Absent Mandate? the Role of Positional Issues in Canadian Elections

Absent Mandate? The Role of Positional Issues in Canadian Elections

Yannick Dufresne

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Political Science University of Toronto

Absent Mandate? The Role of Positional Issues in Canadian Elections

Yannick Dufresne Doctor of Philosophy Graduate Department of Political Science University of Toronto 2015

Positional issues are political issues that generate disagreement among citizens and divide the electorate.

Voting theories conventionally assume that positional issues have little eﬀect on how individuals vote and that parties have little incentive to develop issue-based electoral strategies. Such a characterization of electoral dynamics is particularly prevalent in Canada. An enduring view holds that Canadian parties engage in “brokerage politics” wherein they avoid taking clear policy stances in order to avoid alienating segments of a fragmented electorate.

This thesis challenges the conventional characterization of Canadian electoral dynamics by revisiting the role and impact of positional issues on electoral strategy. First, the results show that once measurement accuracy is improved and the heterogeneity of issue eﬀects is taken into consideration, positional issues have non-negligible eﬀects on individual vote choice. Second, the thesis demonstrates that the behaviour of Canadian political parties does not fully correspond to the brokerage party model. Parties can and do engage in behaviour prescribed by political marketing theory, to which positional issues are central. Positional issues are at the core of an exchange between citizens who care about a small set of issues and political parties who target these citizens using micro-campaigns. Ultimately, this thesis shows that political parties can use positional issues to increase their room for electoral manoeuvre.

ii Acknowledgements

I am deeply indebted to my adviser Neil Nevitte for the transformative role he has played in my life over the last seven years. Dr. Nevitte once told me that he thinks of his function as transferring his knowledge from his mind to the minds of his students. After much patience and perseverance, I hope that he feels that all the time and effort he has invested in my academic development has been worthwhile. To say that he has influenced me as a political scientist would sell short the impact he has had on my development. More than any other person, Dr. Nevitte has changed the way I read, think, and write. His rigour shall remain an inspiration and an example to which I will aspire for the rest of my career. I also wish to thank the other members of my committee: Drs. Richard Nadeau and Peter Loewen. Dr. Nadeau has been a mentor and a model to me for nearly a decade. I will never forget the uncondi- tional support and confidence he has offered me since I was his Master’s student many years ago. Dr. Loewen has provided me with invaluable advice as my project developed from an intuition to a set of coherent ideas. He is one of the most talented young academics I have had the fortune to encounter and there is no doubt in my mind that his name on my doctoral thesis will be a matter of pride for years to come. I am greatly indebted to the internal and external reviewers of my dissertation, Drs. Nelson Wiseman and Eric´ Bélanger. Dr. Wiseman’s influence on my development as a scholar of Canadian politics began early and has persisted in shaping the way I understand the Canadian political landscape. His guidance during my years as a teaching assistant for his Canadian politics course provided me invaluable pedagogical training that I am certain will be an asset to me all my career. Moreover, Dr. Eric´ Bélanger’s careful and insightful evaluation of my thesis gave it the final pushes it needed to become a work of which I am wholeheartedly proud. Finally, I want to thank my colleagues and friends, old and new. Attempting to list all their names runs the risk of forgetting some. It would be an omission for which I would not be able to forgive myself. But all of them know who they are and how grateful I am for their encouragement and support. This is what friends are for. I have saved the last word of acknowledgment for my family, whose role in my life has been, and continues to be, immense. Merci.

iii Contents

Contents iv

List of Figures viii

List of Tables x

1 Introduction 1 1.1 Research Question ...... 4 1.2 Case Selection: Canada and Brokerage Politics ...... 7 1.3 Data ...... 9 1.4 Thesis Overview ...... 11

2 The Citizen Competence Problem in Voting Theories 16 2.1 From the Democratic Citizen to the Incompetent Voter ...... 21 2.1.1 The Rationality of the Classical Democratic Citizen ...... 22 2.1.2 The Columbia School: The Predetermination of the Self ...... 24 2.1.3 The Michigan School: Partisan First ...... 26 2.2 Giving the Voters a Break? Nuance in the Columbia and Michigan Models ...... 29

iv CONTENTS

2.2.1 Loosening Issue-Voting Requirements ...... 30 2.2.1.1 Information shortcuts: Making the Most of Limited Infor- mation ...... 31 2.2.1.2 Retrospective Voting and Valence issues ...... 32 2.2.1.3 Heterogeneity and Issue Publics ...... 34 2.2.2 Measures and Conceptualization ...... 37 2.3 Discussion ...... 41

3 Political Marketing: In Search of Room for Manoeuvre 43 3.1 The Advent of Political Marketing: Segmentation and Positioning . . . . . 44 3.1.1 What is Political Marketing? ...... 46 3.2 A Changing World? Post-Industrial Elections ...... 53 3.2.1 Convergence or Fragmentation? ...... 55 3.2.2 A Recent Increase in Issue Voting? ...... 59 3.3 Discussion ...... 61

4 A New Look at an Old Concept: Conceptualization, Measurement and Eﬀects of Positional Issues 65 4.1 The Conceptual Status of Positional Issues: Ontological and Causal Con- fusion ...... 66 4.1.1 The Quest for the Essence of Vote: Rival concepts ...... 66 4.1.1.1 Party Identiﬁcation: Causal Confusion ...... 68 4.1.1.2 Values: Ontological Confusion ...... 71 4.1.2 Messy Territory: Conceptualizing Positional Issues ...... 74 4.2 Hypotheses ...... 78 4.3 Methodology and Data ...... 79

v CONTENTS

4.4 Results ...... 80 4.5 Discussion ...... 88

5 Public Issues or Issue Publics? The Distribution of Genuine Political Attitudes 91 5.1 Heterogeneity and the Citizen Competence Problem ...... 94 5.1.1 A Black-And-White Electorate: Elite-Based Heterogeneity . . . . . 95 5.1.2 A Fragmented Electorate: Salience-Based Heterogeneity ...... 97 5.2 Hypotheses ...... 102 5.3 Methodology and Data ...... 102 5.4 Results ...... 104 5.5 Discussion ...... 108

6 Detecting Potential for Growth: Immigrants as Targetable Segments 112 6.1 Theory ...... 114 6.1.1 Immigrantness and the Welcoming Eﬀect ...... 114 6.1.2 Minority Issues and the Alienation of Racial Minority Voters . . . . 116 6.1.3 Cross-Pressured Targetable Segments ...... 118 6.2 Hypotheses ...... 122 6.3 Methodology and Data ...... 122 6.4 Results ...... 124 6.4.1 A Welcoming Eﬀect? ...... 126 6.4.2 Immigrantness Versus Race ...... 129 6.4.3 Racial Minorities as Targeted Cross-Pressured Voters ...... 132 6.5 Discussion ...... 140

vi CONTENTS

7 Digging Deeper: The Micro-Targeting of Jewish and Muslim Voters 142

8 Conclusion 150 8.1 Macro- and Micro-Campaigns ...... 153 8.2 Potential for Future Research ...... 156 8.3 Normative Implications ...... 160

Appendix: Figures 161

Appendix: Tables 174

References 197

vii List of Figures

2.1 The Funnel of Causality ...... 28

3.1 The Political Marketing Exchange Process ...... 50 3.2 Three Types of Party Orientation ...... 52

4.1 The Various Foci of Issue Typologies ...... 77 4.2 Stability of Issue Attitudes ...... 82 4.3 Effect of Issue Items and Scales ...... 84 4.4 Effect of Issues, Values and Partisan Identifications ...... 85 4.5 Evolution of Issue Effects for Each Party ...... 87

5.1 Accessibility and Stability of Issue Attitudes ...... 106 5.2 The Eﬀect of Issue Salience: Issue publics? ...... 107 5.3 Elites or Issue Publics? ...... 109

6.1 Minorities’ Partisan Support Over Time ...... 125 6.2 Independent Eﬀects of Immigrantness and Race on the Vote ...... 131 6.3 Immigrant and Racial Minorities’ Ideological Positions and Resistance to the Conservatives ...... 134

viii LIST OF FIGURES

6.4 Riding-Level Eﬀect of Issue-Public Proximity on Being a New Conservative Voter...... 139

7.1 Jewish and Muslim Voters’ Ideology and Propensity to Vote Conservative . 145

1 Effect of issues on vote choice (Breakdown by issues) ...... 162 2 Effect of issues, values and partisan identifications on vote choice (Break- down by values and party identifications) ...... 163 3 Evolution of voters’ most important issue ...... 164 4 The effect of voters’ most important issue ...... 165 5 Increase in Issue Impact? (1) ...... 166 6 Increase in Issue Impact? (2) ...... 167 7 How Citizens Perceive Parties’ Stands on Minority Matters ...... 168 8 Minorities’ Partisan Support Through Time (Three main parties) . . . . . 169 9 Effect of Positional Issues on Issue publics in “very ethnic” targeted ridings 170 10 East-Asian and Hispanic Canadians’ Ideological Positions and Resistance to Conservatives ...... 171 11 Evolution of the effect of being Jewish/Muslim on the vote for the Conser- vatives ...... 172 12 Evolution of the effect of being Jewish/Muslim on the vote for the Liberals 172 13 Evolution of the effect of being Jewish/Muslim on the vote for the NDP . . 173

ix List of Tables

4.1 Eight Positional Issues ...... 81

6.1 The Welcoming-Party Eﬀect: A Bonus for the Party in Power? ...... 128 6.2 Who Shifted? The New Conservative Voters of the 2011 Election ...... 133 6.3 Targeted Ridings: Race, Proximity on All Issues, or Issue Publics? . . . . . 138

7.1 The Canadian Jewish and Muslim Vote (1968-2011) ...... 146

1 United States vs. Canada: The effect of scales on explained variance . . . . 174 2 Issue scales information (1) ...... 175 3 Issue scales information (2) ...... 176 4 Issue scales information (3) ...... 177 5 Issue scales information (4) ...... 178 6 Scaling information (5) ...... 179 7 Evolution of the Effect of the Economy Issue Position on Attitudes Toward Political Parties ...... 180 8 Evolution of the Effect of the Environment Issue Position on Attitudes Toward Political Parties ...... 181

x LIST OF TABLES

9 Evolution of the Effect of the Foreign/US Relations Issue Position on At- titudes Toward Political Parties ...... 182 10 Evolution of the Effect of Law and Order Issue Position on Attitudes To- ward Political Parties ...... 183 11 Evolution of the Effect of Minority Issues Position on Attitudes Toward Political Parties ...... 184 12 Evolution of the Effect of Moral Issues Position on Attitudes Toward Po- litical Parties ...... 185 13 Evolution of the Effect of Social Programs Issue Position on Attitudes Toward Political Parties ...... 186 14 Evolution of the Effect of Women Issues Position on Attitudes Toward Political Parties ...... 187 15 Accessibility and Attitude Stability on the Same-Sex Marriage Issue . . . . 188 16 The Effect of Issue Salience: Full Regression Results ...... 189 17 The Effect of Immigrantness and Ethnicity on the Liberal Vote ...... 190 18 The Effect of Immigrantness and Ethnicity on the Conservative Vote . . . 191 19 The Effect of Immigrantness and Ethnicity on the NDP Vote ...... 192 20 Who Shifted? The New Conservative Voters in the 2011 Election . . . . . 193 21 Targeted Ridings: Race, Proximity on All Issues, or Targeted Issue Publics?194 22 Left-right scale information ...... 195 23 Jewish and Muslim Effect on Conservative Vote: Religious or Issue-Public? 196

xi Chapter 1

Introduction

There is a persisting idea in Canada that the major Canadian political parties engage in brokerage style politics. In a brokerage politics system, parties appeal to different voters by not pronouncing themselves on divisive issues that might exacerbate tense social cleavages (Carty, Young and Cross, 2000; Clarke et al., 1984). As a result, parties end up competing for the same policy space and the same voters, and few ideological differences are developed between them. The brokerage politics perspective views Canadian parties as “centrist, non-ideological election machines” (Bickerton and Gagnon, 2004a, 241). Electoral choice in such a context is influenced largely by leader images and valence issues (Clarke et al., 1996). Valence issues, unlike positional issues, are defined by their non-polarizing nature. They are issues upon which everyone agrees—such as the need for a good economy and clean environment—and do not address the means by which these ends should be achieved. As such, brokerage parties do not fit the rational party model inherent in the classical theory of democracy (Downs, 1957). They do not develop positions on issues in an attempt to target the median voter. Rather, brokerage parties avoid taking any positions

1 on issues at all. Yet classic democratic theory expects citizens to make vote choices based on parties’ positions on issues. If parties do not articulate clear issue positions, how can political issues aﬀect the voting behaviour of citizens? The implication of the brokerage politics model seems to be that positional issues cannot aﬀect voting behaviour in any substantial way.

This thesis argues that, contrary to the brokerage model, positional issues not only matter to Canadian voters, but are also fundamental tools in the strategic toolboxes of political parties. Positional issues can affect the voting decisions of some citizens and can thus be used by parties to increase their room for manoeuvre in attaining their electoral goals. It is a well-known fact that the major Canadian political parties now use political marketing tools to segment the electorate and target specific groups of voters (Flanagan, 2011; Goodyear-Grant, 2013; Marland, Giasson and Lees-Marshment, 2012; Paréand Berger, 2008; Watt, 2006). Positional issues are at the centre of this type of micro-targeting strategy:

Microtargeting allows candidates to surgically deliver diﬀerent messages to diﬀerent constituencies, thus expanding the arsenal of potential wedge issues that can be used in the campaign. With direct mail, email, telephone calls, text messaging, Web advertising, and the like, candidates can narrowcast issue messages to some voters even if others in their coalition might disagree or consider the issue less important. (Hillygus and Shields, 2008, 151)

Evidently, such behaviour contrasts with the expected behaviour of parties in a brokerage politics system. How are positional issue-based strategies reconcilable with the assumption of traditional voting models that issues have little impact on the vote?

The main argument of this thesis is that the potentially impactful role of positional issues

2 in Canadian politics has been overlooked. Most voting behaviour scholarship developed in reaction to the ﬁnding that citizens were falling short of their expected roles in classic democratic theory (Saris and Sniderman, 2004). Citizens lack of political competence shifted academic focus away from electoral strategies and toward more essential or structural determinants of the vote, such as sociodemographics, values, party identiﬁcation, and the state of the economy (Campbell et al., 1960; Fiorina, 1981; Lazarsfeld, Berel- son and Gaudet, 1948). Such perspectives have been fruitful, as they have shed light on the long-term factors that explain the greatest portion of voting behaviour. Perspectives focusing on short-term factors in light of citizen incompetence have looked to cognitively- tractable factors such as valence issues and leader evaluations.

Theoretical models describing party behaviour, such as the brokerage model, also developed with the assumption that positional issues play a minimal role in political behaviour. The argument presented in this thesis stipulates that if one is interested in examining the narrow room for electoral manoeuvre that parties might have, one must consider positional issues. Though positional issues might only explain a small percentage of overall electoral change, small percentages can be decisive.

Approaching the subject from this perspective, the thesis argues that the study of the effect of positional issues on the vote has suffered from three factors: imprecise measurement, unfitting conceptualization, and, more generally, a lack of theoretical interest for marketing-based electoral strategies used by parties on the field. The thesis argues that better measurement of positional issues, the consideration of the possible heterogeneity of their effects, and the use of more appropriate data to observe micro-phenomena can provide new insights on how parties can use positional issues to their electoral advantages. An underlying aim of this thesis is to reconcile more traditional voting models with po-

3 litical marketing approaches. A ﬁrst step in reaching this aim is clarifying that the two lines of reasoning have diﬀerent academic goals. While traditional voting models focus on the macro-level factors impacting elections, political marketing models focus on parties’ abilities to generate room for manoeuvre by also running micro-campaigns. Nevertheless, the two can serve to inform each other in consequential ways.

1.1 Research Question

In an endeavour to redress the historical lack of attention paid to positional issues, this thesis aims to answer the following questions: Can positional issues matter? If so, for whom? And under what conditions? This thesis examines both the eﬀects of positional issues on voting intentions and the roles that positional issues play in parties’ electoral strategies.

There are both theoretical and empirical reasons for why the effects of positional issues might have been overlooked in Canada. First, positional issues have been traditionally operationalized and measured in ways that might have led to the underestimation of their effects. By accounting for this measurement error, it might be possible to observe more stable and consistent issue attitudes than those that have been captured by previous Canadian electoral studies. And by using multi-item issue scales, it might be possible to observe much stronger effects of issues on vote choice. Once properly measured, the effects of positional issues might even compare with those of conventional vote predictors, such as values and party identification. Second, how positional issue effects operate might have been conceptualized in ways that obscured their strength. Central to this thesis is the finding that positional issue effects can operate heterogeneously. An effect can be said to

4 operate heterogeneously if its impact on an individual with a certain characteristic depends on another characteristic of that individual, such as her level of political sophistication or the salience she accords to the issue under investigation. Heterogeneity of this kind is observed when, for instance, the effect on vote choice of an individual’s position on an issue is dependent on the importance of that issue to the individual in general. This perspective echoes the issue-publics perspective, which assumes that voters care about, and are thus affected by, only a small set of issues that are important to them (Krosnick, 1990). If we disaggregate the population by issue importance, the effects that arise from examining issue publics individually could cumulatively have an effect on vote choice, even if the effect of issue positions in the aggregate (without accounting for issue publics) is minimal.1 It is this idea that underlies the concept of salience-based heterogeneity presented in this thesis.

This thesis demonstrates that accounting for measurement error and the heterogeneity of effects leads to the observation of significantly greater positional issue effects on vote choice. These findings add some nuance to the theoretical frameworks inherited from the Columbia and Michigan schools, which hold that the homogeneous effect of positional issue is negligible. However, they support and contribute to the political marketing perspective, which assumes that voter preferences and party strategies have an impact on electoral outcomes (O’Shaughnessy, 1990; Scammell, 1999). Political marketing focuses on the interaction, or exchange, between voters and political parties. In such an interaction,

1It is important to distinguish between issue publics and public issues. The latter concept refers to widely-shared concerns of the day. Single, highly-salient issues have played a crucial role in some Canadian elections. For instance, the Free Trade Agreement with the United States was a critical issue in the 1988 Canadian federal election (Johnston et al., 1992), health care was critical in 2000 (Nadeau, B´elangerand P´etry,2010), and corruption and accountability were critical in 2006 (Andrew, Maioni and Soroka, 2006). In fact, the Free Trade Agreement with the United States in 1988 is considered a classic example of a highly important public issue in a Canadian federal election (Johnston et al., 1992). However, the same public issues are not necessarily salient to voters from one election to the next.

5 positional issues are key elements in party’s toolboxes and are used to attract or mobilize voters.

In order for parties to be able to use positional issues, these issues must affect citizens’ voting behaviour in a substantial way. This thesis shows that some positional issues can affect some citizens greatly and that it is thus possible for parties to capitalize on positional issues in their quest for electoral success. The thesis also shows that the actions parties take to capitalize on positional issues, as described by the political marketing literature (see, for instance, Baines, 1999), are justified. According to the political marketing literature, the development of new market research technologies has led political parties to increasingly micro-target heterogeneous segments of the population (Hillygus and Shields, 2008). Such micro-targeting is more impactful if issue effects actually operate heterogeneously. This thesis explains party behaviour by arguing that issue effects can indeed operate heterogeneously.

The idea that Canadian political parties use positional issues to target segments of the electorate runs contrary to the conventional view of electoral strategies in a brokerage party system (Carty, 2013). Indeed, the main deﬁning characteristic of a brokerage party is that it avoids taking clear stances on issues:

A brokerage party system deﬁnes national problems in the vaguest and most general of terms and produces band-aid politicies to deal with them. It is often the preferred strategy of parties during election campaigns to avoid commit- ment to speciﬁc solutions of any depth [...] (Clarke et al., 1984, 181)

Yet positional issues are not the most important determinants of electoral results. This thesis does not challenge the consensus that long-term structural and institutional eﬀects, as well as cognitively-tractable short-term eﬀects, like valence issues, matter more. Issue

6 effects might be, as Holbrook (1996) argues, substantively similar to ‘campaign effects’, which, even when they are as low as 3%, can have a real impact on election results: “[...] although national conditions set the parameters of likely outcomes and contribute more to the eventual outcome, the campaign can provide the votes that swing the outcome one way or the other.” (Holbrook, 1996, 158) Positional issues might not always swing an election, but focusing on the dynamics of the electorate and on parties’ potential for growth opens up new research avenues that can reinvigorate the study of political strategy. The idea that positional issues have systematic effects on electoral politics offers great potential to political research by relaxing some of the constraints that have historically impeded the empirical study of electoral strategy and, more broadly, challenged the assumptions of political marketing. If citizens actually have stable and consistent issue attitudes, then positional issues have the potential to render closer links between voter behaviour and party behaviour.

By focusing on the relationship between positional issues, voter behaviour and party behaviour, it might be possible to uncover why a party with traditional moral values like the Conservative party of Canada, for example, took a stance in 2006 on the issue of gay marriage but avoided taking a stance on abortion. Is this kind of party behaviour based on myths, or does the nature of public opinion justify such campaign strategies?

1.2 Case Selection: Canada and Brokerage Politics

Canada is a particularly appropriate setting for the investigation of the relevance of issues for the simple reason that it is a tough case: positional issues should not really matter in Canada. The conventional wisdom is that positional issues do not explain electoral

7 behaviour in the country because a peculiar political context discourages mainstream political parties from taking strong stands on issues (Clarke et al., 1996). Different reasons have been offered to explain why brokerage politics developed in Canada. One line of reasoning is based on the idea that the country has long been defined by its racial, cultural and regional heterogeneity (see, for instance, Bourassa, 1902; Mallory, 1954; Siegfried, 1907). If parties were to take clear and unambiguous positions on issues, they would risk alienating segments of the highly fragmented Canadian electorate or affecting national unity. Indeed, in Canada, a politics of accommodation originated from the perceived need for “regional brokerage” (Carty, Young and Cross, 2000, 17), a strategy some consider optimal for winning elections (Clarkson, 2005). Emerging from that line of reasoning is the idea that Canada is particularly susceptible to valence politics, in which policy debates revolve around the ‘who’ and ‘what’ rather than the ‘how’ (Clarke, Scotto and Kornberg, 2010). Political parties all campaign on the same uncontentious valence issues (Stokes, 1963, 1992). At the same time, uninformed voters minimize their electoral calculations by voting mainly on perceived competence of the parties (Petrocik, 1996) or on incumbent performance and leader evaluations (Fiorina, 1981), thereby exacerbating the irrelevance of positional issues. Some scholars contend that the sociodemographic characteristics and values of Canadians also preclude issues from mattering (Blais et al., 2002). The argument presented here contests the view that voters do not have stable issue preferences to which parties can appeal.

A brokerage politics context and a lack of citizen competence are not the only reasons for which positional issues have been understudied. Positional issues have also been understudied because they are idiosyncratic to each country and thus not always amenable to comparative evaluations. The idiosyncratic nature of positional issues stands in contrast to the ubiquitous nature of valence issues, which are better suited for comparative

8 investigation (see Lewis-Beck, 1988; Powell, Jr. and Whitten, 1993). Positional issues are often heavily dependent on specific “cultural” contexts (Leege et al., 2002). Issues such as race in the United States, Turkish integration in the European Union, the asylum seekers debate in Australia, and the Aboriginal question in Canada are all difficult to examine cross-nationally. The solution might be to move up the ladder of abstraction (see Sartori, 1970) by developing concepts with less defined attributes, such as ‘tolerance toward ethnic minorities’. Such an approach can shed light on some research questions, like that examining the global evolution of values in post-industrialized countries (see, for instance, Inglehart, 1977). However, it is difficult to defend the idea that nothing can be gained by moving down the ladder of abstraction and developing more precise concepts that retain the contextually-derived symbolic dimensions of issues. The potential impact of issues on the vote cannot be divorced from the way that issues are understood by voters. By not taking into account the unique cultural values inherent to different contexts, we might forego insight into important psychological processes. For instance, without understanding the cultural values of a country, we might not understand why issue framing might heterogeneously affect a demographically similar group of people (Gamson and Modigliani, 1987).

1.3 Data

This thesis uses data drawn from all available Canadian Election Study (CES) investigations from 1965 to 2011. The ﬁrst two empirical chapters address the measurement of positional issues and the heterogeneity of their eﬀects on vote choice. The empirical evidence for these chapters relies primarily on the 2004 and 2008 Canadian Election Stud- ies (CES). These elections are both recent and are not considered to be positional-issue

9 oriented (see Gidengil et al., 2012). Looking for effects of positional issues in contexts in which they are less likely to be highly determinant seems like a good place to start. The CES datasets have several advantages: they include numerous questions probing respondents’ issue attitudes. They also include a set of conventional controls such as those for sociodemographic factors, party identification, and political sophistication. The 2004 and 2008 CES also include panel studies, which make it possible to explore the stability of issue attitudes over time. Moreover, the CES question wordings of key items were crafted with a concern for cross-time comparability. This last characteristic is particularly important when pooling data in order to capture heterogeneous effects in larger samples. Pooling these data facilitates the use of group sub-sampling and multiple interactions in statistical models, which can be used to detect heterogeneous effects (see Maxwell and Delaney, 2004).

