, ideology had no discernible impact on issue attitudes. However, opinions on homosexuality are not measured in the 2004 issue index, and the results here suggest possible differences in how ideology impacts attitudes on life-style issues versus others. Second, comparing the results from the tests on attitudes towards homosexuals to those on jobs/standard of living and defense spending, further emphasizes that there are differences in how ideology affects different types of issues. This raises an interesting concern, which future research may be able to clarify. If issue attitudes are influencing 98 ideological orientation, then ideally, an individual would look at a summary of multiple attitudes to derive an accurate ideological placement. They should ask themselves their opinions on this issue and that issue, and then use those opinions to locate a suitable placement along the ideological continuum. Indeed, we should find that an issue index has a stronger impact on ideological placement than a single issue. In 2004, this is the case. The results from Chapter 3, clearly show that even when controlling for such factors as education and conceptualization, a summary of an individual’s issue attitudes has a profound impact on their ideological placement, while ideology does not help in organizing issue attitudes. However, when looking at the single issue of same-sex marriage, ideology plays a role in attitude formation. One could argue that we can’t have it both ways. Either ideology influences attitude formation or it does not. The results presented here suggest that the relationship is more complex than that. Further research should specifically look at how ideology influences different categories or types of attitudes. The analyses on same-sex marriage and rights for homosexuals suggest that particularly when looking at issues which bring forth questions of morality and lifestyles, ideology has an important influence. But, when looking at socioeconomic issues (jobs/standard of living) or foreign policy/defense issues, ideology steps out of the picture. Future research may show that how important a particular category or type of issue is to an individual, influences the extent to which ideology impacts attitudes on that issue. Issue salience may affect the degree to which people are willing to use an abstract, philosophical orientation, such the liberal/conservative continuum, to organize their attitudes. 99 Table 5.1. Summary of Attitudes on Same-sex Marriage Response Frequency Percentage Should be Allowed 400 34.66 Should not be Allowed, but Should Allow Civil Unions 49 4-25 Should Not be Allowed 705 61.09 Source: 2004 American National Election Study 100 Figure 5.1. Modeling the Relationship between Ideology and Opinions on Same-sex Marriage a, Feelings Towards Ideological Groups Ideology A Demographics a3 [61 a 1 Party Identification Z”: a V Feelings Towards Same-Sex 162 Homosexuals Marriage 101 Figure 5.2. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Opinions on Same-sex Marriage R2=.46 Ideology Feelings Towards ‘ Ideological Groups Demographics 1.73 .053 (.004) (.000) Party Identification v Same-sex Marriage ' Feelings Towards Homosexuals 2:.17 Source: 2004 National Election Studies Series Numbers in parentheses are p-values N=684 102 Figure 5.3. Modeling the Relationship between Ideology and Opinions on Rights for Homosexuals Ideology A fir a1 V Rights for Homosexuals 103 2.8. ,3. Feelings Towards Ideological Groups Demographics Party Identification Feelings Towards Homosexuals Figure 5.4. Two-Stage Least Squares Estimates from Modeling the Relationship . between Ideology and Opinions on Rights for Homosexuals R2=.47 Ideology A 1.37 .233 (.000) (.000) V Rights for Homosexuals R2: .09 .057 (.000) Source: 2004 National Election Studies Series Numbers in parentheses are p-values N=7 I3 .001 (.010) 104 ' Feelings Towards Ideological Groups Demographics Party Identification Feelings Towards Homosexuals Figure 5.5. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Opinions on Government’s Role in Providing Jobs and a Good Standard of Living R2=.28 .084 Party Placement on Id 1 ‘___ __ eo ogy (.000) Ideology A \ .064 .221 Demographics (.391) (.022) .. .__ _§:“..1.‘:::°:.:::2°" L1v1ng (.000) Living R2: .26 Source: 2004 National Election Studies Series Numbers in parentheses are p-values N=559 105 Figure 5.6. Two-Stage Least Squares Estimates from Modeling the Relationship between Ideology and Opinions on Defense Spending R2=.30 Party Placement on Ideology Ideology f .089 .344 Demographics (.176) (.048) V .349 . Party Placement on Defense (.000) Defense Spending Spending R2= .19 Source: 2004 National Election Studies Series Numbers in parentheses are p-values N=532 106 Chapter 6 Discussion and Conclusions In this dissertation, I argue that issues play an important role in determining many individuals’ ideological identification. A wealth of research has been written on how ideological identification helps individuals form opinions about political issues. I argue that, at the very least, we should consider endogeneity within this relationship. I also posit that for some people the causal relationship works in reverse, fueled by a focus on cues from political parties, which shape how individuals position themselves on issues and label themselves ideologically. To test this theory, I model a reciprocal relationship between ideological identification and issue attitudes. The results of empirically testing these models are two-fold. First, analyses performed on data from 1972—2000, indicate that during this time period, there existed a true reciprocal relationship between individual ideological placements and individual issue positions. For respondents in these studies, ideology and issue attitudes feed off of each other. Individuals use their attitudes to label themselves ideologically, while using underlying ideological orientations to organize their issue attitudes. These two conditions are not mutually exclusive. Some issues attitudes may be driven by an individual’s ideology, while some may not. Attitude formation has been attributed to multiple factors, including partisanship, socialization, and ideology. Future research will attempt to sparse out which attitudes are more likely to have roots in ideology and which are not. I have started this process by comparing attitudes towards the rights of homosexuals to those on non-lifestyle questions. These analyses suggest 107 that ideology plays a stronger role in predicting life-style issue than it does on socioeconomic or government spending issues. But, more work is to be done, especially on the consideration of issue salience. Second, data analyses on the 2004 American National Elections Study show that when taking into account a reciprocal relationship between self-reported ideology and issue positions, the link from issue positions to ideology is consistently significant for a number of different model specifications and the link from ideology to issue positions disappears. The data suggest that in 2004, individuals use issue positions to structure their ideological orientations, and not the reverse. This corresponds well with Jacoby’s (2006) analysis, which shows that ideology did not affect vote choice in 2004. He additionally finds that the effect of ideology on vote choice was not mediated by education. I find this to be the case with the effect of ideology on issue attitudes, as well. The implications for the literature, here, are important. First, the findings suggest that, beginning in 2004, a large portion of the public does not use ideology to organize their issue attitudes. Consequently, it is important to consider a causal link from issue positions to ideology. When this type of relationship is taken into account, the evidence presented here shows that in almost all 2004 model specifications, issue positions have considerable predictive power in explaining ideological orientations, even when ideology is also working to shape attitudes. Additionally, the issues to ideology link holds even when controlling for how individuals conceptualization the political world and for levels of education. 