The CES data are also supplemented with data from the 2011 Canadian Federal Vote Compass (VC). There are many reasons for using VC data. First, party positions on thirty positional issues are all measured on the same five-point scale, thus allowing for direct comparison and better dimensional analysis. Party positions were determined through analyses of party platforms and other publicly-available documents, as well as through discussion with party representatives. More importantly, the VC questionnaire was designed specifically to probe issue attitudes while, at the same time, including conventional sociodemographic variables and retaining a sample size of more than one million respondents. Large samples permit a fine-grained segmentation of the population, such as by riding, immigration group, or consumption of specific media channels. Thus, VC represents an invaluable source of data for studying the heterogeneity of the electorate. Nevertheless, there are issues related to sample selection bias that must be considered. As with other web-based surveys, Vote Compass respondents are not selected in the random

10 way prescribed by classic survey theory (Couper, 2008). However, several methods now exist that allow us to adjust for such shortcomings (see, for example, Alvarez et al., 2011; Wang et al., 2014). When necessary, this thesis employs demographic weights built from 2006 Canadian census data. The use of this new kind of self-selected ‘big data’ might foster suspicion or doubt in some researchers, but as King (2011) notes: “in many areas of the social sciences, new forms and quantities of information may well make dramatic progress possible” (King, 2011, 721). VC data have already been proven useful in measuring the effects of different media events that occured during the 2011 Canadian election (Dufresne et al., 2012). Others have compared VC data to more traditional online surveys and confirmed the usefulness of VC data for explaining public opinion (Fournier, Gélineau and Harell, 2014). The CES and VC data are ultimately compatible data sources that, when combined, are highly valuable for political behaviour research.

1.4 Thesis Overview

The thesis proceeds as follows. The next two chapters present an overview of the literature on positional issues and thereby set out the theoretical groundwork upon which the thesis is based. The role of this literature review is to demonstrate that, contrary to the conventional characterization of Canadian electoral dynamics, positional issues have a non-negligible impact on voting behaviour. As such, Canadian political parties have an incentive to run micro-campaigns that exploit positional issue effects. Chapter 2 engages in a more specific examination of the reasons why one might expect positional issues to be more or less meaningful than party identification, ideology, and values. It investigates how the lack of citizen competence has rendered a situation in which it is difficult for positional issues to affect voters’ choices. The citizen competence problem holds that

11 the vast majority of voters are not well informed about political issues. If citizens are not informed about issues, it is difficult for issues to affect their voting decisions and it is difficult for parties to use issues in their electoral strategies. Chapter 3 describes political marketing theory and the way it reconceptualizes the relationship between voters and political parties so as to redeem the relevance of positional issues. Contrary to the brokerage politics model, the political marketing approach assumes that parties engage in electoral strategies that very often involve positional issues. Political marketing does not endeavour to supplant or contradict traditional voting and party theories. The two perspectives simply have different academic goals. While traditional theories focus on uncovering the fundamental structure of the vote, political marketing theory focuses on investigating the narrow room for electoral manoeuvre that parties might have.

The next four chapters empirically investigate first, the possibility that positional issues may have a substantial impact on voting behaviour and second, that political parties both use micro-targeting to increase their support among specific groups and are justified in doing so. To begin, Chapter 4 demonstrates how the redemption of positional issue effects is both possible and justifiable. It explores how reconceptualizing positional issues and measuring them in novel ways can expose their full impact. The difficulty in defining the concepts inherent to issues, compared to the relative ease of defining concepts inherent to values and ideology, is discussed. The distinction between valence and positional issue voting is then clarified and disentangled from different typologies in the literature. The principle argument of the chapter is that the effects and stability of positional issues have often been underestimated because of measurement issues. The use of multi-item scales strengthens the effects of positional issues at levels comparable to those of party identification and arguably to those of values.

12 Chapter 5 further investigates how positional issues can matter by examining to whom they might matter. To do so, it examines how genuine political attitudes are distributed in the electorate. It is possible that the effect of positional issues is heterogeneous: the same issues might have different impacts on different voters. But there are two types of heterogeneity to consider: elite-based heterogeneity and salience-based heterogeneity.1 Elite-based heterogeneity views the electorate as being divided into two groups: politically sophisticated elites and politically unsophisticated masses. Positional issues are considered to have a significant impact on political elites and a minimal impact on the uninformed masses. Salience-based heterogeneity assumes a more complex distribution of political attitudes. It takes as its point of departure the “issue-publics” hypothesis (Krosnick, 1990), which posits that different voters care about different issues. It then demonstrates how different voters are differently affected by different issues. The analyses show that the importance accorded to issues can increase the effect these issues have on the vote to a greater degree than can other measures of political sophistication. These results provide some support for the segmentability of the electorate by political parties.

The last two empirical chapters (Chapters 6 and 7) examine the practical implications of the previous ﬁndings for party strategy. They investigate whether the micro-targeting of electoral segments like immigrant and racial minorities is a viable electoral strategy. The argument is that the increased complexity of the electorate combined with the developement of targeting technologies led to the development of two parallel electoral campaigns: a macro- and a micro-campaign. While the former conforms to the brokerage politics model, the latter relies heavily on the impact of positional issues. It is for that rea-

1We might consider a third type of heterogeneity—“cognition-based” heterogeneity—, which captures the diﬀerent ways people organize information and understand politics. However, this type of heterogeneity seems to be quite amorphous and is not considered in this thesis. For more information about this type of heterogenity, see Bartle (2005).

13 son that positional issues should perhaps be considered fundamental elements in parties’ toolboxes.

The thesis concludes with a discussion suggesting that positional issues have been understudied in Canada for two main reasons. The first reason is that citizens were long thought to be unable to reason about positional issues sufficiently to allow such issues to impact their vote decisions. The second reason is the enduring idea that Canadian electoral dynamics exist in a context of brokerage politics reinforced by party discipline, where positional issues are set aside principally in favour of valence issues. In order to remedy this inattention, this thesis examines how reconceptualizing and re-operationalizing positional issues might redeem their impact on voting behaviour. The results show that once positional issues are more accurately measured and once the heterogeneity of their effects is taken into consideration, positional issues have non-negligible effects on individual vote choice. Different voters care about different political issues. When precisely measured, positional issues are revealed to matter to the voters that care about them, or to whom they are salient. In light of these findings, the thesis explores whether parties are justified in engaging in positional issue-based micro-targeting strategies. It first establishes that Canadian political parties indeed employ micro-targeting tactics and thereby do not fully correspond to the brokerage party model. Second, it demonstrates that because different positional issues matter to and affect different segments of the population, political parties have good reason to engage in the tactics prescribed by political marketing theory. Positional issues are thus found to be at the core of an exchange between citizens who care about a small set of issues and political parties who target these citizens using micro-campaigns. Political parties can leverage the knowledge that positional issues affect the behaviour of those who care about them in order to increase their room for electoral manoeuvre. In sum, this thesis aims to offer a new perspective on positional issues and

14 to clarify why and how they can matter to both citizens and political parties.

15 Chapter 2

The Citizen Competence Problem in Voting Theories

The question of the relevance of issues is consequential. Issues are central to political parties’ strategies to attract voters. If voters had no significant attitudes about positional issues, party platforms and campaign promises would be irrelevant. Some of the academic literature suggests that this might indeed be the case. The Columbia and Michigan schools—pioneers in the field of electoral behaviour—argue that voters do not meet classic democratic citizenship standards because they do not have significant attitudes about issues (Berelson, Lazarsfeld and McPhee, 1954). The cumulative empirical findings in the academic literature demonstrate that voters’ choices are more constrained by fundamental factors, such as the state of the economy, sociodemographics, and party attachments, than they are by voters thoughts on issues. As a result, electoral studies from the last few decades have focused on the study of the structural, institutional, and social forces that stabilize electoral dynamics (see Lewis-Beck et al., 2008).

16 This chapter reviews how classical voting theories leave very little room for political parties to impact electoral outcomes, and consequently, little room for positional issues to matter. Such a perspective is apparent in the Columbia school’s Minimal Thesis, which argues that campaigns do not matter (Lazarsfeld, Berelson and Gaudet, 1948), as well as in the Michigan school’s Funnel of Causality (Campbell et al., 1960). In these theories, voting patterns are considered to be so stable and so engrained that it is almost possible to predict electoral outcomes ahead of time. Indeed, both the normal vote perspective (Converse, 1966) and the predictability of electoral campaigns perspective (Gelman and King, 1993) assume that political parties have very little latitude in determining their electoral fate. From these vantage points, studying positional issues or other short-term eﬀects is at best unimportant and at worst completely fruitless.

The citizen competence problem is identified in this chapter as the principle factor explaining the relative lack of scholarly attention given to positional issues and party strategies. The first findings regarding citizen competence were disappointing, if not outright shock- ing. Although Lazarsfeld, Berelson and Gaudet (1948) were expecting to detect the effect of voters’ personalities and media exposure on vote choice, their data instead revealed the decisive influence of social group membership: “[...] a person thinks, politically, as he is, socially. Social characteristics determine political preferences” (p.69). This pessimism about citizens’ competence was fueled by the publication of two subsequent articles by Philip Converse, who demonstrates the disappointing level of political attitude stability (1962) and ideological constraint (1964) in the population. Converse was puzzled by how voter behaviour did not seem to fit classical models of democracy. Classical models of democracy expect the most knowledgeable citizens to be the most likely to shift their political positions upon receiving new political information. Yet Converse (1962) found that it was in fact less knowledgeable citizens—those with only middling levels of

17 information—who are most likely to shift their positions from one election to another. Further research led him to conclude that a very large proportion of citizens exhibit a lack of ideological constraint (Converse, 1964) and form their attitudes about political issues randomly (Converse, 1970). Butler and Stokes (1975) replicated these ﬁndings in the British context, arguing that individual voters have a limited understanding of the left-right ideological spectrum.

This unﬂattering portrait of the electorate has driven generations of researchers to turn their backs on issue attitudes and party strategy in order to examine the more fundamental underpinnings of vote choice (see Kuklinski and Peyton, 2007). Analyses of the fundamental underpinnings of the vote have signiﬁcantly contributed to our understanding of voting behaviour. The micro-sociological approach of the Columbia school was followed by the psychosocial approach of the Michigan school and the macro-sociological approach of researchers such as Lipset and Rokkan (1967). All these approaches explain voting behaviour by minimizing citizens’ individual input; voters’ choices are constrained by social groups (Berelson, Lazarsfeld and McPhee, 1954; Katz and Lazarsfeld, 1955; Lazarsfeld, Berelson and Gaudet, 1948), by psychological ties to parties, acquired through socialization (Campbell et al., 1960; Miller and Shanks, 1996), or by structural factors emerging from historical events such as the industrial revolution (Lipset and Rokkan, 1967). More recently, scholars have uncovered the potential of other structural factors, such as the state of the economy (Fiorina, 1981; Lewis-Beck, 1988) and genetics (Bell, Schermer and Vernon, 2009; Dawes and Fowler, 2009; Hatemi et al., 2010), to explain voting behaviour.

The eﬀects of positional issues on vote choice cannot compare in size to the eﬀects of more fundamental factors discovered through decades of research. However, there are

18 some factors that suggest that the effects of positional issues and campaign strategies in general might have been underestimated. First, the original studies of the Columbia school are much more nuanced with regards to the effects of electoral campaigns than the label ‘minimal effects’ implies. Nowadays, it is perhaps difficult to understand the surprise of the authors of The People’s Choice (1948) when their central hypotheses regarding campaign effects were not supported by the data. However, while only (sic) 8% of voters are said to have been converted by a candidate during the electoral campaign, 84% are shown to have been somehow influenced by the campaign (Lazarsfeld, Berelson and Gaudet, 1948, 103).1 By today’s standards, few would consider these numbers as supporting the well-known motto ‘campaigns do not matter,’ even though they suggest that a campaign can determine the winner of the election. Similarly, Philip Converse recently complained about the way the conclusions of his 1964 article have too often been exaggerated in the literature: “I disavow any reading of the Belief Systems essay that concludes that most citizens lack political attitudes. I think there is a limited stratum–10 percent? 20 percent?–with a very sparse complement of such attitudes, mostly because of an aversion to the whole subject of politics” (Converse, 2007, 149).

Other reasons for why positional issues might have been underestimated might be related to measurement and conceptual issues. Some scholars have shown that the effects of positional issues increase once measurement problems are addressed (Achen, 1975; An- solabehere, Rodden and Snyder Jr., 2008). Others argue that the effects of positional issues might be heterogenous: individual voters care about different issues and thus might be differently affected by different issues (Bartle, 2005). It is uncertain whether media consumption contributes to or simply reflects this fragmentation of the electorate, but

1Lazarsfeld, Berelson and Gaudet (1948) enumerate five different effects of the 1940 Erie County electoral campaign: 1) Reinforcement (53%); 2) Activation (14%); 3) Reconversion (3%); 4) Partial reconversion (6%); and 5) Conversion (8%).

19 the salience of issues varies across audiences and is reﬂected in voting patterns (Baum, 2002; DellaVigna and Kaplan, 2007). Finally, it might be that the eﬀects of positional issues were underestimated because researchers, working within a classical conception of democracy, were not looking for them.

The effects of positional issues have the potential to stimulate research on campaign strategies. Even if fundamental factors determine most of political behaviour, the remaining variance should not be dismissed. Any remaining variance, no matter how small, represents room for political parties to impact their electoral fates. Positional issues represent the link between parties and different types of voters and should thus not be dismissed. Nevertheless, constraints on the effects of issue-based campaign efforts are frequently discussed in the literature. A principle constraint is the citizen competence problem. However, recent works show that positional issues are in fact used by parties to alter voters’ perceptions of their ideological positions (van der Brug, 2004), to attract cross-pressured partisans (Hillygus and Shields, 2008), to distract voters from the state of the economy (Vavreck, 2009), or to highlight particular personal characteristics of candidates (Jacobs and Shapiro, 1994). Some scholars even suggest that issues might be the main cause of conversion from one party to another (Luskin, McIver and Carmines, 1989). This role of issues does not contradict the Columbia school model. Indeed, even if social factors structure electoral dynamics, “social groups need to be treated as live social forces, not static categories” (Blais et al., 2002, 96). The increasing fragmentation of the media space and the development of new communication technologies offers new opportunities for political parties to create micro-campaigns aimed at reaching and influencing different types of voters (Baines, 1999). In constrast with the conception of the traditional macro-campaign, positional issues could be considered as being at the heart of micro-campaigns.

20 This chapter is divided in two main sections. The first section reviews classic voting theories and shows how their development in and application to the Canadian context was influenced by the citizen competence problem. The second section discusses the different suggestions that have been made to improve the measurement of positional issues and the different works that loosen the conditions under which issues can matter.

2.1 From the Democratic Citizen to the Incompetent

Voter

Various theories support the claim that positional issues do not matter. Some authors focus on the nature of campaigns and media coverage to defend this claim. In these cases, the media and political parties are considered responsible for issues not mattering. Some argue, for example, that media attention is more often directed at the electoral horse race (Norris et al., 1999), party image (Egan, 1999) and leaders (Mendelsohn, 1996) than it is at issues. Other authors argue that the eﬀect of positional issues is limited because parties either take similar or ambiguous positions on issues (Page and Brody, 1972). Alvarez (1998) argues that party ambiguity on issues increases voters’ uncertainty and thus decreases voters’ inclinations to vote on the basis of issues. However, most inﬂuential theories of voting behaviour put the responsibility for issues not mattering on citizens.

The key to understanding the limited interest in the study of positional issues is the citizen competence problem: the idea that voters fail to meet their expected role in a democratic society. There are cumulative ﬁndings showing that voters lack political information and are not ideologically constrained (Butler and Stokes, 1975; Converse, 1964; Delli Carpini

21 and Keeter, 1996). This line of reasoning represents an elite-centric perception of the mass public, wherein only a handful of citizens possess substantial political information. One version of the elite-centric perspective suggests that “individuals never think for themselves” (Zaller, 1992, 287) or that “the voice of the people is but an echo” (Key, 1961, 2). As such, only a small proportion of the population is expected to base its vote decisions on issue positions (Goren, 1997; Neuman, 1986). Issues just cannot matter very much in models portraying citizens as “muddle-headed” or “empty-headed” (Saris and Sniderman, 2004, 337). The pessimism concerning voters’ ability to reason politically can be traced back to the origins of modern electoral studies.

2.1.1 The Rationality of the Classical Democratic Citizen

In its simplest form, democracy implies that citizens vote for politicians who share their political preferences. Yet when interpreted literally, this simple task ends up being very demanding for individual voters. As Berelson, Lazarsfeld and McPhee (1954) point out, for individual citizens to meet their expected role in a classic democratic system, they must be politically interested, motivated, knowledgeable, principled, and rational. In such a case, the link between citizens and political parties rests on maximizing one another’s utility. This view was ﬁrst formalized in Anthony Downs’ (1957) Economic Theory of Democracy, which conceptualized voters and political parties as rational actors. This rational view also provided the foundation for a spatial theory of voting (Enelow and Hinich, 1984).

Downs’ economic explanation of voting behaviour is derived from the classical conception of democracy and developed into the rational-choice theory of voting. The theory is based on the idea that electoral behaviour can be explained with the same models that are used

22 to study the economic market. Once the rationality of political parties and voters is assumed, the model has the ability to explain and predict the behaviour of these actors. For the proponents of this theory, there are few, if any, differences between voters and consumers and between political parties and commercial companies (Schlesinger, 1991). Voters, like consumers, seek to maximize their utility, and political parties, like companies, seek to maximize their profits, in this case electoral gains (Downs, 1957, 295-296). Thus, political parties take positions on issues that allow them to gather the largest vote share: “[...] parties formulate policies in order to win elections, rather than win elections in order to formulate policies” (Downs, 1957, 28). In the case that the democratic political system offers a certain level of consistency and differentiation between options, rational citizens can fulfill their classic democratic roles and vote according to their preferences. These assumptions enable the spatial representation of parties and voters’ positions in a given electoral context.

To represent strategic positions, Downs (1957) uses a linear scale that aggregates economic issue positions of both voters and parties. This scale allows for the estimation of the optimal position of parties and, consequently, for the prediction of their relative positionings. More precisely, in a context in which voter distribution is unimodal, Downs (1957) predicts that parties will converge toward the same position. This line of reasoning is further developed by the spatial theory literature. Some scholars expanded the model’s unidimensional aspect (Stokes, 1963). In a multidimensional ideological space, the best position for a party would be at the junction of all spatial dimensions’ averages (Hol- combe, 1989). Such a median voter position minimizes the sum of the distances from all voters. However, the median voter is not considered an optimal position by all scholars in the spatial theory literature. Rabinowitz and Macdonald (1989), for instance, contend that voters do not seek to maximize their proximity to political parties. Instead, they are

23 often more likely to support parties that promise to shift policies in their own preferred ideological direction. For example, if an incumbent is implementing policies that are considered too conservative by a moderate conservative voter, that voter might vote for a progressive party to shift governance in his preferred direction, even if he is ideologically closer to the conservative party.

Despite attempts to reﬁne Anthony Downs’ model of voting, its main weakness remains: the rational voter assumption does not ﬁt the empirical evidence presented by the Columbia school. As Lazarsfeld, Berelson and Gaudet (1948) put it: “The notion that people who switch parties during the campaign are mainly reasoned, thoughtful, conscientious people who were convinced by issues of the election is just plainly wrong” (69).1 The criticisms of this assumption motivated more than 50 years of research aimed at questioning and understanding the citizen competence problem (Kuklinski and Peyton, 2007).

2.1.2 The Columbia School: The Predetermination of the Self

The Columbia school sees individual choice as principally influenced by the sociodemographic group to which the individual belongs: “[...] a person thinks, politically, as he is, socially. Social characteristics determine political preferences.” (Lazarsfeld, Berelson and Gaudet, 1948, 69) This conclusion follows the finding that voters’ personalities and their exposure to media have very little influence on vote choice. According to the proponents

1Many in the literature also criticize rational-choice theory for its failure to explain the most basic questions. For instance, Uhlaner (1989) and Blais (2000) both point out the inability of model to explain why most citizens actually do vote: “[...] unfortunately for the theory, many people do vote” (Blais, 2000, 2). Green and Shapiro (1996) oﬀer a more general criticism of rational-choice theory and its application in political science. According to these authors, rational-choice suﬀers from 1) post-hoc theoretical development; 2) the absence of empirical tests; and 3) selection bias.

24 of the Columbia school, voters are so socially bound that their behaviour can be explained by three sets of variables: 1) Socio-economic status; 2) Religion; and 3) Area of residence. These indicators constitute what Lazarsfeld, Berelson and Gaudet (1948) name the political predispositions of individuals. Therefore, the act of voting is not considered by these scholars as an individual act, but a social one: ”People vote, not only with their social group, but also for it” (Lazarsfeld, Berelson and Gaudet, 1948, 148). This explains why the Columbia school is also known as the micro-sociological approach.

The findings of the early Columbia school studies necessarily discourage the study of campaign strategies and positional issues. In this perspective, the study of intra-group sociological processes is more useful to understanding political behaviour. From the Columbia school findings emerged many important concepts used in modern political communications theory such as selective attention and two-step-flow communications. The latter concept holds that the flow of political information moves from social groups’ opinion leaders to less politically active voters (Katz, 1957; Katz and Lazarsfeld, 1955). Such a process reinforces the maintenance of intra-group homogeneity and the persistence of social group differentiation (Berelson, Lazarsfeld and McPhee, 1954).

The Columbia school does not consider electoral campaigns useless. Their inﬂuence was just found to be so secondary that related factors such as positional issues and campaign strategies did not merit much attention. Nevertheless, the Columbia school model does not well explain the variation in party vote share from one election to the next. The sociological approach can also not explain why the state of the economy aﬀects electoral results or why some people do not vote according to their political predispositions. It is in reaction to this perceived static aspect of the Columbia school model that the Michigan school model developed.

25 2.1.3 The Michigan School: Partisan First

The outline of the Michigan school model of voting behaviour was ﬁrst presented in The American Voter (Campbell et al., 1960). The keystone of the Michigan school model is party identiﬁcation; a concept that adds dynamism to the sociological model while at the same time addressing the citizen competence problem.

The main distinction between the Columbia and Michigan schools resides in the importance accorded to psychology in the vote decision-making process. The core of Michigan school model rests on psychological theories that aim to explain individuals’ aﬀective orientations toward groups:

Both reference group theory and small-group studies of influence have con- verged upon the attracting or repelling quality of the group as the generalized dimension most critical in defining the individual-group relationship, and it is this dimension that we call identification. (Campbell et al., 1960, 121)

Partisanship is psychological and its strength varies across individuals. This emotional attachment to a party is acquired through a socialization process. Miller and Shanks (1996) compare this process to the way individuals are socialized into a religion by their environment. Voters have different levels of attachment to their party just as people have different degrees of religiosity. The strength of such emotional links is heavily influenced by sociodemographic and environmental factors (Campbell et al., 1960). Some even suggest that partisanship is a form of social identification like that proposed by the Columbia model (Green, Palmquist and Schickler, 2002). Berelson, Lazarsfeld and McPhee (1954) have already suggested that voters’ partisan allegiances might drive their ideologies and political positions. The causal order implied by the Michigan school is thus the com-

26 plete opposite of the order described by classical democratic theory, wherein voters’ issue positions drive their support for parties.

A crucial element of the Michigan model is the conceptual distinction between partisanship and vote choice. This distinction means that a voter’s psychological inclination toward a party does not necessarily translate into a vote for that party. Such a distinction adds the dynamic aspect to the Michigan model that is missing in the Columbia model. The Michigan school perspective sees vote choice as less predetermined than does the Columbia school perspective. Like other long-term factors that influence behaviour, such as sociodemographics, religion and values, partisanship might be difficult to alter. But the vote can nevertheless be influenced by short-term factors specific to an election (Campbell et al., 1960; Lewis-Beck et al., 2008; Miller and Shanks, 1996). If short-term factors such as the state of the economy, positions on issues, or the quality of candidates disappoint a voter enough, he or she might decide to vote for another party or simply not to vote, especially if the voter’s partisan attachment is relatively weak. However, according to the proponents of the Michigan school, such disloyal behaviour does not affect the voter’s fundamental partisanship (Green and Shapiro, 1996). The Funnel of Causality (see Fig- ure 2.1) conveys the idea that vote decision is the result of the influence of subsequent factors, the most determinant of which are long-term factors such as sociodemographics and partisanship.

Given the great inﬂuence of the Columbia and Michigan schools on voting behaviour research, it is not surprising that little attention has been paid to the study of positional issues. Research conducted with these approaches has focused on uncovering the fundamental factors underpinning voting behaviour. The Funnel of Causality illustrates well the quest for the structural root of the vote: a person’s vote is inﬂuenced by his or her

27 Figure 2.1: The Funnel of Causality

Time

Socio-Demographics

Party Identification

Issues

Candidates

VOTE

Note: Figure taken from The American Voter Revisited (Lewis-Beck et al., 2008, 23). party identification, which is itself influenced by sociodemographic factors. In an era when citizens were revealed as incapable of fulfilling their democratic roles, it is understandable that scholars such as those belonging to the Michigan School were focused on uncovering structures that stabilize the democratic system.

Not all scholars agree with the assumptions underlying the Michigan model. Some scholars note the diﬃculty of exporting the model outside of the American context, to multi- party systems like Canada (Budge, Crewe and Farlie, 1976; Leduc et al., 1984; Meisel, 1975).1 Others argue that the model might be less relevant today given a recent weakening

1Many studies show that party identiﬁcation is less inﬂuential in Canada than in the United States.

28 of partisan attachment in the United States and elsewhere (Dalton, 1984). But, more importantly here, Fiorina (1981) and many others argue that the model exaggerates the inalterability or the causal priority of partisanship (see also Franklin, 1992; Kiewiet and Rivers, 1985). The next section discusses the reasons why the eﬀects of short-term factors like positional issues might have been underestimated.

2.2 Giving the Voters a Break? Nuance in the

Columbia and Michigan Models

The core of the citizen competence problem is that voters cannot vote according to political issues if they cannot grasp the substance of politics. Many scholars now agree that the citizen competence problem might have been overstated (see Kuklinski and Peyton, 2007). Even Converse (1975) complains that the conclusions of his 1964 article are too often simplistically cited: “They are frequently misinterpreted as saying that not much of anybody has public opinions about much of anything. This is a disastrous misconstruc- tion, for it ﬁts no data at all” (Converse, 1975, 83). But, as Saris and Sniderman (2004) put it: “The dominant themes of two generations of research have been that citizens tend to be muddle-headed (the lack of constraint theme), empty headed (the non-attitude

Early Canadian electoral studies even claim that the concept of party identification “may be almost in- applicable in Canada” (Meisel, 1975, 67). Or that the Canadian electorate is composed of an important number of ‘flexible partisans’ that are more susceptible to short-term forces (Clarke et al., 1996). However, the Canadian exceptionalism regarding weak partisanship and, more specifically, the reconceptualization of party identification based on the distinction between flexible and stable partisans have attracted different criticisms. Most of these criticisms state that Clarke et al. (1996) typology suffers from measurement and conceptualization problems that tend to exaggerate the levels of volatility in Canada: 1) Wiseman (1986) observes that the typology is based on recall information, subject to bias; 2) Johnston (1992) argues that the exclusion from the party identification question of a ‘none’ category produces significant measurement error; and 3) Blake (1982) argues that considering inconsistent party identification between levels of government should not be considered as a measure of flexibility. In Canada, the lack of symmetry in federal and provincial party systems does not necessarily make for unstable voting within level.

29 theme) or both” (Saris and Sniderman, 2004, 337-338). There are two major critiques of these dominant research lines. The ﬁrst is methodological, questioning mostly the measurement of political information. The second is theoretical and endeavours to loosen the requirements for making enlightened political choices.

2.2.1 Loosening Issue-Voting Requirements

The classic democratic citizen is a super-voter that is expected to have genuine attitudes on many diﬀerent political issues and to know party positions on these issues. Given these high requirements, it is no wonder that the Columbia and Michigan schools were able to declare most citizens politically incompetent: “people’s motivation is weak if not absent [...] a sense of ﬁtness is a more striking feature of political preference than reason and calculation” (Berelson, Lazarsfeld and McPhee, 1954, 308-311). If voters show so little political competence, how could their attitudes on positional issues matter?

Despite the citizen competence problem, many citizens do manage to make vote decisions. In fact, Key (1961) points out that people generally tend to vote rationally even if their political ideas are fuzzy. Why might that be the case? Some scholars argue that the expectations imposed on voters might have been exaggerated: “if ‘attitudes’ are taken to mean logically consistent summary evaluations of any conceivable political object [...] then it seems clear to me that even splendidly well-informed, attentive citizens will routinely ﬂunk the test” (Bartels, 2003, 63). The American Voter. As Margolis (1977) points out: “Although voters’ positions on the issues did not provide a very powerful explanation of actual votes cast, the relationship between individuals’ positions on the issues and their votes was by no means random” (Margolis, 1977, 31). Indeed, minimal attitudes, involving meagre thought and information, are probably more common

30 than Converse’s non-attitudes (Luskin, 1987). Also, it has been shown that citizens vary ‘wisely’ in their degree of political sophistication; voters might not be fully informed, but they can nonetheless make smart use of their limited information (Saris and Sniderman, 2004). As Key Jr. (1966) states: “Voters are not fools” (p.7). This use of limited information might explain in part why Carmines and Stimson (1980) observe widespread issue voting within the electorate despite observing relatively low levels of political information and involvement. Ultimately, efforts to mitigate the citizen competence problem can be grouped into three categories: 1) The use of heuristics or information shortcuts; 2) The evaluation of incumbent’s and other candidates’ competences with regards to valence issues; and 3) Issue publics and a focus on a small set of positional issues. These three lines of research all loosen the competence requirements for issues to matter to voters. The third category, which holds that different people care about different issues, underpins the idea of the heterogeneity of effects presented in this thesis.

2.2.1.1 Information shortcuts: Making the Most of Limited Information

Information shorcuts counter voters’ lack of political information (Sniderman, Brody and Tetlock, 1991). Instead of conforming to the ideal of democratic citizenship, voters resort to psychological cues to avoid having to acquire complex political information. Although voters do not have extensive knowledge about politics, they still ﬁnd ways to reason about parties, candidates, and issues (Popkin, 1991).1

Even the most informed voters use psychological shortcuts. The limitations of human thought preclude people from acquiring the kind of political information described in nor-

1Downs (1957) posits the idea of information shortcuts in his rational theory of democracy. It might very well be rational for individuals to ignore things they cannot change and thus be politically ignorant. Indeed, Downs views political ideologies as shortcuts: “With this short cut a voter can save himself the cost of being informed upon a wider range of issues” (Downs, 1957, 98).

31 mative democratic theories (Bartels, 2002). Some scholars argue that group cues provided during a campaign allow less-informed voters to understand the position of their social group on issues and vote accordingly (Lupia, 1994, 1998). Even voters’ gut-level likes and dislikes can ﬁll the gaps between less-informed and more-informed voters (Brady and Sni- derman, 1985; Popkin, 1991; Sniderman, Brody and Tetlock, 1991).1 But others are less certain. For instance, Kuklinski and Quirk (2000) show that heuristics do not perfectly compensate for a lack of information and can introduce bias. Nevertheless, information shortcuts perspectives oﬀer convincing theoretical and empirical evidence that voters can hold political attitudes without being unrealistically politically sophisticated.

2.2.1.2 Retrospective Voting and Valence issues

Other ways have also been proposed to explain how voters make the best use of their imperfect political information. One possiblity is that citizens simply vote according to perceived performance of the incumbent administration (Key Jr., 1966). The ‘reward- punishment’ theory of elections boils citizens’ democratic burden down to the decision about whether or not to ‘throw the rascals out’ (Fiorina, 1981). Long before Fiorina (1981), Downs (1957) also mentioned such retrospective voting as a cost-cutting mech- anism for rational but uninformed voters. Voters do consider the incumbents’ economic performance while in oﬃce when making a vote decision (Lewis-Beck and Stegmaier,

1The literature on information heuristics is vast and different views exist about what qualifies as an information shortcut. For instance, some scholars consider ideologies and values as shortcuts; Hinich and Munger (1994) see ideology as reducing the transaction cost of political exchange, as it reduces uncertainty and information costs. However, considering ideologies or values as ‘cues’ might be tautological, depending on how these concepts are defined. Ideologies and values can be thought of as summaries of preferences as much as they can be thought of as determinants of preferences. The definitional distinction between external and internal cues requires further investigation. Internal cues such as ideologies are not consistent with Popkin’s (1991) external conception of heuristics. Popkin (1991) conceptualizes heuristics as external cues that exploit campaign information to help voters reach conclusions about which candidate will best serve their interests.

32 2009). The ideas posited by Fiorina triggered an entire literature about the eﬀects of economic performance on retrospective evaluations of actors and, ultimately, vote choice (see Lewis-Beck and Paldam, 2000; van der Brug, van der Eijk and Franklin, 2007).

The idea of retrospective economic voting is appealing: uninformed voters just have to look to their pocketbooks when deciding whether or not to vote for the governing party. However, the empirical evidence presents a more complex story. Some scholars have found that prospective evaluations of economic performance can matter as much to vote decisions than retrospective evaluations of an incumbent’s past economic performance (Kuklinski and West, 1981; Nadeau and Lewis-Beck, 2001). In a prospective evaluation, a citizen evaluates the incumbent’s ability to improve or maintain a good economy in the future. Such a consideration is obviously more cognitively demanding than simply remembering or noticing the current state of the economy. The ﬁnding that perhaps runs the most counter to individualistic economic logic is that voters assign more weight to collective economic interests than personal economic interests when evaluating a party or candidate (Kinder and Kiewiet, 1979). Collectively, observations about economic voting highlight the fact that voters do care about the competence of their governors.

Diﬀerent theories in the literature assess citizens’ perceptions of competence on vote choice. These theories extend beyond economic competency and are not restricted to incumbent evaluations. Numerous studies document the importance of valence issues on vote choice (Ansolabehere, 2000; Clarke et al., 2010, 2009; Stokes, 1992). Valence issues, as opposed to positional issues (or ‘pro-con issues’), are deﬁned by Stokes (1963) as issues on which almost no voter disagrees (e.g., economic well-being, accessible health care, protection from terrorism, and honesty in government). Political parties can focus on these consensual issues without alienating any segment of the electorate (Clarke, Kornberg

33 and Scotto, 2008). In turn, uninformed voters can limit their reasoning to just selecting the best candidate to deal with valence issues. Brokerage politics models encourage this kind of limited reasoning as political parties within brokerage systems campaign on valence issues instead of pronouncing themselves on contentious positional issues.

2.2.1.3 Heterogeneity and Issue Publics

Most attempts to deal with the citizen competence problem, like information shortcut and retrospective voting theories, recognize that positional issues’ direct impact on vote choice is negligible. They try to explain how voters are able to cast votes that correspond to their preferences without holding genuine attitudes on positional issues. Most works that account for a lack of political sophistication also criticize the classic democratic citizen proposed by spatial theory models (see Kuklinski and Quirk, 2000).1 Indeed, in revisiting The American Voter (1960), Lewis-Beck et al. (2008) enumerate different reasons for which we should not expect issues to matter to vote choice. Among these reasons are voters’ ‘internal’ limitations, including general political competence, familiarity with issues, and ability to perceive party differences on issues. If a voter is unable to cognitively grasp an issue or its implications, he or she is unlikely to develop a position on that issue and use it to make a vote choice. There are also ‘external limitations’ on the effects of issues. For instance, if many different issues are given equal media attention during a campaign, the explanatory effect of a single issue on the vote might be imperceptible. However, most

1It must be noted that an effort has been made by spatial theory scholars to adapt their models. For instance, Rabinowitz and Macdonald (1989) concede that citizens do not have clear positions on issues. Instead, voters have diffused preferences and choose candidates that advocate the clearest stance in the direction that voters prefer. Another attempt to scale down spatial theory models’ assumptions is found in the work of Hinich and Munger (1994), who argue that what matters in spatial theory is ideology, not issues. Finally, Merrill and Grofman (1999) present a ‘unified model’ that conceptualizes ‘candidate acceptability’ by being a directional model with proximity constraint.

34 of these limitations assume that each issue aﬀects every individual in the same way, or, as articulated by spatial theory, that all issues matter equally for each voter. But what if some issues matter only for some voters? This concept is the driving force behind the issue-publics hypothesis.

Some scholars argue that voters can only care and be knowledgeable about a small set of issues, and that different voters care about different issues (Krosnick, 1990). According to Krosnick, knowledge about specific issues differs depending on the voter, and voters are more receptive to parties’ positions on issues they care about. Issue publics comprise the citizens that are alert, attentive, interested and informed about a given issue (Key, 1961). If such issue publics exist, political knowledge might be more specific than generalized. Some voters might know a lot about one issue without knowing much about other issues. Citizens’ motivations to acquire knowledge might be triggered by specific interests. Some scholars argue that issues need to be considered important to voters to influence their vote (Rabinowitz, Prothro and Jacoby, 1982). Others contend that people are able to translate their policy preferences into voting choices on issues that are important to them (RePass, 1971). If voters care about a variety of different issues, it is not surprising that the effect of issues has not been detected in aggregate homogeneous voting models. Even proponents of the Michigan school admit that knowledge about issues is not evenly distributed in the population: “We still need to determine why familiarity with a single issue or a common set of issues varies from one citizen to the next” (Lewis-Beck et al., 2008, 171).1

The issue publics idea is important because it relaxes the conditions for issue voting im- plicit in spatial models and criticisms of issue voting. Voters do not have to know all

1However, some scholars argue that political knowledge is mostly a general, and not a speciﬁc, phenomenon. These scholars also observe that “[...] more than a small fraction of the public is reasonably well informed about politics—informed enough to meet high standards of good citizenship” (Delli Carpini and Keeter, 1996, 269).

35 parties’ positions on all issues. They can handle being knowledgeable about a small set of issues that are important to them: “people who consider a policy attitude to be important are indeed more likely than those who consider it unimportant to perceive a substantial diﬀerence between competing candidates’ attitudes” (Krosnick, 1990, 66). Stable attitudes on a single issue can exist without ideological consistency. For instance, Jews and Muslims living in Canada might be more informed about the Canadian government’s position on the Israeli issue as it likely triggers more interest in their communities than in other communities. It is worth noting that Converse, who is most often associated with the concept of ‘non-attitudes’, also participated in the development of the ‘issue-public’ concept. In The Nature of Belief Systems in Mass Publics (1964), Converse mentions the possibility of specialized political sophistication: “We do not disclaim the existence of entities that might best be called ‘folk ideologies,’ nor do we deny for a moment that strong diﬀerentiations in a variety of narrower values may be found within subcultures of less educated people” (Converse, 1964, 66).

To summarize, since the early findings of the Columbia and Michigan schools, a great deal of attention has been paid to the citizen competence problem. Different solutions have been proposed to explain how uninformed voters end up making political choices. These include the use of information shortcuts, valence politics, and issue publics.1 Of all solutions, only the one based on the fragmentation of the electorate into different

1The current discussion about the proposed solutions to the citizen competence problem focuses on micro-level or psychological explanations. However, it must be noted that many scholars consider that the solution is in the aggregation of the individual opinions. The main logic is as follows: because most citizens hold ‘random’ attitudes on policy issues (Converse, 1970), these opinions cancel one another out when aggregated and reveal attitudinal consistency (Erikson, MacKuen and Stimson, 2001, 429) or even rationality (Page and Shapiro, 1992). This would explain why, despite individual instability, collective opinion moves in stable, recognizable patterns (Stimson, 1999). The aggregation of opinions or macro- politics would explain economic voting: “It is true that individual Americans have a weak grasp on the essentials of economics and economic policy, and it is also true that Americans, in the aggregate, are highly sensitive to real economic performance” (Erikson, MacKuen and Stimson, 2001, xxi).

36 issue publics considers positional issues to be an important component of vote choice. The literature on information shortcuts usually focuses on how voters can end up making reasonable political choices without holding firm attitudes on issues. By definition, the valence politics model also does not consider positional issues as being important. Yet, as will be shown here, the valence politics model and the idea of issue publics might not be entirely irreconcilable. It is possible that a valence issue-based macro-campaign might coexist with an issue public-based micro-campaign that targets voters sensitive to certain positional issues. More importantly, the issue-publics concept opens the door to the idea of the heterogeneity of issue effects, a concept central to the arguments presented in this thesis but one that is difficult to operationalize. The study of positional issues also faces a variety of methodological and conceptual challenges.

2.2.2 Measures and Conceptualization

Recent advances in measurement (Ansolabehere, Rodden and Snyder Jr., 2008) and the development of the concept of mass public heterogeneity (Bartle, 2005; Goldberg, 2011; Krosnick, 1991) have introduced new ways of thinking about issue eﬀects. As shown below, the methodological solutions to the citizen competence problem are compatible with some of the theoretical responses presented above. Achen (1975), for example, demonstrates that measurement correction could improve the stability and coherence of voters’ political thinking. The idea of the heterogeneity of the mass public is also not new. In 1964, Converse proposed that “[...] we come a step closer to reality when we recognize the fragmentation of the mass public into a plethora of narrower issue publics” (Converse, 1964, 245).

37 Achen (1975) was among the first to challenge the contention that people answer survey questions at random and therefore do not have real attitudes about political matters. Achen showed that both the ‘stability’ and the ‘coherence’ of voters’ political thinking increased when corrected for measurement error. Indeed, some argue that the use of multi-item scales for measuring positional issues might help to uncover their effects on vote choice (Ansolabehere, Rodden and Snyder Jr., 2008). Such a way of measuring and operationalizing issues might be more consistent with current political psychology research. People sometimes misinterpret questions, are not familiar with terminology or have different considerations that frame their answers (Saris and Sniderman, 2004; Tourangeau, Rips and Rasinski, 2000). The use of multiple items serves to correct, in part, these measurement errors. The result is a more abstract measure of positional issues that fits with Bartels’ (2002) conception of an attitude. Indeed, according to Bartels, citizens hold attitudes on issues–that is, psychological tendencies–but do not have elaborated preferences, which are essentialist and discrete expressions.

The debate sparked by measurement theorists is more consequential than it might initially seem, as it implies diﬀerent assumptions about citizens’ reasoning abilities.1 On the one hand, Zaller (1992) rejects the presumption that survey responses are manifestations of underlying ﬁxed attitudes, which can be corrupted by measurement error. On the other hand, scholars like Ansolabehere, Rodden and Snyder Jr. (2008) contend that ”[...] the tension in the empirical literature [on issue voting] stems from measurement error in individual issue items” (Ansolabehere, Rodden and Snyder Jr., 2008, 223). Even in high-quality surveys such as the ANES and GSS, Ansolabehere, Rodden and Snyder Jr. (2008) show that measurement error can explain approximately half of the variance in

1For those interested in this measurement debate, it is striking how the question of issue measurement echoes the debate about whether to include a ‘Don’t know’ option in surveys (see Krosnick, 1991). Both debates involve a disagreement between scholars about whether voters have latent genuine attitudes.

38 responses within a typical issue item. Diﬀerences in preferences are therefore blurred substantively by measurement error. The use of multi-item scales is thus appropriate to correct measurement error (McIver and Carmines, 1981). These scales might oﬀer evidence to support the idea that voters have consistent and stable policy preferences, that issues matter to the electorate, and that the heterogeneity of sophistication among the population matters for issue voting (Ansolabehere, Rodden and Snyder Jr., 2008).

The concept of heterogeneity and its operationalization into models of electoral choice is also a promising avenue for circumventing the citizen competence problem (Bartle, 2005). Heterogeneity is the idea that not all voters reason the same way, and that voters’ political behaviour therefore does not follow the same regularities or decision processes (Rivers, 1988). The concept of heterogeneity can take diﬀerent forms and has been used to criticize the uniform nature of some theoretical models. Some scholars, for example, apply the concept of heterogeneity to spatial theory: “It is possible that some voters evaluate candidates according to the proximity model, whereas others use a directional measure. In other words, heterogeneity among voters is an alternative explanation to individual mixed utilities” (Merrill and Grofman, 1999, 79). Others argue that some heterogeneity exists in the cognitive organization of belief systems (Baldassarri and Goldberg, 2010). This line of research is reminiscent of Axelrod’s Schema Theory, which aims at “providing new insights into how people can and should cope with the highly complex environments typical of political arenas” (Axelrod, 1973, 1249). Similarly, Hochschild (1981) suggests that the lack of consistency found in publics’ belief systems could dissimulate the simple fact that elite standards of ideological consistency do not capture ordinary citizens’ patterns of belief.1 1It needs to be noted that the bases of heterogeneity are not necessarily psychological. Heterogeneity can also be geographically-based. For instance, Gelman et al. (2008) has shown that in the United States the eﬀect of partisanship is much more important in poor states than it is in rich states.

39 It is easy to see that the concept of heterogeneity can render theoretical voting models significantly more complex. To avoid confusion, it is crucial to determine the source of the heterogeneity intended for study. The idea that different voters might be influenced differently by the same issues is not new. Converse’s (1970) Black and White model implicitly considers heterogeneity: a small number of people have political attitudes on issues while others answer at random. Political elites ought therefore to vote more according to their issue positions than do the larger group of uninformed voters. Such elite-based heterogeneity is present in the Canadian electorate (Roy, 2009). From this perspective, the effect of positional issues on the vote is moderated by political information. But is it possible that different types of voters can be influenced by different issues? If so, then issue-based heterogeneity should also be taken into consideration.

Issue-based heterogeneity is intrinsic to the idea of issue publics. As Krosnick (1990) argues, the electorate is fragmented into a multitude of issue publics of different sizes. These groups of voters only care and are only adequately informed about a small set of issues. If Krosnick is correct, it is easy to see why the effect of positional issues might have been underestimated in traditional vote choice models. If moral issues, for instance, are only important to a very small portion of the electorate, their effect on the vote would not appear as strong and significant in homogeneous models. They would still have an impact on the vote, but not one detectable by models that assume that their effects are the same for those who care about moral issues and those who do not.

40 2.3 Discussion

Traditional voting theories do not allow the full potential of positional issue effects to be realised. Since the empirical findings of the Columbia School, research has focused primarily on solving the challenge to democratic normative models caused by the citizen competence problem. This challenge led researchers to pay more attention to fundamental long-term determinants of vote choice than to short-term factors like positional issues. Cumulative empirical evidence established the importance of social group membership and partisanship in structuring voting patterns. Thereafter, the influence of short-term factors, such as the fluctuation of the economy and party competence, have been largely documented (Lewis-Beck, 1988; Petrocik, 1996). Despite the attention given to positional issues in recent years (see Ansolabehere, Rodden and Snyder Jr., 2008; Hillygus and Shields, 2008), this short-term factor (central to the classic democratic model) is still largely neglected in the field of electoral studies. The relative unimportance of positional issues compared to more fundamental factors might partially explain this neglect. The centrality of the citizen competence problem in traditional voting theories might have also contributed to limiting interest in examining the effects of cognitively-demanding positional issues. But recent reconceptualizations and methodological advances in the voting behaviour literature suggest new ways for positional issues to matter. In particular, new developments in measurement theory and the concept of issue-based heterogeneity in the electorate suggest that voters can cogitate, within limits, about positional issues. The literature review presented in this chapter therefore suggests that positional issues may matter. Consequently, it is reasonable to revisit the effects of positional issues on voting behaviour and to re-examine their role in campaign strategies. The next chapter does precisely that and, in doing so, presents a political marketing framework that offers

41 parties issue-based room for manoeuvre.

42 Chapter 3

Political Marketing: In Search of Room for Manoeuvre

The conception of the electorate as heterogeneous is at the core of the political marketing approach. According to political marketing theory, parties benefit from segmenting the heterogeneous electorate, targeting strategic groups and positioning their policy proposals accordingly. This conception clashes with the optimal electoral strategy implied by brokerage politics, which includes avoidance of clear issue positions and focusing on leader competence on vague valence issues (Clarke et al., 1984; Clarkson, 2005). Also, by assum- ing that voters have some attitudes on political issues and that parties have the ability to persuade voters to vote for them, political marketing theory conflicts with much of the previously discussed political behaviour literature (Bartle and Griffiths, 2002). However, conflict between the conventional theories and political marketing seems to be the product of differing research interests. An interest in uncovering the structural essence of voting will lead one to different research foci than an interest in determining the most efficient

43 electoral strategy. The impact of positional issues might have been largely underestimated because most electoral research deals with theoretical frameworks that are not principally interested in electoral strategies.