108 This is not to say that levels of conceptualization and education do not have an effect, indeed they do. Considering the 2004 tests, the models for the most educated and those focused on ideology, when thinking about candidates or parties, show quite convincingly that these individuals are much more consistent in matching ideological labels with their issue positions. The issues attitudes for both groups are highly correlated with their ideological self-placement, more so than any other group in the study. And, the combination of issue attitudes and control variables in the model, explains over 80% of the variance in ideology for the ideologically focused individuals. Education and conceptualization matter because they help individuals accurately and consistently label themselves with the proper ideological term. Individuals take positions on issues and use the ideological labels they learn from political elites to categorize themselves as “liberal” or “conservative” without having a full understanding of what those labels mean. This is consistent with what Levitin and Miller argued in 1979. People can use ideological labels without being inherently ideological. Second, comparing the analyses from Chapter 3 and Chapter 4, suggest that the relationship between ideology and issue attitudes may be changing. There are clear differences between the 1976-2000 and 2004 results. At the very least, this suggests the need for further testing on future data sets. Additionally, a lengthy panel study would provide more applicable data. Tracking the same individuals across a larger time span would allow for a more detailed analysis of how the relationship tracks with changes in attitudes and ideological identification. At the very least. applying the models used here to several more years of non-panel data would still provide a better picture of if and how 109 this relationship is changing. Given these problems, it is still an important intellectual exercise to ask what questions can be raised and what conclusions can be drawn from assuming that the relationship between ideology and issue attitudes has changed. The Questions First, and most importantly, what has driven this change in the relationship between ideology and issue attitudes? One explanation involves the content of political discourse. Elite and media political discourse has become less content oriented and more like advertisements (Jamieson 1992). Patterson (1993) argues that media coverage of presidential elections “misrepresents the choice among candidates by focusing on action, competitive maneuverings, and journalistic interpretation, rather than on dispassionate substantive analysis of policy issues” (Jacobs and Shapiro 1996, 10). This may have an effect on the quality of discussion regarding ideology. If the media’s discussion of ideology has also become less sophisticated, then it would be reasonable to conclude that this would affect how individuals use ideology to structure their political beliefs. If the media’s use of the terms “liberal” and “conservative” has led them to lose much of their meaning, this may lead to lower levels of ideological understanding in the general public. In fact, we would see cases where individuals use ideological terminology to characterize politics with little substance behind the use of the terms. Consequently, the appearance of more “ideologues” among respondents, may only indicate that individuals hear ideological terms more often and use them to describe candidates and parties, but apply less substantive meaning to them. Finkel and Norpoth (1984) argue the more that ideological cues are transmitted to the public by the media or political elites, and the clearer these cues are, the more 110 likely it is that citizens will show some ideological grasp of politics. However, this may simply be in the form of labels, rather than helping to create a well-defined ideological view. These two lines of research together, suggest that the quality of ideological discussion at the elite level has declined, and yet the discussion still has an effect on how individuals evaluate the political world. This may have led to a change in how ideology and issue positions relate to each other, leaving individuals with ideological labels, absent of ideological content (Levitin and Miller 1979), to identify their political attitudes. This also fits well with Jacoby’s (2006) discussion on ideology and vote choice. The individuals in his analysis could accurately place the candidates and parties on the ideological continuum, suggesting that they could use ideological labels to describe political objects. However, ideology was found to have no impact on voting behavior. Voters could properly attach the correct ideological labels to the candidates and parties, but ideology did not extend far enough to influence their voting decisions. Second, if the role of ideology is truly limited in this way, is it still important, and how? Some might view this study as another example of research that argues the unimportance of ideology in American politics. While the 2004 findings may indicate that ideology has little effect on the development of issue attitudes, this is not to say that it cannot have an impact. In fact, the 1972-2000 data suggest that, at least to some degree, ideological identification can and did help individuals organize their political attitudes. And, evidence from analyses on attitudes towards gay rights shows that on particular types of issues, ideology can still have an important effect. However, for reasons listed above, we may have entered a period where political discourse has 111 become “watered down,” stripping ideological discussion of its abstract ideological content. Additionally, if one believes that ideology has little effect on issue attitudes, this study does not make any assumptions regarding the effect that ideology might have on other political orientations. For example, Palfrey and Poole (1987) show that individuals who do take an ideological position on the liberal-conservative continuum, are more likely to vote than individuals who identify as moderates. Verba and Nie ( 1972) show that stronger ideological identification increases levels of political activity in general. Others (notably Holm and Robinson 1978) show that ideology has an independent effect on vote choice, even after the effects of issues and partisanship are taken into account. Third, if there is a reciprocal relationship between ideology and issue attitudes, do changes in each drive changes in the other? The findings in Chapter 3 suggest that this might be the case. The individuals who changed ideological positions between 1972 and 1976, were shown to have a strong link from their issue positions to their ideological placement, suggesting that issue positions could be driving these shifts. Conversely, those individuals who shifted in their issue attitudes, showed a much stronger link from their ideological identification to their issue attitudes, indicating that their ideological identification may be causing these changes. However, these findings are far from definitive. A proper test would be to directly analyze those individuals who shifted ideological positions and their issue attitudes across the same time period. Are shifts in ideology driving changes in attitudes or vice versa? Also, does this change for different groups across the sample? Are changes in ideology more likely to cause changes in 112 issue positions for one group over another? Or, is another group more likely to have their shifts on issues lead to changes in their ideological identification? A simple adaptation of the model presented in Chapter 3 would be to replace all the key variables with the changes in those variables between 1972 and 1976. In other words, replace the party placements with the changes in where individuals placed the parties between 1972 and 1976. In addition, replace the variables for ideology and issue positions with the changes within the same time period. Testing the model should indicate whether changes in ideology are leading to changes on issues or vice versa. Fourth, the results here have something to say about polarization in the American electorate. As Fiorina (2005) notes, we have entered into a political world characterized as