As discussed in the previous chapter, the voting behaviour literature’s focus on the citizen competence problem leaves little room for party strategy development. Most party behaviour models also do not leave much room for positional issues to matter. Theoretical models such as the brokerage party model focus on what might be called macrocampaigns: uniform campaign strategies that are elaborated to affect the electorate as a whole. But parties have the capacity to target more specific groups of voters, especially with the advent of new communication and polling technologies (West, 2013). If different issues matter differently to different voters (see, for instance, Krosnick, 1990; Margolis, 1977), then political parties have the option of engaging in microcampaigns by which they endeavour to reach specific segments of the electorate. Recent research shows that parties indeed engage in microcampaigning (Gould, 1998; Hillygus, 2007; Watt, 2006). The goal of this chapter is to establish how political marketing might represent a promising theoretical avenue for the study of the effects of positional issues.

3.1 The Advent of Political Marketing: Segmenta-

tion and Positioning

What is the goal of political parties? Some posit that the goal of political parties is simply to win elections and that these organizations are, by deﬁnition, only vehicles to obtain power (Riker, 1962; Schlesinger, 1991). However, not everyone agrees with this plain deﬁnition of political parties’ behaviour. Strøm (1990), for example, distinguishes

44 between vote-seeking, office-seeking, and policy-seeking parties. Vote-seeking and office- seeking parties act in order to maximize their vote share or win public office. Policy- seeking parties endeavour to influence the direction and substance of policy debates and governmental programs and are thus more ideologically-driven. Other scholars contend that political parties might have different goals entirely:

If it is a political party ﬁghting an election campaign, it may want to maximize the number of seats it wins, or, indeed, it may as, say, a Green party, be seeking to inﬂuence the political agenda. [But] by waging campaigns, parties may also seek to serve internal party purposes, like maintaining party unity, attracting new members, fund raising, nurturing potential coalition links and so on. [...] Yet success at elections, and the chance to occupy government positions that it provides, is clearly the core objective of parties and candidates. (Schmitt-Beck and Farrell, 2002, 3-4)

Political parties have different goals. But a more fundamental question is whether parties have adequate means to achieve their goals. If parties do not have the tools to accomplish their goals or influence their electoral fates, there is arguably little point in having goals to begin with. Party behaviour is already heavily constrained by a diversity of factors (Lewis-Beck et al., 2008). Some of these are external contraints, such as the state of the economy, and are out of their control. In the current media-saturated climate, political parties also have little control over the flow of election communication (Holbrook, 1996). Parties also face internal contraints. For instance, some parties’ organizational structures give their members a lot of power, and their members are often more ideological than the rest of the population (Herrnson, 2009; Layman et al., 2010). Financial considerations are also often related to the success of electoral parties (Farrell and Webb, 2001). The

45 main assumption of political marketing is that, despite the constraints parties may face, they ultimately have control over their electoral fates.

3.1.1 What is Political Marketing?

At the core of any attempt to understand the phenomenon of political marketing is, of course, a definitional exercise. Therein lies one of the main shortcomings of political marketing research. Although somewhat developed as a discipline, political marketing is difficult to define. (Henneberg, 2002, 98)

Political marketing as a field of study is growing in importance. This is exemplified by the creation of the Journal of Political Marketing, the annual International Political Marketing Conference, and an increasing number of political marketing books. Yet the subject has received limited attention in Canada (see Marland, Giasson and Lees-Marshment, 2012). It is important to distinguish between the definition of political marketing used outside the field and the definition used by scholars in the international political marketing literature. Political marketing has many distinct features. It is defined both by a set of techniques and as a conceptual framework. Contrary to popular usage, political marketing is not only about the elaboration and study of political communication techniques. It also includes cross-disciplinary analytical tools for studying party and voter behaviour (Scammell, 1999, 718).

Confusion exists regarding the term political marketing. Some authors note that political analysts and journalists tend to use the term simply as a ‘buzzword’ (Henneberg, 2002, 93) or ‘convenient hold-all’ (O’Shaughnessy, 2002, 213). It is also often confused with political communication (Albouy, 1994; Lock and Harris, 1996), as “being simply about

46 advertising, slogan, sound-bites and spin doctors” (Lees-Marshment, 2001, 2), or “used more or less interchangeably with propaganda” (Scammell, 1999, 723). This conception has perhaps been inﬂuenced by earlier narratives about particular political campaigns. For instance, Richard Nixon was said to be sold to the US electorate “like soap powder or a can of beer” (McGinnis, 1968).

Political marketing parallels the Marketing Revolution that occurred in the American business community in the 1950s (see Keith, 1960). The Marketing Revolution followed the development of market research techniques such as polling. These technical advancements made it possible to segment the market and identify the needs and wants of consumers in a more precise way. The practical application of market research techniques led to the growing acceptance of the idea that the key to business success is a shift away from emphasis on production eﬃciency and toward market research and customer satisfaction. Marketing is diﬀerent from product promotion; it places the consumer at the beginning of the production-selling process (Newman, 1994; Steger, Kelly and Wrighton, 2006).1

To have any definitional value, political marketing must refer to the application of a consumer-centric concept to the political realm (O’Shaughnessy, 1990). As Scammell states, “[the] ‘Marketing concept’ is the key to understanding political marketing” (Scam- mell, 1999, 726). This is what distinguishes political marketing from propaganda or political communication. Business management scholars were the first to apply the ideas coming from the business marketing revolution to politics (see Newman, 1994; O’Cass, 1996). The edited volume, Handbook of Political Marketing (1999b) offers detailed de- scriptions of marketing concepts and management applications developed specifically for the political realm. Political marketing started to attract more attention when some of

1A consumer-centric strategy has also been applied outside the business world, for instance, to non- proﬁt organizations (Burnett, 2007; Kotler, 1975; McLeish, 1995).

47 its ideas and techniques were attributed to highly publicized electoral successes such as Bill Clinton’s in 1992 and Tony Blair’s in 1997. Some consider the 1992 American presidential election as the beginning of a “new era in politics” (Newman, 1999a, 27). In that election, Bill Clinton presented himself as a ‘New Democrat’, distancing himself from the traditional Democrat ideology by endorsing more centrist social positions and neoliberal economic principles. This new positioning was thought to be more in line with the average American voter. The British Labour Party rebranded and repositioned its ideology in a similar fashion prior to its 1997 electoral success (McLaughlin, Mungie and Hughes, 2001). The American Democrats and British New Labour even relied on many of the same campaign strategists (Gould, 1998).

There has been some resistance to the idea of applying marketing to politics. For instance, there has been some debate surrounding the conceptualization of the political product. Wellhofer (1990) contends that the unique nature of the political realm makes it incomparable to the business realm. Others claim that the political product is irreconcil- ably different from commercial products.1 The political product has been conceptualized as the composition of two elements that tap into different types of wants and needs: the political offer and the political image (Bernier, 2001; Henneberg, 2002; Newman, 1994, 1999a). The political offer represents the set of promises made by a political party or a candidate. Positional issues are thus at the heart of the political offer. This definition parallels the conception of the political offer presented by works drawing from classical economic models (see Downs, 1957; Schlesinger, 1991; Stokes, 1963). These classic economic

1Consequently, some of the marketing models designed for commercial goods are not directly applicable to politics (See, for instance, the ‘Four Ps’ (or Marketing Mix) in Newman (1994)). The political product is less similar to consumer goods than it is to services offered by lawyers or financial consul- tants (Henneberg, 2002, 111). For this reason, the service branch of marketing, or more specifically relational marketing, is often considered as more relevant to politics (Henneberg and O’Shaughnessy, 2009; Johansen, 2005).

48 works imply the existence of a unique optimal political position to maximize vote share. They also consider the political image an essential component of the political product. Some scholars following this line of reasoning even contend that the political image is more important than the political offer (see Egan, 1999). The importance of image presents an opportunity to adapt brand management to politics (White and de Chernatony, 2002). The two elements of the political offer are interconnected. As Jacobs and Shapiro (1994) show, the priming of positional issues can be used to affect political image. This dual conception of the political product lessens the rigidity of the more economic conception of the term. In that vein, some scholars make the case that political marketing has the potential to reconcile economic and socio-psychological models of electoral behaviour “by not [being] dogmatically committed to any single model of man” (Bartle and Griffiths, 2002, 20). Note that this perspective concedes that the heterogeneity of the electorate lies at the heart of political marketing.

Whether political marketing represents a unified general theory is a matter for debate. Given the complexity of the political environment, it is overly optimistic to expect a general theory of political marketing. There is not even a unified theory of commercial marketing (Grönroos, 1994). Instead, there are various schools of commercial marketing theory, including several postmodernist schools (Brown, 1993). Nevertheless, most political marketing scholars agree with Scammell (1999) that the marketing concept and exchange process constitute the “common core” of the term (see Figure 3.1).1 This exchange process can also be found in some political scientists’ works that describe political parties as ‘market-based organizations’ or as office-seeking instruments (Strøm, 1990). As

1Political marketing scholars deal with more objects of study than just elections (electoral strategy and tactics). Among other things, these scholars examine party financing, e-government and cyber- democracy, media effects, and governance (see, for example, Newman, 1999b). It is difficult to determine whether this multiplication of objects of study is only due to an excess of enthusiasm over a so-called new sub-discipline.

49 Schlesinger suggests, “Elections are a type of political market, in which parties oﬀer their candidates and their policies in exchange for the votes needed to gain oﬃce.” (Schlesinger, 1991, 12)

Figure 3.1: The Political Marketing Exchange Process

STRATEGIC MARKETING Developing the political product POLITICAL VOTERS PARTIES TACTICAL MARKETING Promoting the political product

Even though political marketing borrows the vocabulary and some of the assumptions of rational-choice theory, it cannot be easily described as a variation of rational-choice theory (Savigny, 2004). Least of all, it cannot be described as such because political marketing does not necessarily assume the rationality of political actors. Indeed, even if the political marketing model states that the use of rational pragmatism by a political party tends to produce greater competitiveness, political marketing scholars habitually consider political parties as restrained by some factors such as ideologies, party constitutions, and party activists (Bowler and Farrell, 1992). Non-rational behaviour is integrated in the Lees-Marshment typology (see Figure 3.2). This model presents three types of parties at different stages of marketing optimization. A product-oriented party is a party that refuses to adapt any aspect of its behaviour to voter’s needs and wants. The party’s motivations are strictly ideological and it might be considered as suffering from a lack of pragmatism. Even if the goal of a party was to simply influence the political agenda or

50 promote an ideology, market research can still oﬀer insight into the best way to do so. The sales-oriented party uses tactical marketing to optimize the promotion of its political product. Finally, the market-oriented party engages in both strategic and tactical marketing. Such a party endeavours to not only promote its political ideas, but also to better adapt them to voters’ needs and wants. In order to do so, it harnesses modern technologies to conduct ﬁne-grained market research and elaborate strategies based on collected intelligence. The originality of political marketing lies in the use of innovative market research techniques, such as segmenting and micro-targeting of the population, to elaborate the political product. It is in this way that political marketing can be distinguished from Strøm’s (1991) party typologies. Ultimately, these theoretical constructs provide explanations of the observed variance in political parties’ ‘rational’ behaviours.

More relevant to this thesis are the positional issues that are at the core of market- orientation processes. While parties’ can strategically position on issues in a way that responds to voters’ needs and wants, they must also remain aware of their larger position in the competitive partisan environment. The consistency and credibility of a party’s positioning are important. Some scholars stress the importance of a party’s political product being based on an ideology that is linked to its history and traditions (Bartle and Griﬃths, 2002; Bernier, 1991; Lees-Marshment, 2006). In 2006, the Conservative Party of Canada repositioned itself, but–like Clinton’s Democrats and Britain’s New Labour–never entirely eliminated its core ideology (Watt, 2006). Instead, the party moderated the ideological component of its political product by incorporating some of the most popular elements of the Liberal platform. This strategy, known as triangulation, was popularized by Bill Clinton’s strategist Dick Morris (Worcester and Baines, 2006). Although the Conservative Party moderated its ideology, it still held on to many traditional conservative positions, such as being tough on crime, increasing the defense budget and re-evaluating the same-

51 sex marriage question. In doing so, the Conservatives came to occupy a unique position in the competitive environment while still retaining much of their ideological core. The idea was not to became a copy of the Liberal Party, but to oﬀer an acceptable conservative alternative to the electorate. Ultimately, this strategy increased the Conservative Party’s political competiveness (Par´eand Berger, 2008).

Figure 3.2: Three Types of Party Orientation

PRODUCT-ORIENTED PARTY

PRODUCT

SALES-ORIENTED PARTY

PRODUCT

MARKETING MESSAGE MESSAGE RESEARCH CONCEPTION DIFFUSION

Tactical Marketing (Product Communication)

MARKET-ORIENTED PARTY

MARKETING MARKET SEGMENT POSITIONING MESSAGE MESSAGE RESEARCH SEGMENTATION TARGETING CONCEPTION DIFFUSION

Strategic Marketing Tactical Marketing (Product Conception) (Product Communication)

Note: Adaptation of the Lees-Marshment Model, Lees-Marshment (2001).

Whether the contextual change that allowed for the advent of political marketing should be considered a new phenomenon is debatable. Some enthusiasts describe the use of

52 the marketing concept in political practice as “the political marketing revolution” (Lees- Marshment, 2004). But others disagree. Some authors trace the concept’s roots to the beginning of the twentieth century (Lock and Harris, 1996). Others trace it back even further, identifying similarities between political marketing and the Sophistic pragmatism of ancient eras (Laufer and Paradeise, 1990). Yet other authors argue that political marketing is as old as democracy itself (Perloff, 1999). This discord in the literature might be the result of definitional problems surrounding political marketing. Some of the disagreement might also arise from the failure to distinguish between strategic and tactical marketing. Recall that what distinguishes political marketing from other approaches is the assumption that the electorate is heterogeneous and that it is thus advantageous for political parties to conduct fine-grained market research. Such market research involves parties segmenting and targeting the electorate and repositioning themselves in order to elaborate their political offers. Parties’ capacities to engage in market research has certainly improved with the advent of new communication and polling technologies (West, 2013). Some scholars argue that these technologies have led to an even greater increase in the heterogeneity or fragmentation of the electorate (Aarts and Semetko, 2003; Bennett, 2012; Gimpel, 1999; Pomper, 1996; Schantz, 1996b). Political marketing theory provides a plausible and promising framework through which to explore the heterogeneous effect of positional issues.

3.2 A Changing World? Post-Industrial Elections

Positional issues are generally considered to not matter very much because voters are assumed to lack the requisite information for them to matter. Some scholars have suggested that the system of brokerage politics operating in Canada has exacerbated the seeming

53 irrelevance of issues, as brokerage parties do not take stances on positional issues and thus eﬀectively remove them from the menu of political choice (Clarke et al., 1984). In deeply divided societies, taking positions on issues can risk alienating substantial segments of the electorate. Evidently, if parties do not take clear stances on issues, it becomes even more diﬃcult for them to matter to an already little-informed electorate. But the Canadian political environment has recently experienced some changes that might open the door to positional issues playing a greater role in party strategy (Bittner and Koop, 2013). At the same time, some scholars argue that the voter information problem has waned due to the general increase in citizens’ levels of education. More educated citizens are considered to be better able to deal with positional issues (Dalton, 1988). This phenomenon is consistent with the cognitive mobilization thesis (Dalton, 1984) and the idea that the political instability of the 1960s increased political interest in the population (Miller and Miller, 1975). In addition to rising education levels, the increase in comprehension of politics is also attributed to the greater availability of mediated information (see Klingemann, 1979a,b). Yet other authors identify even deeper changes in voting behaviour; they suggest that new generations of citizens are more likely to have weaker party ties (Butler and Stokes, 1975; Dalton, 1988) and that traditional social cleavages are weakening (Franklin, Mackie and Valen, 1992). These trends have arguably engendered a ‘new political culture,’ characterized by the electorate’s growing interest in politics and the rise in importance of new issues such as the environment, women’s rights, and immigration (Inglehart, 1990). Taken together, these changes suggest that positional issues might have the space today to play a greater role in voting and party behaviour than they have in the past.

54 3.2.1 Convergence or Fragmentation?

A tension persists in the literature between two conceptions of the consequences of this modernization of politics. Some scholars argue that the structural impacts of post- industrialization have a homogeneous eﬀect on society:

The cultural and structural diﬀerences among western countries [...] have de- clined in some respects. The diﬀusion of values, the comparable economic changes, and the development of rapid transportation and almost instan- taneous communication seem to be producing a common western culture. (Nevitte, 1996, 56)

Most structural theories, such as post-materialism (see Inglehart, 1977, 1998), describe a convergence in citizens’ attitudes and behaviour. But the effect of industrialization might be twofold. On the one hand, the fact that many post-industrial countries are experienc- ing a reduction in the stress related to fulfilling basic needs (Abramson and Inglehart, 1995; Inglehart and Abramson, 1994) and an increase in education levels (Dalton, 2007, 1988; Nie, Junn and Stehlik-Barry, 1996) surely led to comparable macro-level patterns of political behaviour and value development. On the other hand, industrialization and modernization might also engender more complexity at the micro-level. According to Dalton: “Electoral politics is moving from cleavages defined by social groups to value and issue cleavages that identify only communities of like-minded individuals” (Dalton, 1988, 174). Indeed, the fragmentation of the media space and the specialization of the workplace, among other things, could favour the development of different political belief networks (Baldassarri and Goldberg, 2010). Others suggest that rising immigration levels have encouraged the “balkanization of the population,” and that this has had an impact on electoral politics (Gimpel, 1999, 115). In Canada, Henderson (2004) argues

55 that interprovincial diﬀerences in political attitudes are either stable or declining. She identiﬁes nine relatively homogeneous cultural clusters that can account for the variations in political attitudes among Canadians. In the commercial world, consumer behaviour is increasingly studied at the micro-level with emphasis on personality traits and lifestyles (Kahle, Beatty and Homer, 1986; Sandy, Gosling and Durant, 2013). The theoretical assumptions of both postmaterialism theory and commercial Value and Life Style theory are based on Maslow’s (1954) concept of a hierarchy of needs. The postmaterialist considerations that emerge in a wealthy society might encourage the fragmentation of citizen interests:

The issue public hypothesis therefore suggests that unanimity among Amer- icans in terms of which policy attitudes are personally important to them is unlikely to have been the case during much of recent history. [...] But when such prominent events or problems [such as a major war or economic depres- sion] do not focus national attention, people are unlikely to focus their passions on the small set of policy debates. (Krosnick, 1990, 74)

It is reasonable to suppose that mass media might well be associated with the mass marketing of politics (Newman, 1999a). Television has long represented the citizens principal source of campaign information (Ansolabehere, Behr and Iyengar, 1993). Recently, web-based news sources and social media have ascended to become critical components of modern campaigns (West, 2013). Bartels (2003) explains how such a political environment shapes voter attitudes. In his view, voter attitudes cannot be considered as preferences; they must be considered as psychological tendencies inﬂuenced by one’s political environment: “[...] the context dependence of preferences is an unavoidable consequence of basic cognitive and evaluative processes” (Bartels, 2003, 64). Iyengar (1991) underscores the

56 importance of context in attitude formation: “While there can be no denying the inﬂuence of stable dispositional characteristics or cultural norms on attributions of responsibility for political issues, there is compelling evidence that contextual or circumstantial factors are equally if not more important [...] the most important of these contextual inﬂuences is television” (Iyengar, 1991, 10). The media of course helps people to make sense of their own experience within a larger society (Mutz, 1994). Consequently, political parties have incentives to develop more thorough communication strategies. As Norris et al. (1999) point out, it is increasingly important for parties to run campaigns that are thematically undistracted.

The increased complexity of the media environment could encourage the fragmentation of the electorate and could increase the importance of issues in electoral politics (Geer, 2006; Stroud, 2008a). The development of a more complex media space multiplies the opportunities for parties to reach voters that are uninterested in, and pay little attention to, politics (Baum and Kernell, 1999; Ladd, 1995). These voters are arguably the most susceptible to new information (Zaller, 1992), but they are also the toughest to reach: “The group which the campaign manager is presumably most eager to reach—the as yet undecided—is the very group which is less likely to read or listen to this propaganda” (Lazarsfeld, Berelson and Gaudet, 1948). On the one hand, the increased complexity of the media space oﬀers uninterested citizens a greater opportunity to avoid political media (Baum and Kernell, 1999; Prior, 2007). In fact, the multiplication of TV and radio channels coupled with the structure of the internet makes it possible for voters to entirely avoid electoral campaign coverage. On the other hand, the popularity of soft news and the proliferation of political advertising represents a new opportunity for politically disengaged segments of the populations to acquire political knowledge. Indeed, both soft news (Bastien, 2013; Baum, 2002) and political advertisements (Ansolabehere

57 and Iyengar, 1995) have been shown to increase political information. And Just (1996) demonstrates that political interviews on popular talk shows usually focus more on issues than on the candidate’s character. Just argues that Bill Clinton increased his presence on talk shows in 1992 in order to focus the debate on the economy and away from the Paula Jones scandal. Similarly, Geer (2006) shows that negative advertisements provide voters with policy information, sourcing for facts, direct quotes, and information that generally holds incumbents accountable. A greater fragmentation of the media space could also potentially result in greater coverage angles or agenda-settings (see, for example, DellaVigna and Kaplan, 2007).1 If most people engage in selective exposure (Stroud, 2008b), then it is plausible to suppose that the increased complexiﬁcation of the media space reinforces the heterogeneity of eﬀects in the electorate.

The fragmentation of postindustrial societies might have in turn caused the fragmentation of the electorate. Dalton (1988) argues that there has been a multiplication of what he calls ‘apartisan’ voters. These voters differ from traditional independent partisans in that they are sophisticated, well-educated, active citizens. They refuse to identify with a party not out of ignorance or unconcern but out of discontent with what the parties offer. These citizens are less likely to vote as predicted by traditional voting models and are more likely to make discriminating, thoughtful voting choices: “citizen voting behavior is more dependent on the attitudes and perceptions of each individual” (Dalton, 1988, 177). In Canada, Nevitte (1996) observes a similar trend between 1981 and 1990. Similarly, McAllister (2002) observes that a significant proportion of ‘late deciders’ are not irrational, floating voters but ‘calculating’ voters who pay close attention to campaigns and who are thus susceptible to campaign influence. Greater electoral

1Contrary to this view, Soroka (2003) observes a convergence in Canada among the media when it comes to the importance accorded to issues.

58 volatility means that electoral strategies have greater opportunities to impact electoral outcomes beyond just activating socio-political cleavages (Dalton and Wattenberg, 2001). And these opportunities place pressure on political parties to adopt sectional strategies (Pomper, 1996; Schantz, 1996b).

3.2.2 A Recent Increase in Issue Voting?

Many theories suggest an increase in issue voting among citizens, but the empirical findings on the matter are inconclusive. Franklin, Mackie and Valen (1992) observe the decrease of the explanatory power of social cleavages and an increase in the importance of issues to the vote. Särlvik and Crewe (1983) show an increase in the importance of issue voting in Britain during the 1970s. Bartels (2006) observes a similar phenomenon in the United States in the 1980s, but also notes that the economy remains the most important short-term influence on voting behaviour. Bélangerand Nadeau (2009) point out that the influence of issues on the vote has also recently risen in Quebec, despite the effect of economic voting remaining relatively constant over time (Bélangerand Nadeau, 2009, 41).1 A similar pattern has been observed in 23 other democracies (Budge and Farlie, 1983).

Yet not all scholars agree with the view that there has been a signiﬁcant increase in the impact of issues (see Lewis-Beck et al., 2008; Miller and Shanks, 1996). The hypothesis of an increase in issue voting is based on the theoretical assumption of an increase in political knowledge. But levels of political knowledge among American citizens seem to not have changed over the past 50 years (Delli Carpini and Keeter, 1996). Converse and

1“Mais contrairement àla conjecture economique, dont l’effet semble s’êtrefait sentir de manière relativement constante dans le temps, il semble que l’influence des autres enjeux sur le vote au Québec soit plus importante aujourd’hui que par le passé”(Bélangerand Nadeau, 2009, 41).

59 Markus (1979) test the alleged rising importance of issues in elections, but reach the same conclusion as those articulated in The American Voter (1960). However, using the same dataset (NES), Page and Jones (1979) reach different conclusions. They argue that the influence of policy preferences on voting decisions has been underestimated and the influence of party attachments, overestimated. The methodologies and measurements used to uncover the effect of issues on vote choice seem to have a great impact on conclusions. In this regard, Margolis (1977) argues that methodologies used in the 1960s overestimated the effect of issues and can be seen as a cause of the observed increase in issue voting; the effect of issues increased in the beginning of the 70s and receded in 1976, but most studies do not find the ‘issueless’ state of the 1950s.

The understanding of issue voting presented in The American Voter (1960), some analysts suggest, is too simplistic (Margolis, 1977). Issue voting is more subtle and complex than is typically presumed. Although citizens’ positions on issues are not powerful predictors of voting behaviour, they are not random. Recall that even Converse (2007) recently complained that his own conclusions about voter competence have been exaggerated by other scholars. Moreover, it is important to remember that the Michigan model was originally a critique of the Columbia model, which was considered too static. And, for proponents of the Michigan school, the key to the dynamics of voting lies in leader personalities and salient issues (Gidengil, 1992). These two dynamic factors—leader personalities and salient issues—are similar to the two components of the political product deﬁned in the political marketing literature. There seems to be a debate within the political marketing literature about what constitutes the political product. Some scholars consider the issue positions presented in electoral platforms to be central elements of the product (Baines and Egan, 2001), while others focus rather on images and leaders (Scammell, 1999). How- ever, most scholars agree on a deﬁnition of the political product that incorporates both

60 elements (Bartle and Griﬃths, 2002; Henneberg, 2002). These two elements also might not be entirely independent of one another. Jacobs and Shapiro (1994), for example, explain how positional issues have strategically been primed to shape voters’ perception of candidates’ personalities. Priming strategies are also used to boost the salience of issues in the whole population (Druckman, Jacobs and Ostermeier, 2004) or among targeted subgroups of the electorate (Hillygus and Shields, 2008).

Political marketing focuses on how political parties do, or at least can, improve their competitive advantage. The key seems to lie in parties having a more subtle understanding of their electoral market. The brokerage strategy of taking ambiguous positions on issues to avoid alienating segments of the electorate is not necessarily the most eﬃcient strategy. Alvarez (1998), for instance, shows that the more uncertain a voter is about a candidate’s positions, the less likely the voter is to support that candidate. Is it possible that the potential for issue voting has always been present but some opportunities, such as those that targeting technologies represent, were not? Issue voting did not necessarily emerge from a ‘Silent Revolution’ and its corresponding value shift (see Inglehart, 1971, 1977). It might have always existed at the micro-level, among smaller groups of voters concerned about a particular set of issues.

3.3 Discussion

This thesis argues that positional issues might matter more than the voter and party behaviour literatures assume. The key to understanding the historical lack of interest in positional issues is the citizen competence problem. The dearth of political sophistication among citizens seems to have encouraged analysts to focus on uncovering the fundamental

61 factors underpinning vote choice, instead of the effects of positional issues. These factors were mainly structural and left little room for campaign activities and other electoral strategies to matter. The structural perspective suggests that the fate of political parties is mostly determined by factors that parties cannot control. Even when positional issues are considered to matter, the strategic prescription for parties is often to converge toward median positions. In Canada, the concept of brokerage politics has been used to describe why parties traditionally avoid taking clear issue positions (Bickerton and Gagnon, 2004b; Carty, Young and Cross, 2000; Clarke et al., 1984). Few conjectures imply the need for a thorough study of positional issues. On the one hand, the voting behaviour literature suggests that positional issues have little effect. On the other hand, the party behaviour literature reveals few incentives for parties to take distinct issue stances. By contrast, the political marketing perspective provides a theoretical rationale for supposing that the effect of positional issues might have been underestimated.

The core presumption of political marketing is that the electorate is heterogeneous and that there is no “single model of man” (Bartle and Griffiths, 2002). The implication that follows is that parties should base their political strategies on thorough market research. Such a conception of the electorate supposes that different voters care about, and are affected by, different issues. It is important to note the difference between the public salience of an issue and its salience at the personal level. Although the salience of public issues in certain elections has been acknowledged, the effect of issue salience among issue publics has generally been overlooked. But some criticise political marketing on the grounds that it relies too heavily on dependent variable selection and post-hoc rational- izations (Henneberg, 2002). The basic assumptions underlying political marketing, then, need to be investigated. Also, the extent to which positional issues constitute central elements of the exchange between voters and parties must be examined. Ultimately, if

62 some positional issues can matter to some voters who care about them, parties might have more room to position, grow and shape their images.

In the Michigan School tradition, the effect on vote choice of short-term factors, such as issues, is considered small compared to the effect of long-term factors. The present thesis does not contradict this notion. When it comes to predicting electoral outcomes, positional issues cannot compete with structural factors such as the state of the economy. However, some importance must still be allotted to positional issues. Positional issues are important elements in parties’ strategic toolboxes. As Holbrook (1996) observes, a factor does not need to be the most important determinant of electoral outcomes in order to still play an important role in shaping public opinion. Indeed, the effects of issues that the authors of The American Voter Revisited (2008) observe are not small1, even when the heterogeneity of the electorate is not taken into account. The relative absence of positional issues in traditional voting and party behaviour theories and their assumed presence in political marketing theory can in part be explained by differing research foci. While traditional theories have sought to uncover the essential foundations of the vote, political marketing has sought to examine any issue-based leverage that might remain for parties.

Positional issues are difficult to study in a comparative perspective because they can be unique to contexts and times. For example, Abramowitz’s (1995) study of the effect of the abortion issue during the 1992 American presidential election could not easily have been examined in a cross-national context. Stewart (2002) identifies this situation as the

1On each one of the issues included in their survey, at least around 40% of voters meet the three conditions identiﬁed by Lewis-Beck et al. (2008) to be considered a potential issue voter: 1) Familiarity; 2) Opinionation; and 3) Diﬀerentiated perceptions of candidates. The results range from 37.8% for “Government assistance to blacks” to 62.1% for “Diplomacy versus military action” (Lewis-Beck et al., 2008, 181).

63 nuance paradox: the more subtle the understanding of a particular political context, the harder it is to compare to other political contexts. Generalizable theoretical concepts can nonetheless be extracted from context-specific findings. Methodological and measurement challenges are inherent to the study of positional issues. The next chapter is dedicated to such methodological questions. The following chapters test the underlying assumptions of political marketing. First, the concept of the heterogeneity of effects is explored. Different forms of heterogeneity are discussed and the issue-public hypothesis is revisited. A final chapter focuses on issue-based party strategies and looks more specifically at the dynamics of sociodemographic segmentation.

64 Chapter 4

A New Look at an Old Concept: Conceptualization, Measurement and Eﬀects of Positional Issues

This chapter discusses the main conceptual and measurement issues that have arisen in studies of political issues. Following Ansolabehere, Rodden and Snyder Jr. (2008), issue scales are used to test the stability and relative strength of positional issues compared to rival concepts, such as values and party identification. Political issues have been defined and measured in a large variety of ways across time and subfields. The complexity of the definitional exercise goes well beyond the difference between brokerage-style valence issues and positional issues. However, once these ontological and methodological challenges are overcome, the effects and stability of issues seem to be significantly greater than it is conventionally assumed.

65 4.1 The Conceptual Status of Positional Issues: On-

tological and Causal Confusion

There is no consensus in the literature about how to conceptualize political issues. The confusion partly arises because there are so many different angles from which issues can be studied. For example, the way one thinks about an issue will vary greatly depending whether one’s approach is micro-sociological, socio-psychological, or rational. The advent of survey methodology triggered a debate about the nature of citizens’ issue preferences. And positional issues are at the centre of the citizen competence problem. Different labels have been used to define the issue positions citizens report in surveys.1 These different conceptualizations entail different operationalizations and model constructions. It is thus important to determine the ontological contours of the different definitions of political issues and the causal structure they involve before delving into empirical analysis of issue effects. This section first discusses the links between issues and rival concepts such as party identification and values. Then, the different ways that issues have been conceptualized in the literature are analyzed before the measurement of issues is tested.

4.1.1 The Quest for the Essence of Vote: Rival concepts

Voters’ issue preferences are often described in the voting literature as manifestations of more deeply rooted political inclinations such as party identiﬁcation, ideologies, or values. If these factors are the cause of voters’ opinions on issues, then what causes these factors? Scholars inﬂuenced by the Columbia and Michigan schools see group membership as the

1To name a few: preferences (Downs, 1957), random non-attitudes (Converse, 1964), considerations (Zaller, 1992), latent dispositions (Krosnick, 1991), and psychological tendency (Bartels, 2002).

66 main root of issue positions (Campbell et al., 1960; Lazarsfeld, Berelson and Gaudet, 1948). But some argue that greater structural factors are at play. Historical circumstances, like the industrial revolution for instance, can cause sociodemographic changes that provoke value shifts, which impact voters’ issue positions (Inglehart, 1977; Lipset and Rokkan, 1967). A more recent line of research explores the possibility that another structural source—genes—might inﬂuence issue attitudes (Dawes and Fowler, 2009). When these historical, sociological and biological factors are taken into account, very little room is left for individuals to independently determine their issue preferences. One implication of such a line of speculation is that political parties cannot inﬂuence election outcomes beyond mobilizing their existing electoral base. Nevertheless, when one is more interested in examining campaign strategy than in uncovering the underlying essence of the vote, positional issues are crucial. Indeed, they are one of the few factors over which parties have any control.

No one disputes the fact that political parties take issue positions to attract certain voters. But we cannot rule out the idea that issue-based strategies might be successful just because citizens’ preferences are influenced by party identification and values. In fact, most issue-based strategies operate through party identification and values. Parties have been known to use issues to attract cross-pressured partisans (Hillygus and Shields, 2008) and to increase the salience of certain values that are advantageous to them (Abramowitz, 1995). Other scholars have shown how parties have primed positional issues in order to highlight their candidates’ positive character traits (Jacobs and Shapiro, 1994). Such party strategies do not fit the logic of conventional voting models. The relationship between parties and voters entailed by these strategies is the inverse of the relationship between parties and voters entailed by conventional voting models: positional issues are used to tap into values shared by certain sociodemographic groups in order to alter or

67 bolster their party identiﬁcation.

4.1.1.1 Party Identiﬁcation: Causal Confusion

The literature about the impact of party preferences on political behaviour has been described as “vast and confused” (Johnston, 2006, 336). The concept of party identification, which lies at the core of the Michigan model, has been endlessly scrutinized but a consensus has yet to emerge. Most of the disagreement emerges from the difficulty of determining the causal position of partisanship in models of political behaviour. From a rational standpoint, it appears reasonable to assume that issues can affect voters’ attitudes toward parties and that, in turn, these attitudes can affect vote choice (Fiorina, 1981). However, the Michigan school’s classic Funnel of Causality implies a reverse causal order: party identification has an indirect effect on candidate evaluations and issue orientations and a direct effect on vote choice (Campbell et al., 1960; Miller and Shanks, 1996). This theoretical model is consistent with the observation that voters tend to project their own issue positions onto candidates (Berelson, Lazarsfeld and McPhee, 1954). Ultimately, party identification has frequently been considered the key to solving the citizen competence problem. Moreover, while people have often been shown to have unstable attitudes on issues, they have often been shown to have stable party affiliations, and in several different political contexts (Butler and Stokes, 1969).

The greatest contribution of the Michigan school’s socio-psychological model was the establishment of party identification as a concept indispensable to voting studies. But confusion persists in the literature about the relationship between party identification and issue positions, and the respective relationships of these two concepts to the vote. Johnston (2006) distinguishes two ways of conceptualizing party identification: as an

68 unmoved mover and as a sum of preferences. Many scholars disagree with the idea, as suggested in The American Voter (1960), that party identification is an unmoved mover influencing attitudes on issues and candidates. These dissenters argue that party identification is better understood as a sum of preferences or a ‘running tally’ (Fiorina, 1981). The sum of preferences view argues that voters are rational actors who update their party affiliations when they assimilate new information (Gerber and Green, 1998). It has been shown that party identification is indeed updated continuously as new information about economic performance and presidential approval is received (MacKuen, Erikson and Stimson, 1989). In a similar vein, Jackson (1975) argues that positional issues cause party identification. Others observe that party identification can be influenced by short-term factors more broadly (Page and Jones, 1979). The two latter views give greater credence to the role of positional issues in voting models. These views imply that political parties have some leverage to increase their partisan base in the long or medium term.

Many scholars argue for a more dynamic conception of party identification (Clarke and McCutcheon, 2009; Franklin and Jackson, 1983; Niemi and Jennings, 1991), that high- lights the importance of other factors, such as leaders and valence issues, to vote choice (Clarke, Kornberg and Scotto, 2008; Clarke et al., 2010; Green, 2007). But a dynamic conception of party identification is difficult to demonstrate empirically. The stability of party positions on issues makes it difficult to separate conceptually the effects of party identification from the effects of positional issues. Stable party positioning may mask stable party attachments. A fiscally conservative voter, for example, might identify with a party because that party constantly takes fiscally conservative positions and other parties do not. Even when parties do change their positions, the effect of the changes is lagged and it takes some time for citizens to notice. Carmines and Stimson (1989) use aggregate and time-series data to show the long evolution of the effect of the American parties’

69 repositionings on racial issues in the 1960s. When parties change their issue positions, voters do not realign en masse in one election. Instead, the effect of issue repositioning on party identification is gradual, and thus difficult to gauge.

Uncertainty about the causal position of party identification in voting models is consequential. If issue concerns affect voter attitudes toward political parties instead of attitudes toward parties affecting issue concerns, then controlling for partisan attitudes in voting models would be at best, redundant and at worst completely wrong (King, 1991). Some analysts regard themselves as “agnostic about the causal relations among the variables” in voting models (Ansolabehere, Rodden and Snyder Jr., 2008, 227). These analysts add party identification to voting models only to test the robustness of their results. The causal confusion surrounding party identification is further complicated by unobserved heterogeneity in individual partisan stability (Neundorf, Stegmueller and Scotto, 2011). Not all voters are moved the same way by party identification. The observed decline in people’s psychological ties to political parties, for instance, does not occur at the same rate in all segments of the electorate (Abramson, 1983; Dalton, 1999). Similarly, younger voters’ partisan identification seems to be more easily shifted by political events than older voters’ party identification (Franklin and Jackson, 1983). Also, some heterogeneity might arise from different causal orders (Bartle, 2005, 656). Soroka (2003) notes that different issues exhibit different patterns of causal influence. Therefore, party identification can intervene in different stages of voters’ decision-making processes (Rivers, 1988). Some scholars point out that heterogeneity might even pervade spatial-theory models: “It is possible that some voters evaluate candidates according to the proximity model, whereas others use a directional measure” (Merrill and Grofman, 1999, 79). Many scholars disagree with any theory that moderates the effect of partisanship (see Johnston, 2006). Miller (1991), for instance, still considers party identification to be the strongest predictor of

70 vote choice across all demographic groups. Other more recent studies have also shown that party identiﬁcation is stable and impervious to short-term inﬂuences, and that the typical voter is a ‘tethered partisan’ (Green, Palmquist and Schickler, 2002).

4.1.1.2 Values: Ontological Confusion

In the quest to uncover the determinants of vote choice, party identification can be considered rival to positional issues. Voters’ attitudes on issues might simply be superficial manifestations of partisan allegiances. But the independent effects of issues on vote choice are also challenged by two other concepts: ideologies and values. These psychological structures guide uninformed voters during the development of their political beliefs. The causal confusion between party identification and issue positions is thus compounded by an ontological confusion. The conceptual distinctions between ideologies, values, and issue attitudes are difficult to make.

Numerous academic attempts to solve the citizen competence problem have led to the establishment of abstract concepts such as ideologies and values in the literature. Ideolo- gies and values seem to have many advantages over positional issues. While randomness seems to underly individual issue preferences, a stable, constrained structure seems to underly ideologies and values (Jackson, 1983). Moreover, theories based on values and ideologies often take into consideration voters’ low levels of political sophistication. Con- cretely speaking, values and ideologies show greater stability in voting models and seem to have stronger eﬀects on vote choice than do positional issues (Blais et al., 2002; Feldman, 1988). Attempts have even been made to elaborate spatial theories wherein ideologies replace issues as the most important factors (Hinich and Munger, 1994). Because values and issues have a higher degree of abstraction than positional issues, they are easier to ex-

71 amine in a comparative perspective (Haller, 2002). Positional issues are often products of the peculiarities of political contexts and are thus difficult to compare (Heath, Fisher and Smith, 2006). From this perspective, values can be considered as moving up on the ladder of abstraction (see Sartori, 1970). Values have fewer concretely defined attributes than do issues and can thus be applied to a greater diversity of contexts (Davidov, Schmidt and Schwartz, 2008). Ultimately, it is easy to see why values became such popular starting points within studies of comparative politics (see, for instance, Abramson and Inglehart, 1995; Inglehart, 1990). In their level of abstraction, values might be considered as being mid-way between positional issues and ideologies (Gerring, 1997). Even Converse (1964) perceived fundamental values such as egalitarianism, collectivism, statism, individual- ism and minoritarianism as ‘crowning postures’ that helped structure political attitudes. Other scholars argue that values act as bridges between the sociodemographic and partisan bases of voting behaviour: “[values] provide one important avenue for bringing the sociological and social-psychological approaches back together” (Gidengil, 1992, 241). Fi- nally, Alvarez and Brehm (2002) show how even the effect of political information on issue preference is moderated by values.

The ontological confusion between ideologies, values, and issues is reflected in the confus- ing ways these three concepts have historically been operationalized. Values have often been measured with a subset of the positional issue survey items used to measure ideologies. Not much has changed since Rokeach (1968) argued some 45 years ago that academics “do not as yet have methods for assessing values in a manner that would be distinct from the assessment of attitudes” (Rokeach, 1968, 547). Values seem to be more theoretically convenient than positional issues and ideologies insofar as they circumvent the citizen competence problem. Values, acquired and internalized at a young age, require little political information and cognitive involvement in order to be effective influences

72 on vote choice (Kroh, 2009). As a result, values might be treated as sociodemographic characteristics or permanent personality traits. In that case, they would appear at an even earlier stage in the causal order explaining issue attitudes and, ultimately, vote choice.1 Hence, according to Inglehart (1990), the development of post-material values causes the increase in the salience of new issues such as the environment, women rights and immigration.

Values help clarify voters’ preferences by linking political preferences to broad social objectives (Popkin, 1991; Sniderman, Brody and Tetlock, 1991; van Deth and Scarborough, 1995). But if positional issues had stable and strong independent effects on vote choice, there would be no justification for ignoring them. In some circumstances, positional issues might be of even greater interest than rival concepts. The study of electoral strategies in particular would benefit from a greater focus on concepts like positional issues that are more concrete and more directly observable than ideologies and values. Positional issues are both easier to track and to measure. Moreover, the role of positional issues as tokens of exchange between parties and voters has the potential to nuance rejections of the classic theory of democracy. If issues were actually as consequential and as stable as values, it would be difficult to justify studying values over issues. There seems to be little reason to study an abstract concept when a relatively more concrete concept shows comparable stability and consistency. However, one might choose to do so if ones research question is better suited to abstraction. One would understand positional issues differently depending on whether one was trying to uncover the underlying structure of the vote or understand campaign strategies. Therefore, the choice to study values over positional issues can be considered to be dependent on the researcher’s objectives.

1Confusion exists in the causal order explaining vote choice as well. For example, some scholars hold that partisan considerations aﬀect values (see Goren, Federico and Kittilson, 2009).

73 The citizen competence problem has driven scholars to concentrate their efforts on trying to uncover the essence of the vote. Specifically, it motivated academics to try to answer the following question: if voters are not informed enough to live up to the expectations of classical democratic theory, then upon which factors do they base their voting decisions? An entire field of study was born in response to this particular question. Indeed, the sociological, socio-psychological, economic and, more recently, genetic theories of voting all originated from this question. Each of these theories try to uncover the fundamental structures of voting decisions. But what do the underlying assumptions of these theories mean for political parties and electoral strategies? This question has often been avoided by repeating the mantra: campaigns do not matter. Nowadays, few scholars deny that electoral strategies have at least some effect (Wlezien, 2005). Political parties do have some leverage over electoral results. This room for manoeuvre is heavily constrained by the more fundamental determinants of vote choice. Nevertheless, the modest effects of campaigns have been shown to often be decisive in determing electoral outcomes (Hol- brook, 1996). Positional issues are short-term factors that—if they have any independent effects—definitely deserve scholarly attention. The following section reviews how positional issues have been treated in previous investigations.

4.1.2 Messy Territory: Conceptualizing Positional Issues

The various ways that issues are conceptualized and operationalized in the literature reflects a lack of consensus about their effects on vote choice. The term ‘issue’ is used in the literature to refer to many different concepts and theories. For example, it is difficult to argue that the same ontological understanding of issues is shared by theories of issue ownership, issue evolution, and issue publics. Complicating the situation further are a

74 multitude of issue typologies, of which the positional-valence dichotomy is only one. Of course, different definitions and operationalizations of issues lead to different observed effects on vote choice. This section explores the different ways that political issues have been conceptualized and measured in the literature.

Issues are not as well defined as other political factors. Neither sociodemographics, nor party identification, nor values suffer from the same conceptualization problems as issues. The peculiarity of issues has led to frequent academic efforts to create typologies of issues that distinguish between issues that matter and those that do not. A brief scan of the literature reveals that different typologies capture different aspects of political issues. Figure 4.1 displays some of the typologies used to define issues. It is evident that different typologies focus on and thus capture different aspects of issues. The following paragraph describes the difficulties encountered by those seeking for a common definition of political issue.

Four aspects of political issues are assessed by the different issue typologies found in the literature. Some scholars focus on the substance or content of issues. For instance, some make the distinction between traditional issues and ‘new’ issues, such as immigration issues, women’s issues and environmental issues, all of which emerged as a result of post- industrialization (Hooghe, Marks and Wilson, 2002; Inglehart, 1990). Others argue that the symbolic dimensions of certain ‘cultural’ issues can have peculiar effects on voters. In the United States, cultural issues such as nationalism and patriotism, race, gender and religion are all considered to have contributed to the Republican party’s electoral success between 1968 and 1988 (Leege et al., 2002). Differing complexity in the substance of issues has also been categorized into typologies. In that vein, Carmines and Stimson (1980) distinguish between ‘easy’ and ‘hard’ issues, with easy issues being those that require

75 little cognitive engagement to grasp and hard issues being those that require comparatively more.1 Abramowitz (1995) considers abortion, for instance, as an ‘easy issue,’ one which impacted the results of the 1992 American presidential election. Baum (2002) defines another class of issues that has the capacity to attract the attention of the politically less interested: ‘soft issues.’ Substantive distinctions between issues are particularly relevant to the solution of the aforementioned citizen competence problem. These distinctions add another dimension to the idea that issues cannot matter to politically unaware voters. If issues are grasped by different kinds of voters, their effects on voters might be different and the mechanisms by which they operate might also be different. Moreover, not all issues trigger the same interest in voters. Some scholars evaluate the impact of issues based on their levels of salience among voters. Soroka (2003) distinguishes between ‘prominent’ and ‘sensational’ issues. Other authors focus on examining the effects on voters of ‘public issues’ of the day, such as the issue of the Free Trade Agreement during the 1988 Canadian federal election (Johnston et al., 1992), health care in 2000 (Nadeau, Bélangerand Pétry, 2010), and corruption and accountability in 2006 (Andrew, Maioni and Soroka, 2006). But a salient issue is not necessarily a public issue and can be different for different groups of voters or ‘issue publics’ (see Krosnick, 1990). Finally, other issue typologies deal with the strategic value of certain issues. These typologies discuss ‘wedge’ issues, for example, which have the potential to influence voters to defect from their parties (Hillygus and Shields, 2008). They also discuss ‘insurgent’ issues (Vavreck, 2009), which divert attention away from ‘unobstrusive’ issues such as the economy (Baum, 2002) and which are more susceptible to agenda-setting.

The issue typology at the centre of this thesis is the positional-valence issue typology.

1Further proof of the lack of a common deﬁnition of positional issues lies in the fact that some authors, like Luskin, McIver and Carmines (1989), consider structural factors such as the economy ‘hard’ issues.

76 Figure 4.1: The Various Foci of Issue Typologies

SUBSTANCE SALIENCE ● Sociotropic / Egotropic ● Issue publics / Public issues (Nannestad and Paldam, 1994) (Key, 1961, Krosnick, 1990) ● Prospective / Retrospective ● Prominent / Sensational / Governmental (Florina, 1981; Kuklinski and West, 1981) (Soroka, 2003) ● ‘New’ issues (Hooghe et al., 2002; Inglehart, 1990) ● Easy / Hard issues (Carmines and Stimson, 1980) ● Soft issues (Baum, 2002) POSITION ● Positional / Valence issues (Stokes, 1963) ● Consensual issues (Dunlap, 1989)

STRATEGY ● Wedge issues (Hillygus and Shields, 2008) ● Cultural issues (Leege et al., 2002) Unobtrusive (Baum, 2002) ● Insurgent issues (Vavreck, 2009)

This typology seems to encompass all other issue typologies. Valence issues are most often described as public issues, which by deﬁnition are salient for the whole population. Moreover, the ‘valence politics’ model describes the strategic value of valence issues in elections (Clarke et al., 2010; Clarke, Scotto and Kornberg, 2010). This model assumes that valence issues matter and that positional issues do not. Carmines and Stimson (1980) also note that easy issues are often framed as positional issues while hard issues are often framed as valence issues. However, these same authors also deﬁne easy issues as those that deal with ends rather than the means to ends, a characteristic often attributed to valence issues.