red vs. blue, conservatives vs. liberals, the Christian right vs. the secular left. However, this is a battle elites are fighting, and the general public has been left behind. Fiorina argues that while we may see polarization at the elite level, among members of Congress, presidential candidates, or party activists, we do not see it among the general electorate, even on issues such as abortion. If ideology is no longer driving political attitudes, the findings in this paper support F iorina’s conclusion. The failure of ideology to drive attitudes to the extremes on the left or the right, would be exemplified in a public that lacks issue polarization. Indeed, that is what F iorina finds. Finally, this may provide an explanation for why we see devolution from the original definitions of ideological terms and issue inconsistency. If conservatives are supposed to be for less government intervention, then why are so many in favor of governmental restrictions on same-sex marriage and abortion? The answer may be a simple one. If the terms “liberal” and “conservative” are employed solely as labels that 113 characterize a set of issue attitudes, and not as underlying belief systems, then they can be “adjusted” to label particular attitudes that might not fall within their original definitions. This means that political parties and candidates can dictate, to a degree, what positions are considered “liberal” and what positions are considered “conservative.” They can pick an issue, such as same-sex marriage, and say one side is the conservative side and one side is the liberal side. This may have a profound influence on partisan attitudes toward the issue, especially since the analysis here, and in other places, shows that perceptions of where parties stand on issues affect individual attitudes. At the very least, the findings in this paper suggest that future research should consider the ramifications this argument. The Conclusions One result of this study is that it may raise more questions than answers. This is not to say that we are not to draw any important conclusions from the work done here. It is important to note what the findings here say and do not say. First, the results indicate that the relationship between ideology and issue positions is not simple. Of course, political scientists have known this for a long time. However, it is important to note that not only do the results suggest, at least in 2004, the relationship works in the opposite direction of that which may have previously been assumed, but that perhaps this relationship has even changed over time. As noted, an accurate longitudinal study would provide us with a better grasp of the latter conclusion, but we can at least say that things were different in. 2004 than they were in 1976 or even 2000. 114 Second, the findings underline the importance of testing for reciprocal relationships between important political orientations. Research has long considered the possibility of a reciprocal link between issue positions and partisanship, however, little research has considered whether the same type of feedback exists between issues and ideology. Part of this has been dependent upon how we conceptualize ideology. When conceptualized as an underlying belief structure from which attitudes are formed and organized, it is hard to develop a model that would allow attitudes to turn around and influence ideology. The important distinction found in this study, is not just that for some ideology no longer helps individuals form issue attitudes, but that in fact it is issue attitudes that help individuals label themselves ideologically. If this distinction holds, at least two important conclusions can be drawn from it. First, as noted earlier, it suggests that even if we find more individuals using ideological terms to describe political subjects, this does not necessarily mean that people have become more ideological. It may simply imply that individuals have been conditioned to use these terms by political elites, but attach very little substantive meaning to them. They are able to apply the labels in a way that may accurately describe their issues positions, based on cues they pick up from elites. Beyond that the labels carry very little intrinsic value. Of course, the results in Chapter 4, also indicate the accuracy to which individuals can apply these labels to their attitudes is conditioned upon their level of education and how often they use these terms to describe the political world. Second, it indicates that a closer look of how ideology interacts with individual conceptualizations of the political world is in order. Admittedly, my measure of 115 conceptualization is different than that originally used by Campbell et al. (1960). My measure of ideologues captures the degree to which an individual will use ideological terms to describe candidates and parties over their use of other terms, such as issue references. Whereas the original ideologue variable measured whether an individual could use at least one ideological term and then pair it with an appropriate issue position or reference group. This difference is important, and should be kept in mind when interpreting the results presented here. Access to the open ended responses to the 2004 candidate/party likes and dislikes in the ANES data would provide a more accurate way to directly code the levels of conceptualization, and more concrete conclusions could be drawn about the extent to which the levels still influence ideological thinking in the American electorate. F inally, conceptualizing the relationship between ideology and policy positions as reciprocal challenges the traditional view that ideology is a predictor of issue positions and should be considered a standard control variable when modeling attitudes on issues. The results presented here show that future projects that include an analysis of the relationship between ideology and issues, should consider a model that incorporates a reciprocal relationship between the two. Further research will determine the extent to which the link from ideology to issues has eroded, if indeed it has. However, analyses in Chapter 3 and, most importantly, Chapter 4, strongly suggest that, to the very least, those modeling issues and ideology incorporate a non-recursive element between the two. 116 Appendix A I. The following is the question wording and possible responses for the issue questions used in the multiple analyses from Chapter 3. All were coded so that “right” leaning answers were associated with higher numbers, and re-scaled from 0 to 1. 1972-1976 Data Jobs and Standard of Living Some people feel that the government in Washington should see to it that every person has a job and a good standard of living. Suppose that these people are at one end of this scale—at point number 1. Others think the government should just let each person get ahead on his own. Suppose that these people are at the other end—at point number 7. And, of course some other people have opinions in between. Where would you place yourself on this scale, or haven’t you thought much about this? 1. Government see to job and good standard of living 2 3. 4. 5 6 7. Government let each person get ahead on his own Tax System As you know, in our tax system people who earn a lot of money already have to pay higher rates of income tax than those who earn less. Some people think that those with high incomes should pay even more of their income into taxes than they do now. Others think that the rates shouldn’t be different at all—that everyone should pay the same ' portion of their income, no matter how much they make. Where would you place yourself on this scale, or haven’t you thought much about this? 1. Increase the tax rate for high incomes 2 3. 4. 5 6 7. Have the same tax rate for everyone 117 Busing to Achieve Integration There is much discussion about the best way to deal with racial problems. Some people think that achieving racial integration of schools is so important that it justifies busing children to schools out of their own neighborhoods. Others think letting children go to their neighborhood schools is so important that they oppose busing. Where would you place yourself on this scale, or haven’t you thought much about this? 1. Bus to achieve integration 2 3. 4. 5 6. 7. Keep children in neighborhood schools Health Care There is much concern about the rapid rise in medical and hospital costs. Some feel there should be a government insurance plan which would cover all medical and hospital expenses. Others feel that medical expenses should be paid by individuals, and through private insurance like Blue Cross. Where would you place yourself on this Scale, or haven’t you thought much about this? 1. Government insurance plan 2 3. 4. 5 6 7. Private insurance plans Rights of the Accused Some people are concerned with doing everything possible to protect the legal rights of those accused of committing crimes. Others feel that it is more important to stop criminal activity even at the risk of reducing the rights of the accused. Where would you place yourself on this scale, or haven’t you thought much about this? Protect rights of the accused l. 2. 3. 4 118 5. 6. 7. Stop crime regardless of rights of accused Aid to Minorities Some people feel that the government in Washington should make every possible effort to improve the social and economic position of blacks and other minority groups. Others feel that the government should not make any special effort to help minorities because they should help themselves. Where would you place yourself on this scale, or haven’t you thought much about this? 1. Government should help minority groups 2 3. 4. 5 6 7. Minority groups should help themselves Urban Unrest and Rioting Over the past few years there has been much discussion about the best way to deal with the problem of urban unrest and rioting. Some say it is more important to use all available force to maintain law and order—no matter what the results. Others say it is more important to correct the problems of poverty and unemployment that give rise to the disturbances. Where would you place yourself on this scale, or haven’t you thought much about this? 1. Solve problems of poverty and unemployment 2 3. 4. 5 6 7. Use all available force 1984 Data Spending on Services Some people think the government should provide fewer services, even in areas such as health and education, in order to reduce spending. Suppose these people are at one end 119 of the scale at point number 1. Other people feel it is important for the government to provide many more services even if it means an increase in spending. Suppose these people are at the other end of the scale, at point 7. And of course, some other people have opinions somewhere in between at points 2, 3, 4, 5, or 6. Where would you place yourself on this scale. or haven’t you thought much about this? 1. Government should provide many fewer services; reduce spending a lot 2 3. 4. 5 6 7. Government should provide many more services; increase spending a lot Aid to Minorities Some people feel that the government in Washington should make every possible effort to improve the social and economic position of blacks and other minority groups. Others feel that the government should not make any special effort to help minorities because they should help themselves. Where would you place yourself on this scale,'or haven’t you thought much about this? 1. Government should help minority groups 2 3. 4. 5 6 7. Minority groups should help themselves Defense Spending Some people believe that we should spend much less money for defense. Suppose these people are at one end of a scale, at point 1. Others feel that defense spending should be greatly increased. Suppose these people are at the other end, at point 7. And, of course, some other people have opinions somewhere in between, at points 2, 3, 4, 5 or 6.Where would you place yourself on this scale, or haven't you thought much about this? Government should decrease defense spending 1. 2 3. 4. 5 6 120 7. Government should increase defense spending Couperation with Russia Some people feel it is important for us to cooperate more with Russia, while others believe we should be much tougher in our dealings with Russia. Where would youp lace yourself on this scale, or haven’t you thought much about this? 1. Try to cooperate with Russia 2 3. 4. 5 6. 7. Get much tougher with Russsia Role of Women in Society Some people feel that the government in Washington should make every effort to improve the social and economic position of women. Others feel that the government should not make any special effort to help women because they should help themselves. Where would you lace yourself on this scale, or haven’t you thought much about this? 1. Government should help women 2 3. 4. 5 6 7. Women should. help themselves Jobs and Standard of Living Some people feel that the government in Washington should see to it that every person has a job and a good standard of living. Suppose that these people are at one end of this scale—at point number 1. Others think the government should just let each person get ahead on his own. Suppose that these people are at the other end—at point number 7. And, of course some other people have opinions in between. Where would you place yourself on this scale, or haven’t you thought much about this? 1. Government see to job and good standard of living 3. 121 Government let each person get ahead on his own 1996 Data Spending on Services Some people think the government should provide fewer services, even in areas such as health and education, in order to reduce spending. Suppose these people are at one end of the scale at point number 1. Other people feel it is important for the government to provide many more services even if it means an increase in spending. Suppose these people are at the other end of the scale, at point 7. And of course, some other people have opinions somewhere in between at points 2, 3, 4, 5, or 6. Where would you place yourself on this scale, or haven’t you thought much about this? 1. Government should provide many fewer services; reduce spending a lot 2 3. 4. 5 6 7. Govermnent should provide many more services; increase spending a lot Abortion There has been some discussion about abortion during recent years. Which one of the opinions on this page best agrees with your view? You can just tell me the number of the opinion you choose. 1. By law, abortion should never be permitted. 2. The law should permit abortion ONLY in case of rape, incest or when woman's life is in danger. 3. The law should permit abortion for reasons OTHER THAN rape, incest or danger to the woman's life, but only after the need for the abortion has been clearly established. 4. By law, a woman should always been able to obtain an abortion as a matter of personal choice Defense Spending Some people believe that we should spend much less money for defense. Suppose these people are at one end of a scale, at point 1. Others feel that defense spending should be greatly increased. Suppose these people are at the other end, at point 7. And, of course, some other people have opinions somewhere in between, at points 2, 3, 4, 5 or 6.Where would you place yourself on this scale, or haven't you thought much about this? 1. Government should decrease defense spending 2 3. 4. 5 6 7. Government should increase defense spending Jobs and the Environment Some people think it is important to protect the environment even if it costs some jobs or otherwise reduces our standard of living. (Suppose these people are at one end of the scale, at point number 1.) Other people think that protecting the environment is not as important as maintaining jobs and our standard of living. (Suppose these people are at the other end of the scale, at point number 7.) And, of course, some other people have opinions somewhere in between, at points 2,3,4,5 or 6. Where would you place YOURSELF on this scale, or haven't you thought much about this? 1. Protect environment, even if it costs jobs, standard of living 2 3. 4. 5 6. 7. Jobs. standard of living more important than environment Environmental Regulation Some people think we need much tougher government regulations on business in order to protect the environment. (Suppose these people are at one end of a scale, at point 1.) Others think that current regulations to protect the environment are already too much of a burden on business. (Suppose these people are at the other end of the scale, a point number 7.) And, of course, some other people have opinions somewhere in between at points 2,3,4,5 or 6. Where would you place YOURSELF 011 this scale, or haven't you thought much about this? 123 l. Tougher regulations on business needed to protect the environment 2 3. 