The lack of consensus surrounding the deﬁnition of positional issues has led to positional

77 issues being operationalized and measured in a diversity of ways. As such, it comes as no surprise that the findings regarding positional issue effects have been mixed. The persis- tent disagreement in the literature about the actual role of positional issues in electoral politics calls for further investigation of the topic. Different measures of positional issues are used in different voting models. Some scholars use the issues that survey respondents consider most important to them personally (see Bélangerand Nadeau, 2009) while others use respondents’ attitudes toward public issues of the day, measured with single questions (Blais et al., 2002). Ansolabehere, Rodden and Snyder Jr. (2008) prescribe the use of multi-item issue scales instead of single-issue measurements, and advocate the inclusion in these scales of more issues than just the ones people deem most salient. The analyses presented in this chapter put the last two prescriptions into practice. They show that the stability and effects of positional issues on vote choice differ substantially depending on how issue positions are measured and conceptualized.

4.2 Hypotheses

The literature on issue eﬀects leads to at least two research hypotheses amenable to empirical investigation. The ﬁrst hypothesis revisits certain assumptions about the stability of positional issues once conceptual and measurement matters are taken into account. The second hypothesis examines the possibility that positional issues might have a considerable impact on the vote:

Hypothesis 1 : Issue stability is substantively increased with better measurement

Hypothesis 2 : Positional issues have a considerable eﬀect on vote choice

78 4.3 Methodology and Data

The data used to explore the hypotheses tested in this chapter come from the 2004, 2008 and 2011 Canadian Election Study (CES).1 These three studies contain a large number of issue items that have consistent question wordings. These characteristics allow us to construct consistent and comparable issue scales. Moreover, the 2004 and 2008 CES waves also include a panel composed of respondents who answered the same issue questions in four-year intervals, which allows us to explore the stability of issue attitudes.

The analysis presented in this section uses issue scales composed of multiple measures. Some Canadian scholars have already called for an improvement in the quality of measurements used in Canadian voting behaviour research; of these, some have suggested the use of covariance structure analysis (Gidengil, 1992). Ansolabehere, Rodden and Snyder Jr. (2008) show that there are clear advantages to using issue scales composed of multiple measures instead of individual survey items. The scaling of survey items also eﬀectively manages the problem of measurement error (see Achen, 1975). Nevertheless, it is still possible for nonrandom measurement error to emerge from unidimensional scales (Green and Citrin, 1994). To account for that possibility, the reliability of scales needs to be assessed. The scales used in this analysis are built using the factor scores for the ﬁrst factor of principal factors factor analyses.2

1Many issue attitudes are surveyed in CES mail-back questionnaires. The 2006 CES is not used for this analysis because it does not include a mail-back questionnaire. 2Cronbach α’s and factor loadings are used to evaluate the internal validity and unidimensionality of the scales. All items’ factor loadings are higher, and most of the time, much higher, than the acceptance level of 0.3 (Kim and Mueller, 1978, 70). For Cronbach α scores, there is no sacred level of acceptability (Schmitt, 1996). Nevertheless, all of the α scores in the present analysis, except for those related to the environmental issue scale, are above 0.5. The relatively low Cronbach α scores can be explained by the sensitivity of the coeﬃcient to the low number of items included in some scales (Garson, 2010). Also, the factor corresponding to the unidimensional scale are all in accordance with the Kaiser’s rule (Eigenvalue > 1). The issue scales used in this analysis are standardized to facilitate interpretation and comparison of the results, but Ansolabehere, Rodden and Snyder Jr. (2008) ﬁnds identical results using both normalized

79 Pearson’s correlations and logistic regressions are then used to assess the stability and strength of positional issues. The issue scales are used to compare the effects of positional issues to rival predictors of vote choice. The proper combination of statistical controls is difficult to determine. There is a debate in the political methodology literature between proponents of fully specified models, measured by a high R2, and those favoring specific causal inferences by including only control variables that are theoretically prior to the focal independent variables and correlated with both the dependent and independent variables (King, 1991).1 The choice of proper controls when assessing the effects of positional issues is further complicated by ongoing debates about the exact causal orders of rival concepts such as party identification, values, and positional issues in vote choice models. It is unclear whether these rival concepts need to be included as controls in a statistical model testing for the effect of positional issues on vote choice. To deal with this uncertainty, different models including many different causal orders are tested. Such a procedure can be considered as a partial sensitivity test for the robustness of the results. The next section presents the substance of the issue scales used in the analysis and the results from the tests of the different hypotheses.

4.4 Results

Eight positional issue scales are used in this analysis. The choice of these issues can be justiﬁed in part by their relevance to the political context and in part by the limits imposed by the available questionnaires. The number of issues and items is obviously limited by the CES questionnaire, which was not designed with the precise idea of building issue and standardized scales. 1See Achen (2005) for a discussion of the problems associated with using too many control variables in regression analyses.

80 scales in mind. Consequently, the number of items per scale varies per issue and is not as high as the multi-item scales used in other studies (see Ansolabehere, Rodden and Snyder Jr., 2008). Despite these limitations, the issue scales in the following analysis are suﬃciently robust to test the stability and strength of positional issues. Table 4.1 shows the issues, the direction of their position, and the number of items used to build the diﬀerent scales.

Table 4.1: Eight Positional Issues

Issues Position Number of items Economy Free market 6 Environment Environmentalist 2 Foreign/US relations More involvement, closer ties 5 Law and order Tough on crime 3 Minority issues More acceptance 8 Moral issues Traditional 4 Social programs No cut 6 Women issues Feminist 7

Source: Canadian Election Study, 2004-2008. See Tables 2-6 in the Appendix for details.

Positional issues are conventionally included as single items in fully specified regression models. The issue with this is that concepts operationalized by single items are prone to measurement error. The results of the present analysis show that the stability and the strength of positional issues are greatly enhanced when measurement error is taken into account. When multi-item issue scales are used, the effects of positional issues are clearer than when single-item questions are used. These results are important; they provide some support for the conclusions presented by Ansolabehere, Rodden and Snyder Jr. (2008). They also suggest that results found in the United States apply to the Canadian setting as well. Positional issues appear more stable and their effect on vote choice is

81 signiﬁcant.

Figure 4.2: Stability of Issue Attitudes 0.8 ● ● ● ● ● ●

0.7 ● ● ● ● ●

● ● ●

0.6 ●

● ● ● ● ● ● 0.5 ● ● ● ● ● ● ● ● ● ● ● ● ● 0.4 ● ● Stability (Correlation coefficient) ● ● ●

● 0.3 0.2

1 2 3 4 5 6 7 8

Number of items

Source: Canadian Election Study (CES), 2004 and 2008 (Panel respondents). Method: Pearson’s correlations. Factor scaling (if number of items > 2).

The data reported in Figure 4.2 also contribute to the literature in a different way; they illustrate the stability of various issue attitudes between 2004 and 2008. In this case, stability is assessed by examining Pearson’s correlations.1 The higher the score on the vertical axis, the more stable the attitude. These results confirm that aggregating multiple items increases the stability of positional issues. Correlations have been calculated for all possible unique combinations of items for a given issue. For instance, the CES contains seven different measures of attitudes toward the economy, which can be, and were,

1For the justiﬁcation of this measure see Converse (1964), Achen (1975), and Ansolabehere, Rodden and Snyder Jr. (2008)

82 combined in 127 unique ways1 for which Pearson’s correlations are calculated. Figure 4.2 shows the correlations between the 756 unique combinations that can be constructed from the eight positional-issue scales. When it comes to the number of items in the scales, the adage “the more the better” seems to apply (Ansolabehere, Rodden and Snyder Jr., 2008, 223). Certainly, issue attitudes appear more stable than expected once measurement error is taken into account. But this stability does not necessarily mean that positional issues matter.2 And one way to ascertain whether positional issues matter is to examine the eﬀects of the same issues on vote choice.

The tests for the effects of issue scales are summarized in Figure 4.3. Logistic regression models are estimated to evaluate the statistical significance and the impact of issues on vote choice. Three models, one for each major political party, are estimated for each issue variable.3 One way to circumvent the methodological debates associated with model specification (see Achen, 2005; Schrodt, 2010) is to incorporate different sets of statistical controls. This procedure can be considered as a partial4 sensitivity test that assesses the robustness of the findings to the inclusion or exclusion of control variables. Figure 4.3 (a) reports the statistical significance of the individual issue items and of the issue scales. Figure 4.3 (b) shows the strength of the effect on vote choice for the same variables. Once again, the key finding is that scales clearly perform better than individual items when

1Broken down: 7 individual items + 21 possible two-item scales + 35 possible three-item scales + 35 possible four-item scales + 21 possible ﬁve-item scales + 7 possible six-item scales + 1 possible seven-item scale = 127 possible unique combinations. 2Actually, the stability of issue attitudes is not necessarily an indicator of ‘real attitudes’ as understood by Converse (1964). People might change their minds for a number of reasons: “In the case of the ‘stability theorem’ in mass belief studies, the empirical status of the theorem is clear only if we assume the existence of a rigid reference coordinate that holds the object of appraisal unchanged either absolutely or in relation to the appraiser during the interval between measurements. If we allow for change in the object of appraisal, then the stability theorem can no longer claim an absolute or unambiguous signiﬁcance for the concept of stability” (Bennett, 1977, 478). 3For the results broken down by issues, see Figure 1 in Appendix. 4Contrary to a sensitivity analysis, control variables were not added individually. The controls were grouped and ordered according to the Michigan school’s Funnel of Causality.

83 Figure 4.3: Eﬀect of Issue Items and Scales

(a) (b)

Significance of issues, 2004 and 2008 Strength of issues, 2004 and 2008

● 10 1.0 ● ● ● ● ● 8 0.8 6 0.6 4 0.4

● ● Statistical significance (p−value)

Effect on vote (|Beta coefficients|) on vote Effect ● 2 0.2 0 0.0

(Items) Scales (Items) Scales

Source: Canadian Election Study (CES), 2004 and 2008 (Pooled data). Method: Logistic regression (summary of 2808 models). Factor scaling. Dependent variables: Vote(Conservative, Liberal, NDP). Independent variables: (Items) (Issue items [included individually]); Scales (Issue scales [included individually]) Controls: 1) SES only (468 models); 2) (...) + Values (468 models); 3) (...) + Party identification (468 models); 4) (...) + Economic perception (468 models); 5) (...) + Leader evaluation (468 models); 6) (...) + Incumbent evaluation (468 models). Note: The three vertical lines in graphic (a) represent the statistical significance levels: from bottom to top, p < .001; p < .05; p < .1. assessing the effects of positional issues on vote choice. And positional issues do seem to have significant effects on vote choice. These results are consistent with those found by Ansolabehere, Rodden and Snyder Jr. (2008) in the United States.1

How do these eﬀects compare with other predictors of vote choice? Figure 4.4 reproduces the results for the issue scales and compares them to the statistical signiﬁcance and

1Ansolabehere et al. (2008) results are only based on two issue scales: Economy (13 items) and Moral (6 items). The authors also tried “adding a Foreign Policy Scale but it was never statistically or substantively significant in 1992 or 1996” (Ansolabehere et al, 2008: 226). Contrary to these results, all eight positional issue scales are statistically or substantively significant in Canada in at least one of the elections under study. For a comparison of effect of scales on the explained variance, see Table 1 in Appendix.

84 strength of diﬀerent values and partisan identiﬁcations.1

Figure 4.4: Eﬀect of Issues, Values and Partisan Identiﬁcations

(a) (b)

Significance of issues, values and party identification, 2004 and 2008 Strength of issues, values and party identification, 2004 and 2008

● ● 1.0 ● 10 ● ● ● ● ● ● ● ● ● 8 0.8 ● ● ● ● ● ● ● ● ● 6 0.6

● ●

● ● 4 0.4 ● ●

● Statistical significance (p−value)

● (|Beta coefficients|) on vote Effect

● 2

0.2 ● ● 0 0.0

Scales Values Party Id Scales Values Party Id

Source: Canadian Election Study (CES), 2004 and 2008. Method: Binary logistic regression (summary of 2808 models). Factor scaling. Dependent variables: Vote(Conservative), Vote(Liberal), Vote(NDP). Independent variables: Scales (Issue scales [included individually]); Values (Value scales [included individually]); Party Id (Partisan identification [included individually]) Controls: 1) SES only (468 models); 2) (...) + Values (468 models); 3) (...) + Party identification (468 models); 4) (...) + Economic perception (468 models); 5) (...) + Leader evaluation (468 models); 6) (...) + Incumbent evaluation (468 models). Note: The three vertical lines in graphic (a) represent the statistical significance levels: from bottom to top, p < .001; p < .05; p < .1.

But is it possible that these results just reflect peculiarities of the 2004 and 2008 elections? And is it possible that there has been a recent increase in issue voting? To test these possibilities, the impacts of the eight issue scales are observed over time. The Figure 4.5 shows the evolution of the effect for each positional issue from 2004 to 2011. Unfortunately, the cross-time variability of the available CES issue questions makes it possible to only develop comparable issue scales for three elections. The findings presented do not appear

1For the results broken down by values and partisan identiﬁcations, see Figure 2 in Appendix.

85 to support the idea that there has been a general increase of issue effects. The impact of issue effects seems to vary by election and by party and positional issue.1 It is not surprising that the economic positional issue had a significantly more powerful impact in 2011 than in other years, given that the world was facing a major economic crisis at the time.2 Even so, the effect of economic positions is only greater in 2011 than in 2004 for attitudes toward the Conservative Party and the NDP. But the data do underscore a significant point; it seems that contextual factors matter. For instance, the effect of the environmental issue significantly increased from 2004 to 2008 for the Liberals, but not for the other parties. It is possible that the emphasis Liberal leader Stéphane Dion put on imposing a carbon tax contributed to this situation, especially given that the effect of the environment issue receded after the 2008 Federal Election. Overall, the directions of the issue effects do generally seem to reflect the policy positions adopted by the three parties. It is perhaps interesting to note, however, that every positional issue has its strongest effect on attitudes toward the Conservative Party. Whether such a situation is the result of the Conservatives’ global positioning or their electioneering style remains an open question.

1Testing the evolution of issue effects in the Canadian context is complicated by the fact that no positional issue question has been asked in every CES wave. The only issue-related question that has been consistently included in CES questionnaires since 1965 is the one asking voters to identify the issue most important to them. When we examine that question’s impact on explained variance in models of vote choice, we observe a small increase in the effect of the most important issue. And, again, there are differences by parties. However, because this question is an issue-salience measure that is not necessarily linked to a particular positional issue, the results must be interpreted with great care. For further information, see Figures 3 and 4 in the Appendix. 2The statistical significance of the results can be assessed in the full regression results displayed in Tables 7 to 14 in Appendix. The standard errors of the coefficients for the interactions between the electoral years and the positional issues are also represented graphically in Appendix in Figures 5 and 6.

86 Figure 4.5: Evolution of Issue Eﬀects for Each Party

Economy Environment

0.5 ● 0.5

● ● ● ● ● ● ● ● ● ● ● 0.0 ● ● 0.0 ● ● ● ●

−0.5 −0.5 Attitudes toward partyAttitudes toward partyAttitudes toward

2004 2008 2011 2004 2008 2011

Foreign/US relations Law and order

● 0.5 0.5 ● ●

● ● ● ● ● 0.0 ● ● 0.0 ● ● ● ● ● ●

−0.5 −0.5 Attitudes toward partyAttitudes toward partyAttitudes toward

2004 2008 2011 2004 2008 2011

Minority issues Moral issues

0.5 0.5 ● ● ● ● ● ● ● ● 0.0 ● 0.0 ● ● ● ● ● ● ● −0.5 −0.5 Attitudes toward partyAttitudes toward partyAttitudes toward

2004 2008 2011 2004 2008 2011

Social programs Women issues

0.5 0.5 ● ● ● ● ● ● ● ● ● ● ● ● 0.0 0.0 ● ● ● ● ● ● −0.5 −0.5 Attitudes toward partyAttitudes toward partyAttitudes toward

2004 2008 2011 2004 2008 2011

Source: Canadian Election Study (CES), 2004, 2008 and 2011. Method: Linear regression models are used in order to facilitate the direct comparisons between coeﬃcients from diﬀerent years. Note: Solid line = Conservatives; Dashed line = Liberals; Dotted line = New Democrats.

87 4.5 Discussion

The preceding analysis challenges the enduring view that positional issues do not matter much in Canadian elections. The results show that better conceptualizations and measurements of positional issues lead to the exposure of both their stability and their significant effect on vote choice. The significance and strength of issues is found to be comparable to even the significance and strength of more conventional predictors of vote choice like values and partisan identification. In fact, it turns out that issues are more often statistically significant than values. However, the conceptual confusion between values and positional issues might explain this surprising observation. One might argue that issue scales are actually measuring values. After all, moral and economic issue scales are often built using almost the same items as moral traditionalism and market liberalism value scales (see, for instance, Gidengil et al., 2012). There is some support in the literature for the proposition that these scales are measures of values (McClosky and Zaller, 1984). But there is also support for considering these scales as measures of issue positions (Ansolabehere, Rodden and Snyder Jr., 2008).1

It is possible that the analysis showing that positional issue attitudes are stable and have independent eﬀects on vote choice is simply the product of the use of multi-item scales. The observed strength of issues could be an artifact of applying the scaling technique typically used to measure values to positional issues. But even so, the idea that scaling techniques determine the strength of issues has little theoretical consequence for the causal order of conventional voting behaviour models. It is discussed above that single-item issue indicators are highly prone to measurement error and are thus likely to produce noisy

1The question thus remains: when should issues be considered values and when should values be considered issues? For a more detailed discussion about the ontological distinction between values and positional issues, see the corresponding section in the Appendix.

88 attitudinal signals. These signals become clearer with proper scaling. That noise does not mean that positional issue attitudes are less genuine. Many scholars after Converse (1964) defended the idea that citizens do have political attitudes despite their apparent lack of political sophistication and ideological consistency (Achen, 1975; Lane, 1962; Saris and Sniderman, 2004). Issue scales such as those used in this analysis, as well as those used by Ansolabehere, Rodden and Snyder Jr. (2008), seem to relax the requirements for political sophistication enough to render citizens able to vote on positional issues. The more items in a scale, the more opportunities to capture citizens’ real attitudes. The implication of that line of speculation is that scales might need to include more items in order to correctly measure positional issues. Including several questions about gay rights in a scale, for instance, might help to ensure that it is indeed attitudes about gay rights that are being measured and not attitudes about other, related moral issues, such as abortion. The same pattern should emerge if there are more items available in a scale to measure attitudes toward health care independently from attitudes toward other social programs.

There are many aspects of issue effects that require further investigation. But the principle question at hand is: do different issues affect all voters equally? Different ways of assessing the effects of issues on voters vary both in terms of measurement techniques and conceptualization. The positions and the personal saliences of issues constitute two different dimensions, which need to be measured separately (Miller and Peterson, 2004). These two aspects—position and salience—refer to two of the three elements that limit the potential effect of issues, according to the authors of The American Voter (1960).1

1The discussion of the effect of issues is complicated by the fact that 1) a small proportion of voters is considered as having positions on issues; 2) party identification is considered as heavily influencing these positions; and 3) voters are not concerned with the same issues (Campbell et al., 1960; Lewis-Beck et al., 2008).

89 The next chapter focuses on these two dimensions more directly as it examines the interaction between positional issues and personal salience. The heterogeneity of issue eﬀects is also considered.

90 Chapter 5

Public Issues or Issue Publics? The Distribution of Genuine Political Attitudes

Explaining vote choice necessarily involves examining behavioural patterns. But do all voters follow the same patterns? Most voting models assume that they do. For instance, the Funnel of Causality posits that the order in which various factors come into play in determining vote choice is the same for all citizens (Campbell et al., 1960). But some scholars argue that voting behaviour is more complex (Bartle, 2005). Different types of voters can be affected by different factors or behave according to different causal orders. Such a line of speculation, however, significantly complicates voting theories. In fact, the complexity of voters’ decision-making poses fewer problems to statistical modelling than the diversity of reasoning paths that might precede vote choice (Rivers, 1988). Heterogeneity in voters’ decision making processes has at least two causes. First, heterogeneity might originate

91 in differences in voters’ general ability to reason about politics. This ability to reason is assessed with measures such as political knowledge (Delli Carpini and Keeter, 1996; Roy, 2009), awareness (Converse, 1964; Zaller, 1992), and education (Sniderman, Brody and Tetlock, 1991). Second, voters may have more specific or specialized knowledge than general knowledge, and they might accord different levels of salience to different issues (Krosnick, 1990; RePass, 1971). These two types of heterogeneity are referred to here as elite-based heterogeneity and salience-based heterogeneity, respectively.

There are some similarities between the two types of heterogeneity.1 Both relate to information effects. Greater information about issues increases the weight that issues have in voters’ decision-making processes. Elite-based heterogeneity recognizes that political knowledge is concentrated among specific segments of the population and implies a rather dichotomous view of the electorate (see, for example, the Black-and-White model in Con- verse, 1970).2 Alternatively, the salience-based conception of heterogeneity implies a more specialized political knowledge. Many voters may only care about a limited set of issues and thus might act as issue publics (Krosnick, 1990). The concept of issue publics is particularly relevant here because it relaxes some of the limitations to positional-issue voting identified in the literature: those that relate to the general lack of political attitudes.3 Indeed, despite the fact that few issues concern the entire electorate enough to trigger

1In his typology of heterogeneity, Bartle (2005) actually considers these two types of heterogeneity as a single type. A third type could be cognition-based heterogeneity, which involves different ways of organizing information and understanding politics; this might be considered a more amorphous type of heterogeneity (see also Goldberg, 2011). 2To be fair, Converse (2007) recently expressed concern about reductionist interpretations of his theories and “disavow[ed] any reading of the Belief Systems essay that concludes that most citizens lack political attitudes.” It must be remembered that in the Belief Systems essay (1964), Converse does argue that some issues, such as race, are very important to structuring political attitudes in the United States. 3For example, in The American voter Revisited (2008), Lewis-Beck et al. identify three individual limitations that prevent issues from significantly mattering to vote choice: 1) the lack of issue intensity; 2) the limited familiarity with an issue; and 3) the difficulty in distinguishing between candidate’s positions on the issues.

92 widespread attitude formations, the issue-public hypothesis contends that even unaware voters can hold attitudes toward a limited number of issues about which they care. Issue publics can sustain stable and consistent attitudes toward a limited set of issues because of environmental and psychological reinforcement processes, such as selective exposure (Iyengar et al., 2008). The issue-public perspective also implies heterogeneous issue effects. The level of salience accorded to issues may vary by the voter and by the issue. The effect of environmentalist attitudes on voting behaviour, for example, might exhibit very little homogeneous effect. But, as an issue, the environment might be decisive for a small proportion of the electorate that cares specifically about that particular issue. This is especially the case when a party takes a clear position on a particular issue or already has a degree of ownership over the issue (Bélangerand Meguid, 2008; Petrocik, 1996). According to Delli Carpini and Keeter (1996), people usually tend to be generalists rather than specialists about politics. Moreover, some research seems to indicate that there is a fragmentation of political knowledge along lines that are different from the education stratification implied by the elite-based heterogeneity concept. Soft news in the United States, for example, has been shown to increase the information level of otherwise politically inattentive individuals on certain issues like foreign affairs and law and order (Baum, 2002).

This chapter explores the possibility that positional issue eﬀects operate in a heterogeneous manner. More speciﬁcally, the chapter investigates the presence or absence of elite-based and salience-based types of heterogeneity in the Canadian electoral context. Salience-based heterogeneity supposes an electorate fragmented into various issue publics, which can be consequential for electoral strategy research.