4. 5 6 7. Regulations to protect environment already too much of a burden on business 2000 Data Spending on Services Some people think the government should provide fewer services, even in areas such as health and education. in order to reduce spending. Suppose these people are at one end of the scale at point number 1. Other people feel it is important for the government to provide many more services even if it means an increase in spending. Suppose these people are at the other end of the scale, at point 7. And of course, some other people have opinions somewhere in between at points 2, 3, 4, 5, or 6. Where would you place yourself on this scale, or haven’t you thought much about this? 1. Government should provide many fewer services; reduce spending a lot 2 3. 4. 5 6 7. Government should provide many more services; increase spending a lot Aid to Minorities Some people feel that the government in Washington should make every possible effort to improve the social and economic position of blacks and other minority groups. Others feel that the government should not make any special effort to help minorities because they should help themselves. Where would you place yourself on this scale, or haven’t you thought much about this? 1. Government should help minority groups 2 3. 4. 5 6 7. Minority groups should help themselves Defense Spending Some people believe that we should spend much less money for defense. Suppose these people are at one end of a scale, at point 1. Others feel that defense spending should be greatly increased. Suppose these people are at the other end, at point 7. And, of course, some other people have opinions somewhere in between, at points 2, 3, 4, 5 or 6.Where would you place yourself on this scale, or haven't you thought much about this? 1. Government should decrease defense spending 2 3. 4. 5 6 7. Government should increase defense spending Jobs and Standard of Living Some people feel that the government in Washington should see to it that every person has a job and a good standard of living. Suppose that these people are at one end of this scale—at point number 1. Others think the government should just let each person get ahead on his own. Suppose that these people are at the other end—at point number 7. And, of course some other people have opinions in between. Where would you place yourself on this scale. or haven’t you thought much about this? 1. Government see to job and good standard of living 2 3. 4. 5 6 7. Government let each person get ahead on his own 11. The index of attitudes for each year was constructed using the items listed above. The responses were added together and then divided by the number of questions the respondent answered. Because the questions were already re-scaled from 0 to 1, the index itself was on a scale of 0 to 1. For example, if Respondent A answered all four questions, in 2000, as follows: 2, 2, 1, 1, the index score for Respondent A would be .084, indicating that they displayed left-leaning attitudes. 125 Table A.2. Two-Stage Least Squares Estimates From Modeling the Relationship between Ideology and Issue Positions for 1972-1976 Respondents (Lagged Variables and Party Placements as Instruments) Issue Index 2SLS Coefficient Ideology 2 SLS Coefficient Ideology Issue Index Ideology 1972 Issue Index 1972 Party Placement on Issues Party Placement of Ideology Income Education Age Gender African-American llispanic Asian Native American Other Constant R2 N .167 (.000) .467 (.000) .242 (.000) .002 (.081) -004 (.050) -.001 (.1 12) -007 (.470) —.053 (.027) -047 (.525) .195 (.034) No Observations No Observations .151 (.000) .529 687 Source: 1972-1976 American National Elections Study Numbers in parentheses are p-values .354 (.000) .321 (.000) .066 (.000) .000 (.778) .003 (.259) .001 (.108) .026 (.022) -.011 (.700) .112 (.192) .045 (.669) No Observations No Observations Appendix B I. The following is the question wording and possible responses for the seven issue questions that comprise the issue index in Chapter 4. All were coded so that “right” leaning answers were associated with higher numbers. Spending on Services Some people think the government should provide fewer services even in areas such as health and education in order to reduce spending. Suppose these people are at one end of a scale, at point 1. Other people feel it is important for the government to provide many more services even if it means an increase in spending. Suppose these people are at the other end, at point 7. And, of course, some other people have opinions somewhere in between, at points 2, 3, 4, 5 or 6. Where would you place YOURSELF on this scale, or haven't you thought much about this? 1. Government should provide many fewer services 2 3. 4. 5 6. 7. Government should provide many more services Defense Spending Some people believe that we should spend much less money for defense. Suppose these people are at one end of a scale, at point 1. Others feel that defense spending should be greatly increased. Suppose these people are at the other end, at point 7. And, of course, some other people have opinions somewhere in between, at points 2, 3, 4, 5 or 6.Where would you place YOURSELF on this scale, or haven't you thought much about this? 1. Government should decrease defense spending 2 3. 4. 5 6. 7. Government should increase defense spending Jobs and Good Standard of Living Some people feel the government in Washington should see to it that every person has a job and a good standard of living. Suppose these people are at one end of a scale, at point 1. Others think the government should just let each person get ahead on their own. 135 Suppose these people are at the other end, at point 7. And, of course, some other people have opinions somewhere in between, at points 2, 3, 4, 5, or 6.Where would you place YOURSELF on this scale, or haven't you thought much about this? 1. Government should see to jobs and good standard of living 2 3. 4. 5 6 7. Government should let each person get ahead on their own Assistance to Blacks Some people feel that the government in Washington should make every effort to improve the social and economic position of blacks. (Suppose these people are at one end of a scale, at point 1.) Others feel that the government should not make any special effort to help blacks because they should help themselves. (Suppose these people are at the other end, at point 7.) And, of course, some other people have opinions somewhere in between, at points 2, 3, 4, 5, or 6.Where would you place YOURSELF on this scale, or haven't you thought much about this? 1. Government should help blacks soweww . Blacks should help Women’s Role in Society Recently there has been a lot of talk about women's rights. Some people feel that women should have an equal role with men in running business, industry, and government. (Suppose these people are at one end of a scale, at point 1.) Others feel that a woman's place is in the home. (Suppose these people are at the other end, at point 7.) And, of course, some other people have opinions somewhere in between, at points 2, 3, 4, 5 or 6.Where would you place YOURSELF on this scale, or haven't you thought much about this? 1. Women and men should have equal roles 2. 3. 4. 5. 6. 136 7. A woman's place is in the home Diplomacy Some people believe the United States should solve international problems by using diplomacy and other forms of international pressure and use military force only if absolutely necessary. Suppose these people are at one end of a scale, at point 1. Others believe diplomacy and pressure often fail and the US. must be ready to use military force. Suppose these people are at the other end, at point 7. And, of course, some other people have opinions somewhere in between, at points 2, 3, 4, 5, or 6. Where would you place YOURSELF on this scale, or haven't you thought much about this? 1. Should solve with diplomacy 2 3. 4. 5 6. 7. Must be ready to use military force Abortion There has been some discussion about abortion during recent years. Which one of the opinions on this page best agrees with your view? You can just tell me the number of the opinion you choose. 1. By law, abortion should never be permitted. 2. The law should permit abortion only in case of rape, incest, or when the woman's life is in danger. 3. The law should permit abortion for reasons other than rape, incest, or danger to the woman's life, but only after the need for the abortion has been clearly established. 4. By law, a woman should always be able to obtain an abortion as a matter of personal choice. 137 II. The index on attitudes was constructed using the seven items listed above. The responses were added together and then divided by the number of questions the respondent answered. Because the questions were already re-scaled from 0 to l, the index itself was on a scale of O to 1. For example, if Respondent A answered all seven questions as follows: 2, 2, l, 1, 2, 3, 2, the index score for Respondent A would be .155, indicating that they displayed left-leaning attitudes. III. The following is the question wording for candidate and party likes and dislikes questions. Respondents were given the option to list up to five reasons why they liked or disliked the object in question. Respondents could have had up to 40 possible responses. Each of the reasons was coded as an issue reason, an ideological reason or other. The number of issue reasons, ideological reasons, and other reasons were then counted. Individuals that gave more issue reasons than the other two categories were considered to be issue focused. Individuals who gave more ideological reasons than the other two categories were considered to be ideologically focused. As previously noted, those individuals who placed an ideological label (liberal or conservative) at higher than 80% and the other ideological label at lower than 20%, on their respective feeling thermometers were also included in the ideologically focused category. Candidate/Party Likes and Dislikes Now I'd like to ask you about the good and bad points of the major candidates for President. 13 there anything in particular about George W. Bush that might make you want to vote FOR him? Is there anything in particular about George W. Bush that might make you want to vote AGAINST him? Is there anything in particular about John Kerry that might make you want to vote FOR him? Is there anything in particular about John Kerry that might make you want to vote AGAINST him? Is there anything in particular that you like about the Democratic Party? 13 there anything in particular that you dislike about the Democratic Party? 13 there anything in particular that you like about the Republican Party? 13 there anything in particular that you dislike about the Republican Party? Table 8.1 Two-Stage Least Squares Estimates for All Respondents Equation: Issue Index Ideology ZSLS Coefficient 2SLS Coefficient Ideology .057 (.536) Issue Index .965 (000) Party Placement on Issues .494 (000) Party Placement on .036 Ideology (.000) Income .003 .003 (.008) (.007) Education -.016 .002 (.000) (.653) Age .001 .001 (.031) (.076) Gender -.025 .009 (.011) (.493) African-American -.061 .090 (.000) (.000) Hispanic -.017 -.050 (.401) (.051) Asian -.053 .005 (.111) (.918) Native American -.027 .118 (.533) (.753) Other -.010 .041 (.756) (350) Constant .195 -.157 (.000) (.001) R2 .413 .463 N 819 819 Source: 2004 American National Elections Study Numbers in parentheses are p-values 139 Table B.2. Two-Stage Least Squares Estimates for Issue Focused People Equation: Issue Index Ideology 2SLS Coefficient 2SLS Coefficient Ideology .003 (975) Issue Index .846 (.000) Party Placement on Issues .482 (000) Party Placement of .042 Ideology (.000) Income .004 .002 (.004) (.221) Education -.018 .004 (.000) (.572) Age .001 .001 (.039) (.052) Gender -.030 .009 (.019) (.610) African-American -.052 .064 (.013) (.030) Hispanic -.050 -.041 (055) (841) Asian -.050 -.014 (.257) (.812) Native American .010 -.O24 (.887) (.785) Other .025 .063 (.543) (.229) Constant .210 -.129 (.000) (.454) R2 .385 .456 N 468 468 Source: 2004 American National Elections Study Numbers in parentheses are p-values 140 Table 8.3. Two-Stage Least Squares Estimates for Ideology Focused People Equation: Issue Index Ideology 2SLS Coefficient 2 SLS Coefficient Ideology -.073 (.841) Issue Index .991 (000) Party Placement on Issues .716 (062) Party Placement of .060 Ideology (.000) Income .007 .002 (175) (464) Education -.029 -.018 (.146) (.118) Age -.000 .000 (.961) (.901) Gender -.027 -.001 (.474) (.959) African-American -. 197 .099 (.043) (.232) Hispanic .083 -.086 (.278) (.183) Asian -.157 .050 (.119) (.503) Native American .017 .120 (.930) (.387) Other -.344 -.047 (.1 18) (.756) Constant .248 -.l 17 (.051) (.262) R2 .521 .827 N 97 97 Source: 2004 American National Elections Study Numbers in parentheses are p-values 141 Table 8.4. Two-Stage Least Squares Estimates for Respondents with No High School Diploma Equation: Issue Index Ideology 2SLS Coefficient 2SLS Coefficient Ideology - .876 (549) Issue Index 2.262 (254) Party Placement on Issues .727 (354) Party Placement of -.050 ' Ideology (.330) Income .014 .007 (500) (577) Age .003 -.004 (.289) (.431) Gender -.003 .193 (.980) (.339) African-American -.15 5 .3 87 (125) (323) Hispanic -.209 -.l98 (554) (261) Asian -.186 .387 (.532) (.497) Native American .118 -.711 (.746) (.277) Other Constant R2 N No Observations .442 (239) .000 52 Source: 2004 American National Elections Study Numbers in parentheses are p-values 142 No Observations 4434 (597) .000 52 Table 8.5. Two-Stage Least Squares Estimates for Respondents with a High School Diploma but No College Education Equation: Issue Index Ideology 2SLS Coefficient 2SLS Coefficient Ideology .076 (.460) Issue Index .921 (000) Party Placement on Issues .461 (000) Party Placement of .042 Ideology (000) Income .001 .005 (.625) (.002) Age .001 .001 (.210) (.091) Gender -.037 .028 (.011) (.181) African-American -.054 . 107 (.021) (.001) Hispanic -.023 -.006 (.435) (.891) Asian .031 .085 (.696) (.435) Native American -.026 .091 (.616) (.205) Other .030 .1 17 (.590) (.1 12) Constant .238 -.223 (.000) (.029) R2 .352 .397 N 377 377 Source: 2004 American National Elections Study Numbers in parentheses are p-values 143 Table 8.6. Two-Stage Least Squares Estimates for Respondents with at Least Some College Education Equation: Issue Index Ideology ZSLS Coefficient 2SLS Coefficient Ideology -.045 (778) Issue Index .892 (000) Party Placement on Issues .580 (000) Party Placement of .041. Ideology (.000) Income .006 .001 (.000) (.749) Age .001 .001 (.049) (.044) Gender -.013 -.008 (.397) (.623) African-American -.050 .064 (.059) (.044) Hispanic .008 -.052 (.800) (.151) Asian -.075 -.002 (.071) (.970) Native American -.223 -.029 (.042) (.808) Other -.055 .004 (.215) (.944) Constant .302 -. 102 (.000) (.276) R2 .403 .609 N 390 390 Source: 2004 American National Elections Study Numbers in parentheses are p-values 144 Appendix C I. The following are the questions asked in the 2004 ANES measuring opinions on policies to promote the rights of homosexuals. All were coded so that “right” leaning answers were higher numbers, and then re-scaled from 0 to 1. Same-Sex Marriage Should same-sex couples be ALLOWED to marry, or do you think they should NOT BE ALLOWED to marry? 1. Should be allowed 2. Should not be allowed to marry but should be allowed to legally form a civil union 3. Should not be allowed Laws to Protect Homosexuals Against Job Discrimination Do you FAVOR or OPPOSE laws to protect homosexuals against job discrimination? 1. Favor strongly 2. Favor not strongly 3. Oppose not strongly 4. Oppose strongly Homosexuals in the Military Do you think homosexuals should be allowed to serve in the United States Armed Forces or don't you think so? 1. Feel homosexuals should be allowed - strongly 2. Feel homosexuals should be allowed - not strongly 3. Feel homosexuals should not be allowed - not strongly 4. Feel homosexuals should not be allowed - strongly Adoption by Homosexual Couples Do you think gay or lesbian couples, in other words. homosexual couples, should be legally permitted to adopt children? 