93 5.1 Heterogeneity and the Citizen Competence Prob-

lem

One of the core ﬁndings to repeatedly emerge from decades of public opinion research is that many voters lack political sophistication (Campbell et al., 1960; Converse, 1964; Delli Carpini and Keeter, 1996). But no scholar has ever claimed that all voters are politically ignorant. Campbell et al. (1960) found that there is a great divide between average citizens and political elites in terms of their engagement in politics, levels of political information, and ideological connectedness. Compared to elite belief systems, mass public belief systems tend to be more disorganized and more dependent on charismatic leaders than on political substance (Converse, 1964). According to Converse (1970), only around 20% of voters do not answer political questions at random. This dichotomous view of the electorate is known as the Black-and-White model. Neuman (1986) is even more pessimistic than Converse. In his view, roughly 75% of voters are politically uninterested and unsophisticated. Another 20% of voters are completely apolitical; they exhibit only minimal awareness of and concern for politics. Only 5% of the public constitutes what Neuman calls opinion-elite citizens. Yet, this small number of competent citizens is not necessarily alarming: “It is in the nature of representative democracy that only a small proportion of the population can participate in politics to the fullest” (Luskin, 1990, 331). Other scholars are more optimistic. Delli Carpini and Keeter (1996) conclude their study on political information by arguing that, despite the unequal distribution of knowledge across citizens, “more than a small fraction of the public is reasonably well informed about politics–informed enough to meet high standards of good citizenship.” (269) The debate about citizen competence is evidently closely related to the debate about the distribution of political sophistication in the population. Voters need a minimum level of political

94 sophistication to be considered as having genuine attitudes on positional issues.

5.1.1 A Black-And-White Electorate: Elite-Based Heterogene-

ity

Many scholars describe the electorate in a way that leaves little room for issue voting to be important. The Michigan school tradition of research is a well-known example. The authors of The American Voter Revisited (2008) consider the lack of issue familiarity and intensity as individual limitations that explain the absence of issue voting. People do not recognize political issues and can not express degrees of preference for one issue position over another. This situation is a product of the fact that there are often too many issues discussed in a campaign, which is considered an external limitation. More-educated voters can overcome these hurdles, but the pool of potential issue voters is too “small” to allow issues to matter for vote choice (Lewis-Beck et al., 2008, 182). According to one widely shared view, education is the key to positional issues mattering (see Kuklinski and Peyton, 2007). A rise in levels of education should be accompanied by a corresponding increase in the importance of positional issues. Some scholars argue that such an increase in the eﬀect of issues on vote choice has indeed occured (Budge and Farlie, 1983; Franklin, Mackie and Valen, 1992). But others show that higher levels of formal education do not automatically lead to political interest and political sophistication (Neuman, 1986). Education turns out to be an imperfect proxy for more relevant factors such as political awareness or political interest (Highton, 2009; Kwak, 1999). More politically aware individuals are those that are motivated enough to both gather information and put cognitive eﬀort into developing ideas about politics (Taber and Lodge, 2006; Zaller, 1990). Hence, political awareness appears as a predictor of the amount of information gained during an electoral campaign

95 (Nadeau et al., 2008).1 Similarly, there is persuasive evidence showing that the knowledge gap widens when a public issue becomes salient (Althaus, 2003). But does having new information automatically translate into having genuine attitudes? There are reasons to doubt that claim.

The media shapes perceptions and influences political attitudes (Mutz, 1998). In Canada, media use is correlated with more negative perceptions of the state of health care (Bli- dook, 2008). But information effects could simply encourage greater conformity with the opinion-elite. Zaller (1992) defines these elites as government leaders, journalists, activists, and policy experts, and Neuman (1986) characterizes them as well-educated males, mature in age, who occupy high status and high income occupations. But how do elites’ views diffuse? Educated voters tend to be more influenced by media content than voters who use a combination of popular wisdom and experiential knowledge to frame political issues (Gamson, 1996). In any case, the elitist view is very present in the literature. Recall, for instance, Key’s (1961) claim that the “voice of the people is but an echo.” Schattschneider (1960) similarly views political elites as defining the scope of conflict and controlling how ordinary voters think about issues. The elitist view sometimes borders on conspiracy theory: “the marketplace of ideas is dominated by the views of elite strata. The more exposed people are to the market, the more likely they are to see the world through the eyes of the upper class” (Ginsberg, 1986, 148). Some even suggest that elites are the guardians of such democratic values as political tolerance and minority rights (Peffley and Rohrschneider, 2007). The claim that elites structure the political debate is not particularly controversial. When elite consensus on an issue exists, sophisticated citi-

1The eﬀect of electoral campaigns on the public level of information is now widely accepted (Holbrook, 1996; Popkin, 1991). However, as stated by Nadeau et al. (2008), it might be important to distinguish between two types of political information: citizens’ general stock of information (GSI) and the campaign- speciﬁc information (CSI) that they might gain.

96 zens tend to conform to elite views (McClosky and Zaller, 1984). Conversely, when elites disagree about an issue, polarization at the public level develops (Entman, 2004). More politically aware individuals tend also to become more ideological about salient public issues than politically unaware individuals (Zaller, 1992).

Diﬀerent theories assume a dichotomy in the electorate in terms of political sophistication. Whether this cleavage is caused by varying education, information, or awareness levels matters less than the idea that only a small proportion of voters have the ability to vote according to positional issues. According to this view, positional issues matter, but only for the most politically enlightened in society. However, another line of research explores a deeper level of heterogeneity, salience-based heterogeneity, which presumes a more fragmented electorate.

5.1.2 A Fragmented Electorate: Salience-Based Heterogeneity

Positional issues can matter for political experts because these experts can access their policy attitudes from memory (Goren, 1997). When a public issue becomes salient, more people take it into consideration when making political decisions. This is the logic behind the priming effect (see Scheufele, 2000). As the authors of The American Voter (1960) ac- knowledge, highly salient public issues contribute to the dynamism of political behaviour. In order for a public issue to become salient, it needs to be extensively diffused through the population: “[...] the more attention the media pay to an issue, the more likely that issue is to come to mind and thus influence people’s political judgments” (Gidengil, 1992, 89). However, as Lewis-Beck and colleagues (2008) note: “[the] widespread availability of information does not, in itself, guarantee that issues will become part of a voter’s con- scious reasons for supporting one candidate over another.”(166) Is it likely, then, that

97 different voters care about different issues? If different voters cared about different issues, each would presumably inform herself about the particular issue she cares about. After all, the acquisition of political information is considered to be the result of complex interactions between individual factors and environmental supply (Huckfeldt and Sprague, 1995). Yet, people are often characterized as being generalists rather than specialists in their political knowledge (Delli Carpini and Keeter, 1993). This characterization means that familiarity with one issue should translate into an increase in the probability of familiarity with another (Lewis-Beck et al., 2008). That may be true but many scholars also argue that issue salience operates at the individual level.

Issues need to be considered important to voters in order for them to be able to influence vote choice (Rabinowitz, Prothro and Jacoby, 1982). Some argue that there are different types of issue saliences. McLeod, Becker and Byrnes (1974), for instance, distinguish between community issue salience, individual issue salience, and perceived issue salience. The first two types of salience are particularly useful as they differentiate public issues from issues that are personally important to an individual voter.1 If citizens’ vote choices were influenced by their attitudes on issues important to them, then individual issue salience would represent fertile ground for research into the extent to which, and how, positional issues might matter to voting behaviour.

Determining why voters care about diﬀerent issues is not a simple endeavour. Even the authors of The American Voter Revisited (2008) were puzzled by the individual variance among voters in familiarity with certain issues. For them, “the existence of cognition is the critical factor” (Lewis-Beck et al., 2008, 171). But the question remains: what could

1The third type–the perceived issue salience–connects the first two types by referring to the individual perception of the community issue salience. Perceived issue salience can be associated with agenda-setting effects (see McCombs and Shaw, 1972) and to the concept of impersonal influence (see Mutz, 1998).

98 account for the difference in cognition? Different analysts supply different answers to that question. First, voters are not all exposed to the same sources of information. In postindustrial societies, the workplace and the media environment are becoming more complex and more fragmented. The mass media is less uniform than it once was, when presidents were sold like “soap powder or a can of beer” (McGinnis, 1968). The multiplication of television channels and the advent of the internet has contributed to an environment that favours the expression of sub-interests. Baum (2002), for example, shows how viewers of soft news become inclined to care about some issues more than others. The fragmentation of the media environment also facilitates selective exposure, which implies a reversed order: viewers choose their source of influence. Indeed, citizens now have more opportunity to avoid political information that does not conform to their worldview. Stroud (2008b) demonstrated how voters’ political beliefs are increasingly related to voters’ media exposure. Selective exposure to media sources is consistent with Festinger’s (1957) cognitive dissonance theory, which argues that people tend to reject information that is not consistent with their belief system. Moreover, there is evidence that voters do not react in the same ways to the same sources of influence (Erbring, Goldenberg and Miller, 1980). For example, elderly citizens are more affected by news reports about social security than are younger citizens, and African Americans are more affected by news about racial discrimination than other citizens (Iyengar and Kinder, 1987). The structural changes associated with the rise of post-industrial societies, combined with psychological concepts such as cognitive dissonance and selective exposure, might jointly contribute to the creation of issue-specific experts.

An issue public is comprised of voters who are more alert, attentive, interested and informed about a given issue (Key, 1961). Ironically, it was Converse (1964) who ﬁrst introduced the concept of issue publics. According to Converse, “[...] we come a step

99 closer to reality when we recognize the fragmentation of the mass public into a plethora of narrower issue publics.”(245) But it was Krosnick (1990) who primarily elaborated the contemporary version of the concept to create the issue-public theory. Salience-based heterogeneity of the eﬀect of positional issues on vote choice is central to issue-public theory:

Policy attitudes that citizens consider important are highly accessible in memory, are highly resistant to change, are highly stable over time, are extensively linked to and consistent with individuals’ basic values, instigate polarized perceptions of competing presidential candidates’ policy attitudes, and are powerful determinants of candidate preferences. (1990, 70)

The issue-public idea complicates elitist views of the electorate. However, the idea of different causal mechanisms operating for different subgroups does not have the same appeal as theories that are more universally applicable and parsimonious. Krosnick’s issue-public theory is important because it relaxes the conditions under which issue voting can matter. All voters do not need to know, as is implicitly assumed by spatial models, the positions of all the parties on all the issues (Lewis-Beck et al., 2008). Voters can handle being knowledgeable about a small set of issues that are important to them: “[...] people probably accumulate large stores of knowledge in memory relevant to important policy attitudes and smaller amounts of knowledge relevant to unimportant policy attitudes” (Krosnick, 1990, 68). The stability of attitudes on a single issue that matters to a voter is easier to manage, even without ideological consistency (Taylor, 1983). In this way, public issues overcome the information-related hurdles facing issue voting. The authors of The American Voter Revisited (2008) detect some variations in the electorate concerning the perception of candidates’ issue positions, and these differences are not fully explained

100 by education. One possibility is that different levels of party identification account for this variance in perception. But it is also possible that the relationship operates in the opposite direction. According to Krosnick (1990), sophisticated perceptions on issues follow personal salience, not the other way around. Consider the case of someone who cares a lot about a particular issue and who is more informed about party positions on that issue. That person would be more likely to support the party that is closer to his own position on that issue. While party positioning on issues is rather stable, it is difficult to detect issue-public voting from crystallized party identification.

The issue-public theory is not necessarily inconsistent with the Michigan school model. Indeed, some Michigan school scholars seem to recognize the role of salience-based heterogeneity in vote decision: “[...] even though familiarity and opinionation are widespread, citizens are much more narrowly focused with respect to the policy questions that arouse strong feelings” (Lewis-Beck et al., 2008, 175). But not all voters are necessarily members of an issue public. Moreover, other factors such as values, party identification, or even elite discourse, can contribute to voters developing attitudes about issues that are less important to them as well. In Krosnick’s view: “It seems more sensible [...] to view linkage between an attitude and values, needs, and goals as one possible cause of importance” (Krosnick, 1990, 60). From this vantage point, issue publics can hardly be considered campaign-specific, short-term factors. The origins of issue publics might be motivated by self-interest and might grow as the relevant issues become more prominent (Key, 1961). In short, the issue-publics literature is ambiguous about where exactly in the causal chain leading to vote choice issue-public membership should be placed. But by considering salience-based heterogeneity, this literature implicitly assumes that attitudes on positional issues precede party identification.

101 5.2 Hypotheses

The diﬀerent theories concerning the distribution of political attitudes provide the foundation for a number of testable hypotheses:

Hypothesis 1 : Members of issue publics should have more stable attitudes and more accessible positional issue attitudes than other voters

Hypothesis 2 : The impact of positional issues on vote choice should be stronger among members of issue publics

5.3 Methodology and Data

The 2004 and 2008 Canadian Election Study (CES) surveys are useful platforms for testing these hypotheses.1 The CES questionnaires include various measures of political sophistication such as education, political information, and political interest. Salience is also measured in diﬀerent ways, which means that it is possible to test for both the subjective and objective types of salience detected by Epstein and Segal (2000). The various questions asking respondents how much importance they accord to issues provide a subjective measure of salience. Furthermore, the CES surveys also include stop watches that measure the time it takes a respondent to answer positional issue questions. These measures can be considered measures of objective salience (Miller and Peterson, 2004). Open-ended questions, like the ones that probe which issue respondents’ consider most important, prevent an over-standardization of the responses (Kuklinski and Pey- ton, 2007). Such a consideration is particularly relevant for exploring the heterogeneity

1Many issue attitudes are surveyed in CES mail-back questionnaires. The 2006 CES is not used for this analysis because it does not include a mail-back questionnaire.

102 of the electorate. Finally, these datasets permit us to use the issue scales presented in the previous chapter to assess potential heterogeneity of positional issue eﬀects in the electorate.

The concept of heterogeneity is not intuitive and it can represent operationalizational challenges. One way to detect heterogeneity is to consider interaction terms that capture non-additive eﬀects (Bartle, 2005, 656). Goren (1997) used similar interval-level knowledge scales and similar formal interaction tests to demonstrate the eﬀects of expertise. In this analysis, positional issues are interacted with the two types of heterogeneity bases. For elite-based heterogeneity, interactions between positions on issues and various measures of political sophistication reveal whether issues matter more for politically sophisticated voters than for unsophisticated voters. More precisely, these measures of political sophistication include political interest scales, dummy variables for education levels, and a political information scale: three indicators that tap voters’ motivation, ability and opportunity to develop general political sophistication (see Delli Carpini and Keeter, 1996; Luskin, 1990).1

When assessing salience-based heterogeneity, interactions between issue positions and individual salience probe whether specific interest in an issue is associated with heterogeneous effects of issues on voting behaviour. Objective individual salience is measured using a scale that includes various items tapping the importance respondents accord to a particular issue. The use of such a scale is motivated by the difficulty inherent in capturing the concept of issue salience with a single item. The use of only the conventional “Most important problem” question to measure importance is problematic (see Wlezien, 2005).

1The choice of keeping multiple measures of political sophistication is justiﬁed by the desire to account for the various ways of conceptualizing and operationalizing political sophistication that exists in the literature. Among others: 1) Education level (Converse, 1964); 2) Interviewer assessment (Zaller, 1992); 3) Direct objective questions (Delli Carpini and Keeter, 1996).

103 It is unclear if this question taps the importance of an issue to the respondent personally or if it simply reflects the respondent’s perception of the most publicly discussed issue. Hence, it is important to employ a measure that distinguishes between the fact of being a member of an issue public, which involves having strong feelings toward a specific issue, and detached knowledge of the public issue of the day. This distinction is not always made in the literature. For instance, Lewis-Beck et al. (2008) consider political information and perception of objective reality on one issue as a measure of personal issue salience when, according to Krosnick (1990), political information should be a consequence of issue salience. Therefore, members of issue publics are defined by their level of attachment to a particular issue. The use of multi-item scales that include questions asked at different points in the survey and with various wordings provide further foundation for a more valid measure of individual issue salience.

5.4 Results

The data reveal evidence of the ﬁrst two types of heterogeneity. Salience-based heterogeneity has a greater impact on the eﬀect of positional issues than elite-based heterogeneity. The evidence thus provides some support for an electorate composed of issue publics.

Citizens who care about an issue tend to think more about it than those who do not care about the issue. As such, these voters are likely to have more easily accessible issue positions stored in their short term memory (Krosnick, 1990). The 2004 CES contains a measure of how long respondents take to answer questions on same-sex marriage. This timing indicator provides an objective measure of personal salience (Epstein and Segal,

104 2000). The speed with which a respondent answers the question on same-sex marriage in 2004 predicts the stability of the attitudes when the same question is asked in 2008. Indeed, those who care the most about the same-sex marriage issue are the least likely to have changed their minds over four years. Figure 5.1 illustrates the effect of response time on the stability of attitudes once such political sophistication indicators such as education, political knowledge, and political interest are controlled. The results show a clear and significant effect of response time on the cross-time stability of attitudes on positonal issues. Attitudes about positional issues are more stable for citizens who care about a particular issue. But does personal salience also have an effect on attitudes toward parties?

The issue-public hypothesis also stipulates that the effect of issue attitudes on vote choice will differ depending on the level of salience voters accord to particular issues of interest. One way to test this proposition is to consider those positional issues clearly associated with a specific party: moral issues for the Conservative Party and the environmental issue for the Green Party.1 These choices aim to avoid the confusion that disputed issue ownership might have on issue voting (see, for instance, Damore, 2004; Holian, 2004). The conventional wisdom is that for an issue attitude to have an effect on the vote, there must be perceived party differences on that issue (Lewis-Beck et al., 2008).

Figure 5.2 suggests that views on moral issues of those who care a lot about these issues have a stronger impact on evaluations of the Conservative Party than do other voters’ views on moral issues.2 A similar pattern emerges with individuals’ positions on the

1Oddly enough, the association of the environmental issue with the Green Party might be less clear in the context of the 2008 Canadian election. In that election, the Liberal Party included the proposition of a ‘Green Shift’ in their electoral strategy. 2Factor scales measuring attitudes toward parties are used as dependent variables instead of vote choice because there are so few Green Party voters in the sample. To be consistent, a similar scale is used to evaluate the impact of attitudes toward moral issues on support for the Conservative Party. But

105 Figure 5.1: Accessibility and Stability of Issue Attitudes

Source: Canadian Election Study (CES), 2004 and 2008 (Panel). Method: Least Squares Regression (Full results in Table 15 in the Appendix). Vertical bars indicate 95-percent conﬁdence intervals. Dependent variables: Stability of attitudes toward same-sex marriage. Independent variables: Respondent’s time to answer the questions related to the same-sex marriage issue in 2004. Controls: Education, Political information and Political interest.

106 Figure 5.2: The Eﬀect of Issue Salience: Issue publics? Positive Positive mean mean Attitude toward the Green Party Attitude toward Attitude toward the Conservatives Attitude toward Negative Negative Progressive mean Traditional Less mean More Attitude toward Moral Issues Degree of Environmentalism

(a) (b)

Source: Canadian Election Study (CES), 2004 and 2008 (Pooled). Method: Least Squares Regression (Full results in Table 16 in the Appendix). Predicted values. Vertical bars indicate 95-percent conﬁdence intervals. Dependent variables: Attitude(Conservative), Attitude(Green). Factor scales. Independent variables: Issue scales (a) Moral Issues (Position and Salience); (b) Environment (Position and Salience). See Tables 2 to 6 in the Appendix for details. Controls: Age, Income, Education, Political information and Political interest. Interactions (Position X Salience): Moral issues (p < .01); Environment (p < .01). The darker lines represent the predicted level of attachment to a given party when the issue is salient for voters, and the lighter lines when the issue is not salient.

107 environmental issue and their evaluation of the Green Party. On the graphs presented in Figure 5.2, the darker lines that represent the voters that accord a high level of salience to the issue have a steeper slope than the lighter lines representing the other voters. There is a statistically significant interaction effect (p < .01). Heterogeneity appears to influence how issue attitudes can matter for electoral politics. The results in Figure 5.2, then, provide some support for the issue-public hypothesis. It is possible that these results might mask the effects of elite-based heterogeneity in political sophistication. Issue voting, after all, is more pronounced among those with higher levels of expertise (Goren, 1997). Is it possible that general expertise about politics matters more than specific expertise on an issue? The results reported in Figure 5.3 indicate that elite-based heterogeneity also matters. The significant finding that emerges from those data is that the personal salience accorded to an issue has a greater interactive effect than different measures of political sophistication.1

5.5 Discussion

The findings presented in this chapter suggest that positional issue effects vary as a function of the extent to which an individual cares about a particular issue. While different positional issues might have a negligible effect on voting behaviour at large, they seem to have substantial effects on the voting behaviour of the particular groups to whom they are salient. This heterogeneity of issue effects can thus be labeled ‘salience-based heterogeneity’. Characterizations of electoral dynamics such as the brokerage-politics note that when vote choice is used as a dependent variable, the substantive results are the same. 1There is a debate in the literature concerning how to measure political sophistication (see Luskin, 1987). We adopt the skeptical view that advanced education does not automatically lead to political interest and political information (Neuman, 1986). Consequently, the components of political sophistication are tested separately.

108 Figure 5.3: Elites or Issue Publics? Positive Positive Positive Positive mean mean mean mean Attitude toward the Conservatives Attitude toward the Conservatives Attitude toward the Conservatives Attitude toward the Conservatives Attitude toward Negative Negative Negative Negative Progressive mean Traditional Progressive mean Traditional Progressive mean Traditional Progressive mean Traditional Attitude toward Moral Issues Attitude toward Moral Issues Attitude toward Moral Issues Attitude toward Moral Issues (a) Salience (b) Education (c) Political interest (d) Political information Positive Positive Positive Positive mean mean mean mean Attitude toward the Green Party Attitude toward the Green Party Attitude toward the Green Party Attitude toward the Green Party Attitude toward Negative Negative Negative Negative Less mean More Less mean More Less mean More Less mean More Degree of Environmentalism Degree of Environmentalism Degree of Environmentalism Degree of Environmentalism (e) Salience (f) Education (g) Political interest (h) Political information

Source: Canadian Election Study (CES), 2004 and 2008 (Pooled). Method: Least Squares Regression. Predicted values. Vertical bars indicate 95-percent confidence intervals. Dependent variables: Attitude(Conservative), Attitude(Green). Factor scales. Independent variables: Issue scales (a) (b) (c) (d) Moral issues (Salience and Position); (e) (f) (g) (h) Environment (Salience and Position). See Tables 2 to 6 in the Appendix for details. Controls: Age, Income, Education, Political information and Political interest. Interacting variables: Salience (High salience); Education (University education); Political interest (High political interest); Political information (High political information). The darker lines represent the predicted level of attachment to a given party when the level of the interacting variable is high, and the lighter lines when its level is not high. model tend to overlook the heterogeneity of issue effects and the ability of political parties to target specific groups of citizens.

The data also demonstrate that there is an unequal distribution of issue attitudes among citizens. That is to say, following the issue-publics hypothesis, people diﬀer in the extent to which they have sophisticated issue attitudes. This attitudinal heterogeneity, however, does not simply reﬂect the classic elite-mass divide on political sophistication. The black-

109 and-white model of the electorate elaborated by Converse (1970) has been shown to be more complex. Citizens’ knowledge about diﬀerent political issues varies as a function of extent to which they care about those issues. Thus, political expertise can be fragmented and dispersed across various issue publics.

Despite the existence of a horizontal fragmentation of political sophistication, an elitist view of the electorate still holds. Variables measuring political sophistication are found to all consistently impact voters’ choices. Certainly, the effects of structural variables such as education must be considered when attempting to uncover the underlying composition of voting behaviour. However, it is reasonable to suppose that the fragmentation of the workplace and the media environment have corresponding effects that make the electorate less homogeneous. In a multichannel universe, voters have more opportunities to avoid topics that do not interest them and focus on topics that do interest them. Exposure to mass media narrows the knowledge gap between the elites and the general public (Kwak, 1999). And it is possible, as Baum (2002) argues, that soft news motivates politically uninvolved voters to become more interested in specific political issues such as foreign crises. Even Zaller (1998) acknowledges that his previous works might have exaggerated citizens’ lack of competence. After seeing Bill Clinton’s approval rating improve in the midst of the Monica Lewinsky scandal, Zaller concluded that citizens might actually be better at handling political substance than he had originally thought. Moreover, Kuklinski and Segura (1995) show that concerns over some issues such as abortion can emerge from the grassroots level. The voice of the people might thus be more than an echo. In any case, if positional issues matter for some voters, political parties have an incentive to care about them. And brokerage parties’ strategies of avoiding taking clear stances on issues might be suboptimal. The next two chapters explore the possibility that positional issues were used by political parties in recent Canadian elections to increase their support

110 among speciﬁc groups.