1. Yes 2. No 2. The index on attitudes towards rights for homosexuals was constructed using the four items listed above. The responses were added together and then divided by the number of questions the respondent answered. Because the questions were already re-scaled 145 from 0 to l, the index itself was on a scale of 0 to 1. For example, if Respondent A answered all four questions as follows: 2, 2, 1, 1, the index score for Respondent A would be .208, indicating that they displayed left-leaning attitudes on the rights of homosexuals. 146 Table C.l. Two-Stage Least Squares Estimates From Modeling the Relationship between Ideology and Opinions on Same-sex Marriage Equation: Same-Sex Marriage Ideology 2SLS Coefficient 2 SLS Coefficient Ideology 1.730 __ (.000) Same-sex Marriage _ .053 (000) Feelings Towards -.004 _ Homosexuals (.000) Feelings Towards _ .001 Ideological Groups (.000) Party Identification -.054 .053 (.191) (.000) Income -.002 .003 (.673) (.015) Education -.032 -.001 (.019) (.915) Age .001 .001 (524) (084) Gender .035 -.008 (.321) (.569) African-American .019 .055 (.816) (.039) Hispanic .072 -.028 (305) (337) Asian .133 -.065 (288) (194) Native American .155 -.039 (326) (554) Other -.019 .024 (870) (616) Constant .163 .045 (396) (339) R2 .170 .455 N 684 684 Source: 2004 American National Elections Study Numbers in parentheses are p-values 147 Table C.2. Two-Stage Least Squares Estimates From Modeling the Relationship between Ideology and Opinions on Rights for Homosexuals Equation: Rights for Homosexuals Ideology 2SLS Coefficient 2 SLS Coefficient Ideology 1.372 __ (000) Rights for Homosexuals _ .233 (000) Feelings Towards -.005 _ Homosexuals (.000) Feelings Towards __ .001 Ideological Groups (.010) Party Identification -.051 .057 (027) (000) Income -.008 .005 (017) (000) Education -.004 -.007 (723) (145) Age -.001 .001 (300) (003) Gender -.024 .003 (.371) (.839) African-American -.073 .072 (242) (003) Hispanic .049 -.023 (379) (409) Asian .082 -.053 (409) (282) Native American .109 -.032 (.388) (.621) Other -.086 .040 (352) (388) Constant .1 3 7 . 104 (380) (014) R2 .090 .466 N 713 713 Source: 2004 American National Elections Study Numbers in parentheses are p-values 148 Table C.3. Two-Stage Least Squares Estimates From Modeling the Relationship between Ideology and Opinions on Government’s Role in Providing Jobs and a Good Standard of Living Equation: Jobs/Standard of Living Ideology 2SLS Coefficient 2 SLS Coefficient Ideology .064 __ (391) Jobs/Standard of Living __ .221 (022) Party Placement on .546 __ Jobs/Standard of Living (.000) Party Placement of _ .084 Ideology (.000) Income .005 -.001 (.140) (.793) Education .005 -.015 (.434) (.047) Age .000 .001 (.641) (.148) Gender -.006 -.027 (.760) (.210) Race (Black=l) -.040 -.067 (.191) (.045) Constant .252 .216 (.002) (.018) R2 .257 .277 N 559 559 Source: 2004 American National Elections Study Numbers in parentheses are p-values 149 Table C.4. Two-Stage Least Squares Estimates From Modeling the Relationship between Ideology and Opinions on Defense Spending Equation: Defense Spending Ideology ZSLS Coefficient 2 SLS Coefficient Ideology .089 __ (.176) Defense Spending __ .344 (.048) Party Placement on .349 __ Defense Spending (.000) Party Placement of __ .082 Ideology (.000) Income -.001 .003 (.745) (.479) Education -.021 -.005 (.001) (.573) Age .001 .001 (.006) (.383) Gender -.017 -.027 (.339) (.232) Race (Black=l) -.017 -.081 (.571) (.023) Constant .393 .1 17 (000) (347) R2 .185 .301 N 532 532 Source: 2004 American National Elections Study Numbers in parentheses are p-values 150 References Alt, James E., Bo Sarlvik, and Ivor Crewe. 1976. “Partisanship and Policy Choice: Issue Preferences in the British Electorate, February 1974.” British Journal of Political Science. 6: 3, 273-290. Bartels, Larry M. 1996. “Uninformed Votes: Information Effects in Presidential Elections.” American Journal of Political Science. 40: 194-230. Bennett, Linda L. and Stephen E. Bennett. 1990. Living with Leviathan: Americans Coming to Terms with Big Government. Lawrence, KS: University Press of Kansas. Brewer, Paul R. 2003. “The Shifting Foundations of Public Opinion about Gay Rights.” The Journal of Politics. 65: 4, 1208-1220. Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960. The American Voter. New York: Wiley. Carmines, Edward G. and James A. Stimson. 1989. Issue Evolution: Race and the Transformation of American Politics. Princeton, NJ: Princeton University Press. Conover, Pamela J. and Stanley F eldman. 1981. “The Origins and Meaning of Liberal/Conservative Self-Identifications.” American Journal of Political Science. 25: 617-645. Converse, Philip E. 1964. “The Nature of Belief Systems in Mass Publics.” In: D. E. Apter (Ed.), Ideology and Discontent. New York: Free Press. Cooperman, Alan. 2004. “Same-sex Bans Fuel Conservative Agenda.” Washington Post. com. Retrieved March 25, 2008, from http://www.washingtonpost.com/wp- dyn/articles/A23672-2004Nov3.html Delli Carpini, Michael X. and Scott Keeter. 1996. What Americans Know about Politics and Why it Matters. New Haven: Yale University Press. Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper and Row. Erikson, Robert S. and Kent L. Tedin. 2005. American Public Opinion: Its Origins, '7l/t Content and Impact, 7 ed. New York: Pearson Longman. F inkel, Steven and Helmut Norpoth. 1984. “Candidates and Issues in the 1980 Campaign: The Ideological Connection.” Political Behavior. 6: 1, 61-77. F leishman, John A. 1986. “Types of Political Attitude Structure: Results of a Cluster Analysis.” Public Opinion Quarterly. 50: 371-386. Franklin, Charles H. 1984. “Issue Preferences, Socialization, and the Evolution of Party Identification.” American Journal of Political Science. 28: 3, 459-478. Free, Lloyd A. and Hadly Cantril. 1967. The Political Beliefs of Americans. New Brunswick, NJ: Rutgers University Press. Gamson, William A. and Andre Modigliani. 1989. “Media Discourse and Public Opinion in Nuclear Power: A Constructionist Approach. American Journal of Sociology. 95: 1-37. Haeberle, Steven H. 1999. “Gay and Lesbian Rights: Emerging Trends in Public Opinion and Voting Behavior.” In Gay and Lesbians in the Democratic Process, eds. Ellen D. B. Riggle and Barry L. Tadlock. New York: Columbia University Press, pp. 146-169. Hamill, Ruth C. and Milton Lodge. 1986. “Cognitive Consequences of Political Sophistication.” In Political Cognition, ed. Richard R. Lau and David 0. Sears. Hillsdale, NJ: Lawrence Erlbaum. Hamill, Ruth C., Milton Lodge, and Frederick Blake. 1985. “The Breadth, Depth, and Utility of Class, Partisan, and Ideological Schemata.” American Journal of Political Science. 29: 850-870. Holm, John D. and John P. Robinson. “Ideological Identification and the American Voter.” Public Opinion Quarterly. 42: 235-246. Hyman, Herbert and Eleanor Singer, eds. 1968. Readings in Reference Group Theory and Research. New York: Free Press. Hymes, Robert. 1986. “Political Attitudes as Social Categories: A New Look at Selective Memory.” Journal of Personality and Social Psychology. 51: 233-241. Jackson, John E. 1975. “Issues, Party Choices, and Presidential Votes.” American Journal ofPolitical Science. 19: 2, 161-185. Jacobs, Lawrence R. and Robert Y. Shapiro. 1994. “Toward the Integrated Study of Political Communications, Public Opinion, and the Policy-Making Process.” PS: Political Science and Politics. 29: 1, 10-13. Jacoby, William G. 1986. “Levels of Conceptualization and Reliance on the Liberal- Conservative Continuum.” Journal of Politics. 48: 423-432. 152 Jacoby, William G. 1988. “The Impact of Party Identification on Issue Attitudes.” American Journal of Political Science. 32: 643-661. Jacoby, William G. 1991. “Ideological Identification and Issue Attitudes.” American Journal of Political Science. 35: 178-205. J acoby, William G. 1994. “Public Attitudes Toward Government Spending. American Journal of Political Science. 38: 336-361. Jacoby, William G. 1995. “The Structure of Ideological Thinking in the American Electorate.” American Journal of Political Science. 39: 314-335. Jacoby, William G. 2002. “Liberal-Conservative Thinking in the American Electorate.” Political Decision Making, Deliberation, and Participation. 6: 97-147. J acoby, William G. 2006. “Ideology in the 2004 Election.” Paper presented at the Conference on “The Wartime Election of 2004.” The Mershon Center, The Ohio State University. Columbus, Ohio, January 12-15, 2006. Jamieson, Kathleen. 1992. Dirty Politics: Deception, Distraction, and Democracy. New York: Oxford. Jefferson, Brandie M. (2006). “Social Conservatives Say Gay Marriage Threaten Religious Freedom.” The Boston Globe. Retrieved March 25, 2008, from http://www.boston.com/news/local/massachusetts/articles/2006/10/ 1 5/social_con servatives_say_gay_marriage_threatens_religious__freedom/. Kinder, Donald R. and D. Roderick Kiewiet. 1978. “Economic Grievances and Political Behavior: The Role of Personal Discontents and Symbolic Judgments in Congressional Voting.” American Journal of Political Science. 23: 3, 495-527. Knight, Kathleen. 1985. “Ideology in the 1980 Election: Ideological Sophistication Does Matter.” Journal of Politics 47: 828-853. Kovenock, David M. and Philip L. Beardsley. 1970. “Status, Party, Ideology, Issues and Candidate Choice: A Preliminary, Theory Relevant Analysis of the 1968 American Presidential Election. Paper presented at the 1970 Meeting of the International Political Science Association. Kuklinski, James. H., Daniel S. Metlay and W. D. Kay. 1982. “Citizen Knowledge and Choices on the Complex Issue of Nuclear Energy. American Journal of Political Science. 26: 615-642. Lau, Richard R., Thad A. Brown, and David O. Sears. 1978. “Self-Interest and Civilians’ Attitudes Toward the Vietnam War. Public Opinion Quarterly. 42: 464-483. 153 Levitin, Teresa E. and Warren E. Miller. 1979. “Ideological Interpretations of Presidential Elections.” American Political Science Review. 73: 757-771. Lewis, Gregory B. and Marc A. Rogers. 1999. “Does the Public Support Equal Employment Rights for Gays and Lesbians?” In Gays and Lesbians in the Democratic Process, eds. Ellen D. B. Riggle and Barry L. Tadlock. New York: Columbia University Press, pp. 118-145. Lipset, Seymour M. 1960. Political Man. Garden City, NY: Doubleday. Luttbeg, Norman R., and Michael M. Gant. 1985. “The Failure of Liberal-Conservative Ideology as a Cognitive Structure. Public Opinion Quarterly. 49: 80-93. Martinez, Michael D. and Michael M. Gant. 1990. “Partisan Issue Preferences and Partisan Change.” Political Behavior. 12: 3, 243-264. McClosky, Herbert. 1958. “Conservatism and Personality.” The American Political Science Review. 52: 1, 27-45. McConahay, John B., and Willis D. Hawley. 1977. “Is it the Buses or the Blacks: Self- interest versus Symbolic Racism as Predictors of Opposition to Busing in Louisville.” Proceedings of the American Psychological Association’s 85th Annual Convention. Miller, Arthur H., Warren E. Miller, Alden S. Raine, and Thad A. Brown. 1976. “A Majority Party in Disarray.” The American Political Science Review. 70: 3, 753- 778. Miller, Warren E. 1976. “The Cross-National Use of Party Identification as a Stimulus to Political Inquiry.” In Ian Budge, Ivor Crewe, and Dennis F arlie, (Eds), Party Identification and Beyond. New York: Wiley. Miller, Warren E. and Teresa E. Levitin. 1977. Leadership and Change. MA: Winthrop Press. Nie, Norman H., Sidney Verba, and John R. Petrocik. The Changing American Voter. Cambridge, MA: Harvard University Press. Niemi, Richard G. and M. Kent Jennings. 1991. “Issues and Inheritance in the Formation of Party Identification.” American Journal of Political Science. 35: 4, 970-988. Page, Benjamin 1. 1977. “Elections and Social Choice: The State of Evidence.” American Journal of Political Science. 21: 639-668. 154 Palfrey, Thomas K. and Keith T. Poole. 1987. “The Relationship between Information, Ideology and Voting Behavior.” American Journal of Political Science. 31: 3, 511-530. Patterson, Thomas. 1990. Out of Order. New York: Knopf. Pomper, Gerald M. 1972. “From Confusion to Clarity: Issues and American Voters, 1956-1968.” American Political Science Review. 66: 415-428. Ray, Leonard. 2003. “When Parties Matter: The Conditional Influence of Party Positions on Voter Opinions about European Integration.” The Journal of Politics. 65: 4, 978-994. RePass, David E. 1971. “Issue Salience and Party Choice.” American Political Science Review. 65: 389-400. Roberts, Thomas and Sean Gibbons. 2004. “Same-sex Marriage Bans Winning on State Ballots.” CNN. com, http://www.cnn.com/2004/ALLPOLITICS/1. l/02/ballot.samesexmarriage/index. htrnl. Schulman, Mark and Gerald Pomper. 1975. “Variability in Electoral Behavior: Longitudinal Perspectives from Causal Modeling.” American Journal of Political Science. 19: 1-18. Sears, David O. 1993. “Symbolic Politics: A Socio-Psychological Theory.” In: Shanto Iyengar and W. J. McGuire (Eds), Explorations in Political Psychology. Durham, NC: Duke University Press. Sears, David O. and Jack Citrin. 1985. Tax Revolt: Something for Nothing in California. Cambridge, MA: Harvard University Press. Sears, David 0., Carl P. Hensler and Leslie K. Speer. 1979. “Whites’ Opposition to ‘Busing’: Self-Interest or Symbolic Politics? American Political Science Review. 73: 369-384. Sears, David O., Leonie Huddie and L. G. Schaffer. 1986. “A Schematic Variant of Symbolic Politics Theory, as Applied to Racial and Gender Equality. In: R. R. Lau and David 0. Sears (Eds), Political Cognition. Hillsdale, NJ: Lawrence Erlbaum. Sears, David 0., Richard R. Lau, Tom R. Tyler and Harris M. Allen Jr. 1980. “Self- Interest vs. Symbolic Politics in Policy Attitudes and Presidential Voting. American Political Science Review. 74: 670-684. 155 Sears, David 0., Tom R. Tyler, Jack Citrin, and Donald R. Kinder. 1978. “Political System Support and Public Response to the Energy Crisis. American Journal of Political Science. 22: 56-82. Sharp, Carol and Milton Lodge. 1985. “Partisan and Ideological Belief-Systems: Do They Differ?” Political Behavior. 7: 147-166. Sniderman, Paul M. 1993. “The New Look in Public Opinion Research.” In Political Science: The State of the Discipline 11. Ed. by Ada W. Finifter. American Political Science Association. ’ Sniderman, Paul M., Michael G. Hagen, Philip E. Tetlock, and Henry E. Brady. 1991. “Reasoning Chains.” In Paul M. Sniderman, R. A. Brody and Philip E. Tetlock (Eds), Reasoning and Choice: Explorations in Political Psychology. New York: Cambridge University Press. 8033, Joe, Laura Langbein, and Alan R. Metelko. 2003. “Why Do White Americans Support the Death Penalty?” The Journal of Politics. 65: 2, 397-421. Sribney, William, Vince Wiggins, and David Drukker. 1999. “Negative and Missing R- squared for ZSLS/IV.” STA TA.c0m. Retrievied March 25, 2008 from http://www.stata.com/support/faqs/stat/2sls.html Stimson, James A. 2004. Tides of Consent: How Public Opinion Shapes American Politics. Cambridge, UK; NY: Cambridge University Press. Stoutenborough, James W., Donald P. Haider-Markel and Mahalley D. Allen. 2006. “Reassessing the Impact of Supreme Court Decisions on Public Opinion: Gay Civil Rights Cases.” Political Research Quarterly. 59: 3, 419-433. Sullivan, Andrew. 2003. “The Conservative Case for Gay Marriage.” Time. com. Retrieved March 25, 2008, from http://www.time.com/time/magazine/article/0,9l 71,1 101030630-460232,00.html. Sullivan, John L., James E. Piereson and George E. Marcus. 1978. “Ideological Constraint in the Mass Public: A Methodological Critique and Some New Findings.” American Journal of Political Science. 22: 233-249. Tedin, Kent L. 1980. “Assessing Peer and Parent Influence on Adolescent Political Attitudes.” American Journal of Political Science. 24: 1, 136-154. Tedin, Kent L. 1994. “Self-Interest, Symbolic Values, and the Financial Equalization of the Public Schools.” The Journal of Politics. 56: 3, 628-649. 156 Tedin, Kent L., Richard E. Matland, and Gregory R. Weiher. 2001. “Age, Race, Self- Interest, and Financing Public Schools through Referenda.” The Journal of Politics. 63: 1, 270-294. Verba, Sidney and Norman H. Nie. 1972. Participation in America. New York: Harper and Row. Wilcox, Clyde and Barbara Norrander. 2002. “Of Moods and Morals: The Dynamics of Opinion on Abortion and Gay Rights.” In Understanding Public Opinion, 2nd ed., eds. Barbara Norrander and Clyde Wilcox. Washington: Congressional Quarterly Press, pp. 121-148. 157 l11111111111111111