111 Chapter 6

Detecting Potential for Growth: Immigrants as Targetable Segments

Political parties increasingly develop campaign strategies that target small electoral segments (Hillygus, 2007). Some recent research has shown that such targeting strategies can effectively influence voter behaviour (Alvarez, Hopkins and Sinclair, 2010). But how do positional issues fit into this landscape? The previous chapters have suggested that positional issues matter to voters more than is conventionally assumed by the literature on voting behaviour and brokerage politics, especially when we consider the salience-based heterogeneity of the electorate. These findings provide some support for the effective use of political marketing strategies and tactics such as micro-targeting. If different issues matter to different voters, parties are well-advised to use market research to evaluate their potential for growth. Such market research would entail identifying profitable groups to

112 target. Like the Columbia and Michigan theories, political marketing theory is skeptical about the ability of political parties to change voters’ preferences. At the core of political marketing is the idea that political oﬀers can be adapted to voters’ needs and wants (O’Shaughnessy, 1990). Political parties are also subordinate to voters, not the other way around. Political marketing approaches grant parties more control over their electoral fates than do the Michigan and Columbia school voting models. Parties might not be able to change voters’ preferences, but they can change their political oﬀer to attract certain groups of voters. Positional issues might be used to attract groups of voters to the party, resulting in the party’s expansion of its electoral base. Inversely, positional issues might lead parties to lose support among certain electoral segments, as the Tea Party has done among more moderate Republicans. The conception of electoral dynamics in political marketing theory is in line with the idea that “social groups need to be treated as live social forces, not static categories” (Blais et al., 2002, 96).

An instance of a political party changing its political oﬀer in order to attract particular sub-segments of the population might be found in the positioning of the Conservative Party of Canada during the 2011 Canadian federal election. Shortly after the election, Flanagan (2011) suggested that the Conservative Party of Canada endeavoured to target immigrants speciﬁcally. Such a claim runs contrary to the conventional voting wisdom in Canada, which has historically considered immigrants to be very strong Liberal partisans (Bilodeau and Kanji, 2010). Indeed, immigrants have been considered to have contributed to several past electoral successes of the Liberal Party of Canada (Blais, 2005; Harrell, 2013). Nevertheless, Flanagan is not alone in believing the Conservatives successfully courted immigrants on the eve of the 2011 election; authors such as Friesen and Sher (2011) and Payton (2012) make similar claims. Given that immigrants are popularly thought to hold right-wing views on many issues (Bricker and Ibbitson, 2013; Flanagan,

113 2011), it would make sense for the Conservatives to endeavour to target these voters. Such targeting would allow Conservatives to move ahead of their principal historical opponents, the Liberals, while remaining close to their ideological core (Paréand Berger, 2008). But immigrants are not a homogeneous group of citizens. Though they all share the fact of being immigrants, they also come from different cultural backgrounds and belong to different racial groups. As such, different issues might be important to different immigrants.

This chapter examines diﬀerent explanations for the alleged recent shift of immigrant voters from the Liberal party to the Conservative party. Racial explanations appear more convincing than structural explanations that are based on the fact of having immigrated. More importantly, the results support the idea that positional issues and targeting strategies might have played a role in the success of the Conservatives in attracting some traditional Liberal voters. Ultimately, it seems as though the Conservatives successfully exploited an opportunity for growth by successfully attracting immigrant voters to the party. These voters appear to have been drawn to the party moreso because of their proximity to the party on issues than because of their race or a unique immigrantness trait.

6.1 Theory

6.1.1 Immigrantness and the Welcoming Eﬀect

The logic of the welcoming effect is based on the literature on incumbency effects. Put simply, uninformed citizens can rely on incumbents’ past performance in office as an

114 information shortcut to help them decide whether to support that party in future elections (Fiorina, 1981). Positive retrospective evaluations favor the party in power. It is easy to derive from this logic an application for the case of immigrant voters: perhaps immigrant voters, thinking retrospectively, support the party that was in power at the moment of their arrival in the country (Wiseman, 2007). For uninformed, non-partisan voters, incumbency can represent a useful cue for competence and stability. For immigrants, an incumbency effect might be more pronounced, as these voters are socialized into a country’s political environment later in their lives and have weaker partisan ties than do native-born voters (Black, Niemi and Powell, 1987; White et al., 2008). Indeed, in Canada, “immigration has added a higher proportion of voters with no established party ties to the electorate than in any other established democracy” (Carty, 2002, 727). If a welcoming effect exists, then Liberal popularity among immigrants would be in part a reflection of having been the party in power during the most important waves of immigration to Canada. Indeed, the Liberals have for long been considered to be Canada’s “natural governing party” and have been in power for a substantial portion of Canada’s history (Carty, Young and Cross, 2000). The welcoming-party effect might concomitantly explain the alleged recent increase in numbers of immigrants voting for the Conservative Party since it gained office in 2006.

The debate around the effect of the party in power on the immigrant voters has theoretical and practical consequences. One implication is that all immigrants, regardless of their ethnic background or time of arrival, are affected by the same structural factors inherent to moving. The voters’ immigrantness would be an independent cause of their vote decision, and their ethnic origins or the parties’ policies would be largely irrelevant. On the eve of an election, the favoured parties would simply have to activate their immigrant electoral base and get them out to vote. This view fits into the theoretical frameworks that assume

115 that positional issues have minimal eﬀects on vote choice. Immigrant voters share a common immigrant experience, which results in similar behavioural characteristics.

6.1.2 Minority Issues and the Alienation of Racial Minority Vot-

ers

The welcoming eﬀect assumes that issues have very little impact on the vote. Its underlying incumbency logic minimizes the impetus for voters to reason about issues when evaluating parties. In such a theory, what parties stand for is secondary. Yet in reality, parties do hold positions on immigration issues. And they hold positions on minority rights issues too. It is reasonable to presume that such issues, which clearly pertain to many immigrants’ lives, would in fact be salient to immigrants and racial minorities. It is possible that immigrant voters have more in common than just the simple fact of all having moved from one country to another.

In Canada, immigrants and racial minorities are typically thought of as a Liberal electoral base (Blais, 2005; Blais et al., 2002). But the origins and causes of the relationship between the Liberal party and minority groups is less clear. It is possible that immigrant voters inclination to vote for the Liberals rather than the Conservatives is attributable to factors other than immigrantness. Given that many immigrants are also members of racial minorities, it is possible that racial factors might influence voting patterns.1 While immigrant and racial categories overlap (Bilodeau and Kanji, 2010), they are obviously not the same. Unlike welcoming effects, racial effects imply that party positions on issues

119.1% of Canadians identify as belonging to a racial minority. Of these, 65.1% are immigrants. The composition of the Canadian immigrant population has changed signiﬁcantly over the last few decades. While 78.3% of immigrants were of European origin before 1971, 86.3% of immigrants that have arrived between 2006 and 2011 are of non-European descent (Statistics Canada, National Household Survey, 2011).

116 matter. If racial considerations are at play because racial minorities care about minority rights issues, these voters need at the very minimum to be able to distinguish parties’ positions on those issues. As Lewis-Beck et al. (2008) points out, the perceived diﬀerence between parties on an issue is a precondition for that issue to impact vote choice.

There has traditionally been a sharp contrast between Canadian parties on the minority rights issue. The Liberal Party is clearly associated with the idea of multiculturalism, which was enshrined in law by that party through the 1982 Charter of Rights and Freedoms (Clarkson, 2005). It should thus not come as a surprise that the Liberals are perceived by voters as better suited to protect minority rights and promote cultural diversity than are the Conservatives.1 When asked to evaluate parties’ policy positions on those issues in the 1984 and 1993 Canadian Election Study, Canadians consistently rate the Conservatives as the least concerned about racial minorities (see Figure 7 in the Appendix). There is also a sharp contrast between immigrants’ and ethnic minorities’ perceptions and other voters’ perceptions of party capacities to handle minority rights issues. Immigrants and ethnic minorities tend to perceive the Liberal party more favourably and the Conservative party more negatively.

The Conservatives’ reputation on minority rights has deep historical roots (Wiseman, 2007). However, the era between 1993 and 2004 when the Reform Party and Canadian Alliance replaced the Progressive-Conservatives as the main Canadian conservative forces certainly worsened that reputation. Before then, Mulroney’s Conservatives saw the potential for growth represented by racial minorities and made eﬀorts to attract these voters: as Loney (1998) points out, “The Conservatives under Brian Mulroney also saw the electoral

1For the 1965-1988 period, ‘Conservatives’ refers to the Progressive-Conservative Party. Between 1993 and 2000: the Progressive-Conservative party and the Reform Party (1993,1997)/Canadian Alliance (2000). And after 2000: the Conservative Party of Canada.

117 advantages to be gained by promoting links with ethnic groups and promoting multiculturalism policies” (Loney, 1998, 148). These efforts resulted in the enactment in 1988 of the Canadian Multiculturalism Act, which legislated government assistance for the promotion and enhancement of multicultural diversity. However, these Conservative efforts to attract racial minorities were annihilated by the advent of the Reform Party, which replaced the Progressive Conservative Party as the main vehicule for conservative ideas in the country. In 1993, the newly created Reform Party advocated the disbanding of official multiculturalism and a reduction in annual immigration quotas, while being highly critical of the Charter of Rights and Freedoms (Ellis and Archer, 1997; Laycock, 2002). Such issue positions contributed to the party’s reputation as being “too extreme” (Nevitte et al., 2000). This image stuck to the party even after its metamorphosis into the Canadian Alliance, which, after merging with the remnants of the Progressive-Conservative Party, became the present Conservative Party of Canada (Greenspon, 2001).

The Conservative Party’s historically divergent position on minority rights issues might have led to their disfavour among racial minorities. Operating unlike a wedge issue, which attracts voters to a party that would typically not receive their vote (see Hillygus and Shields, 2008), the minority rights issue might have driven potentially conservatively- minded racial minorities away from the Conservative Party. The possibility thus arises that immigrants are attracted to the Liberals and deterred by the Conservatives not because of immigrantness, but because of racial factors.

6.1.3 Cross-Pressured Targetable Segments

The Conservatives have recently made noticeable eﬀorts to improve their branding on racial minority issues (Wells, 2006). According to Flanagan (2011), Conservative strate-

118 gists focused their eﬀorts on the “immigrant voter” after several failed attempts to rally francophone Quebec voters in 2006 and 2008. Immigrant and racial minorities were of particular strategic interest to the Conservatives because many of these voters have conservative positions on issues such as the economy, law and order, and moral questions (Bricker and Ibbitson, 2013). Considering that these voters tend to support more leftist parties such as the Liberals (Bilodeau and Kanji, 2010), immigrant and racial minority voters can be considered cross-pressured.1 Therefore, there was a clear opportunity for the Conservatives to enlarge their electoral base without alienating their traditional voting base. After all, Lusztig and Wilson (2008) show that nativist or anti-immigrant sentiments are not a major cause of division among conservative voters in Canada. If the Conservatives could distance themselves from the extremist and intolerant image they inherited from their Reform/Alliance past, they might be able to turn immigrants and racial minorities into natural Conservative voters. Such a conjecture assumes a dynamic relationship between voters and parties, one in which positional issues play a major role.

The advent of micro-targeting technologies and the change in the socio-economic environment has increased the importance of segmentation-based strategies in American politics (Schantz, 1996a). Increasing cultural and ethnic diversity has also encouraged “the kind of special interest centeredness characteristic of so much contemporary American electioneering” (Gimpel, 1999, 237). Indeed, the tendency to target ethnic communities has been increasingly present in the United States (Leighley, 2001). And “new immigrants have proven capable of inﬂuencing the electoral calculations of party leaders” (Tichenor, 2002, 7). It seems that the Republicans are in fact making great eﬀorts to change their image

1Hillygus and Shields (2008) consider that to be considered cross-pressured, “an individual must not only disagree with the position taken by her own party but also agree with the position of the opposition party.” (Hillygus and Shields, 2008, 56)

119 and increase their appeal among racial minorities (Marbut, 2004; Philpot, 2008). Some of these eﬀorts are targeted toward the Hispanic minority. As the largest minority group in the United States, Hispanic Americans necessarily represent a strategic electorate segment that Republicans cannot ignore. Yet there is more. Hispanic voters are also thought to be cross-pressured: these traditional Democratic supporters are more ideologically conservative (de la Garza and Cortina, 2007; DeSipio, 1996) and tend to care more about traditional conservative issues like morality and national security than other voters (Abra- jano, Alvarez and Nagler, 2008). As such, these voters are members of issue publics that represent subtantial room for Republican growth. Accordingly, Hispanic Americans have increasingly been the target of direct mail, door canvassing and other micro-targeting techniques by Republicans in recent years (de la Garza and DeSipio, 2004).

Micro-targeting strategies are not limited to the United States. The Conservatives have already shown their ability to elaborate strategies that involve targeting specific electorate segments. In 2006, they rebranded their party to fit their electoral goals in part by targeting different segments. At the time, Quebec ridings were appealing to Conservatives who hoped to recreate a nation-wide conservative coalition (Hébert, 2007).1 But their image in the province had been sullied by the anti-Quebec sentiments that the Reform Party espoused. The Conservatives repositioned their party in order to attempt to tap the Que- bec issue public. They promised, for example, an “open federalism” that would provide more recognition and autonomy for the province. The Conservatives also promised to address the possible fiscal imbalance problem between federal and provincial governments, a problem especially important to Quebec at the time. The Quebec issue was not the

1According to some observers, Brian Mulroney led the Progressive-Conservative Party to two succes- sive electoral majorities by bringing together different types of conservative voters: Western Reformists, Ontarian Red Tories, Quebec Bleus, and Blue Grit Maritimers (Watt, 2006). The constitutional debates of the era may have contributed to the implosion of the Mulroney Coalition and the subsequent emergence of the Reform Party and the Bloc Québécois.

120 only issue on which the Conservatives repositioned between the 2004 and 2006 elections (Clarke et al., 2006). They also repositioned on support for universal, publicly-funded healthcare and the firm intention to not legislate on abortion rights in order to be more appealing to potential voters (Paréand Berger, 2008). The Conservatives’ attempt to rally French-Speaking Quebeckers is known to have failed after the party cut funding to the arts and to culture on the eve of the 2008 election (Hébert, 2007). The subsequent electoral disappointment the party faced motivated them to turn their strategic focus away from Francophone Quebeckers and toward immigrant minorities (Flanagan, 2011).

The use of positional issues to target voters who disagree with a party’s position might be a way for parties to enlarge their bases of support. These strategies appear to be working. According to Hillygus and Shields (2008), the more information voters receive about issues on which their attitudes diverge from the position of the party they typically support, the more likely these cross-pressured voters are to defect. Similarly, Stimson (2004) calls conflicted conservatives those voters who say they are conservative but hold liberal positions on specific issues. These voters represent a major source of movement in public opinion because they can be appealed to by all sides of the political spectrum. In Canada, some racial minorities are thought to be cross-pressured voters that represent room for growth for the Conservatives. There is some evidence that the Conservatives, like the American Republicans, have targeted racial minorities in recent years. In a package sent by mistake to a political opponent, Conservatives explicitly mention their intention to target ten “very ethnic” ridings during the 2011 election based on an analysis of voting patterns among different racial communities (de Souza, 2011). According to this document, the Conservatives planned to reach targeted issue publics among these communities with massive advertisement buys in ethnic media.

121 Conservative efforts to reach racial minorities might have had some success. Some of the implications of the results presented in the previous chapters is consistent with that interpretation. Different voters care about different issues. If a party targets appropriate issue-public segments, it can impact political attitudes and ultimately, vote choices. From such a perspective, the possibility emerges that the vote choices of immigrants or racial minorities are less a reflection of their immigrantness or race than they are of their community’s political views and priorities.

6.2 Hypotheses

The diﬀerent lines of reasoning found in the literature explaining immigrants’ political behaviour are amenable to empirical investigation:

Hypothesis 1 : Immigrants tend to support the party that was in power at the time of their arrival

Hypothesis 2 : Racial factors explain voting behaviour better than immigrantness

Hypothesis 3 : Immigrant and racial minorities are cross-pressured issue publics targeted by the Conservatives

6.3 Methodology and Data

The diﬀerent research hypotheses are tested using Canadian Election Study (CES) surveys. The data from thirteen waves of the CES1 allow the analysis to extend over a

1More precisely, this analysis uses the CES data collected during these thirteen federal elections: 1965, 1968, 1974, 1979, 1984, 1988, 1993, 1997, 2000, 2004, 2006, 2008, 2011. The 1972 data are not

122 wide time span and reveal changes in trends. Despite there being diﬀerent investigators and some variance in question wordings, the diﬀerent CES waves were conducted using similar methods of investigation and include comparable sets of questions. These questions provide the sociodemographic and attitudinal information required to conduct the following analysis. The questions of particular importance–those regarding ethnicity and year of immigration–were asked in all CES waves since 1974. Finally, the longevity of the CES gives it an advantage over other data sources; it is amenable to the pooling of data samples and permits cross-time comparisons.

In addition to the CES datasets, the present analysis uses the 2011 Vote Compass (VC) data. These data were gathered from a voter engagement application that was developed in co-operation with CBC/Radio-Canada, Canada’s public broadcaster and largest news media corporation. In addition to thirty positional issue questions, a set of questions capturing users’ sociodemographic backgrounds and political preferences were asked. Moreover, VC data also include party positions on the same thirty positional issues. Party positions were determined by analyzing publicly available party statements. These statements were sent to the parties for review and conﬁrmation. This information permits a unique comparison of voter-party proximity on positional issues. However, despite their obvious assets, these types of data must be treated with care. In order to deal with potential selection bias, all analyses using these data must take into account users’ sociodemographic backgrounds such as age, education, income, gender, language, riding of residence, and their levels of political interest. Despite the selection bias pitfall associated with VC data, this dataset still has obvious advantages. The nearly two million responses collected during the 2011 Canadian federal election campaign are useful for studying small available. For the 1974-1979-1980 CES panel, only the ﬁrst respondents’ survey was kept. The 1980 CES has not been used because no new respondents were added to that sample.

123 electoral subgroups. The present analysis uses these new forms of data cautiously: the use of VC data is restricted to supporting or enlarging results of analyses conducted using CES data.

6.4 Results

Do immigrants and racial minorities actually have a special attachment to the Liberal party and/or an aversion to the Conservative Party?1 This conventional belief can be observed in Figure 6.1 that shows Canadian minorities’ partisan support since 1965. The Liberals historical advantage with the immigrant vote can clearly be seen. However, a Conservative disadvantage does not appear quite so strongly. Nevertheless, the Conserva- tives seem to have noticeably improved their vote share among immigrants in the recent election, a ﬁnding that is consistent with aforementioned targeting speculations. In the 2011 federal election, a greater proportion of immigrants than non-immigrants voted for the Conservatives.2 Despite this fact, a greater division appears among racial minorities.3 And this time, the Conservatives’ disadvantage is clear.

Since 1993, the year that saw the emergence of the Reform Party, Liberal support from both minority groups shows a similar pattern. This observed similarity might explain why these two types of voters are often considered as a single category (see Bilodeau and Kanji, 2010). However, apart from this period, the diﬀerence between the voting

1For simplicity’s sake, this analysis focuses primarily on the Conservative Party and their main historical opponent, the Liberal Party of Canada. Most results concerning the New Democratic Party can be found in the Appendix. 2This ﬁnding can also be observed in IPSOS-Reid 2011 Exit Poll data. According to this survey, the Conservatives scored better than the Liberals in 2011 among both recent and older cohorts of immigrants. Only the NDP had better results among immigrants arriving to Canada in the last ten years. 3Immigrant voters are those not born in Canada. Racial minorities are those respondents who indicate that their ancestors’ origins are non-European (Russians are considered Europeans).

124 Figure 6.1: Minorities’ Partisan Support Over Time

Conservatives Liberals

100 100

75 75

50 50

25 25 Percentage of vote Percentage of vote Percentage

0 0

19651968 1974 1979 1984 1988 1993 19972000 2004 20082011 19651968 1974 1979 1984 1988 1993 19972000 2004 20082011

(a) Immigrants

Conservatives Liberals

100 100

75 75

50 50

25 25 Percentage of vote Percentage of vote Percentage

0 0

19651968 1974 1979 1984 1988 1993 19972000 2004 20082011 19651968 1974 1979 1984 1988 1993 19972000 2004 20082011

(b) Racial minority

Source: Canadian Election Study (CES), 1965 to 2011. Note: The black line represents (i) immigrant and (ii) racial minorities, respectively. The dashed line represents all other voters. Immigrant voters are those not born in Canada. Racial minorities are those respondents who indicate that their ancestors’ origins are non-European.

125 patterns of immigrants and racial minorities is clear, especially when looking at support for the Conservatives. It is therefore relevant to explore the factors explaining the voting behaviour of these two groups separately. This section tests three hypotheses related to immigrantness, race, and issue publics.

6.4.1 A Welcoming Eﬀect?

The defining characteristic of an immigrant is the fact of not having been born in the country where one lives. Therefore, the effect of being an immigrant on the vote should be related to that specific characteristic and not to other factors such as race or culture. Many studies indeed show that immigrantness has an independent effect on the vote and on party affiliation (Black, Niemi and Powell, 1987; White et al., 2008; Wong, 2000). Another expectation related to immigrantness is that immigrants support the party that was in power at the time of their arrival in the country. Given the Liberals’ long reign during the most important waves of immigration to Canada, it is expected that most immigrants would have historically voted Liberal. Yet now that the Conservatives are in power, they should be the ones benefiting from this welcoming-party advantage. Such an explanation only holds if a welcoming-party effect endures.

If immigrants vote in gratitude to the party that was in power when they arrived, there should be diﬀerences in voting patterns among immigrants based on their year of arrival. Also, an increase in immigrant support for the Conservatives should be perceivable since the moment the Conservatives took oﬃce in 2006.

In order to examine these possibilities, all immigrants have been assigned to a party according to their year of arrival. Unfortunately, the data do not permit us to determine

126 whether immigrants who immigrated during an electoral year arrived before or after election day.1 Therefore, those who arrived during an election year that experienced a change of government are dropped for the analysis as it was impossible to determine which party was in power at their arrival. The CES data from all available electoral years have been pooled in order to increase the sample size and maximize statistical power. This methodological choice also follows the stipulation of the welcoming-party effect is structural and consequently, time insensitive. The need to distinguish between the immigrants that arrived during the Conservative reign from those who have been welcomed by a Liberal government is consequential. Yet if immigrants indeed tend to favour the party that was in power at the time of their arrival, this effect might be undetectable in conventional models that include only one variable for immigrants. With some immigrants supporting one party, and others opposing it, the immigrantness effect would be cancelled-out. The welcoming-party effect would be hidden by the aggregation of these two types of immigrants into the same variable. Instead of looking at the effect of being an immigrant on voting for a party, the models presented in Table 6.1 look at the welcoming-party effect.

The results shown in Table 6.1 could not be clearer: there is no detectable welcoming-party effect. This finding applies both to voting intentions and to feelings toward the Liberals and the Conservatives. Thus, immigrants do not seem to express electoral gratitude toward the party that was in office at the time they moved to Canada. These results show little support for the idea that new immigrants would simply endorse the Conservatives because they were in power at the moment of their arrival.2 The next section explores the possibility that “racial” elements play a greater role than “immigrantness” in Canadian

1The CES questionnaire only asks the year that an immigrant immigrated, not the month and day. 2These analyses do not include non-citizen immigrants as the CES only surveys those eligible to vote in Canada.

127 Table 6.1: The Welcoming-Party Eﬀect: A Bonus for the Party in Power?

Focal party Liberals Conservatives Vote Feeling Vote Feeling Welcominga -0.04 -0.00 -0.00 -0.00 Focal party in powerb 0.72∗∗∗ 0.02 -0.47∗∗∗ -0.01 Welcoming X Focal party in powerc 0.03 0.01 0.05 -0.00 Controls Quebec 0.29∗∗ 0.01 -0.39∗∗∗ -0.02∗ West -0.67∗∗∗ -0.04∗∗∗ 0.40∗∗∗ 0.02∗∗ Maritimes -0.09∗ -0.01 0.18 -0.02 Woman 0.08 0.01 -0.04 -0.02∗∗∗ French language -0.36∗ -0.01 -0.03 -0.01 Other language 0.13∗ 0.01∗ 0.11† 0.02 Catholic 0.30∗∗ 0.05∗∗∗ 0.12 0.01∗∗∗ Protestant -0.21∗ -0.00 0.89∗∗∗ 0.11∗∗∗ Other religion 0.42∗∗∗ 0.05∗∗∗ 0.15 0.04∗∗∗ Less than 34 years-old -0.12 0.00 -0.11 0.01 More than 55 years-old 0.04 0.00 -0.06 0.01 Below High School 0.15† 0.03∗∗ -0.26∗∗ 0.00 University Degree -0.10 0.02∗∗∗ 0.13† 0.02 Low Income -0.01 -0.01 0.17∗ 0.02 High Income 0.10 0.00 0.18 0.01 Urban 0.19∗∗ 0.02∗∗∗ -0.12† -0.01 Years in country -0.01∗∗∗ -0.00∗∗∗ 0.01∗∗ -0.00 (Intercept) -0.71∗∗∗ 0.48∗∗∗ -0.88∗∗∗ 0.41∗∗∗ N 5536 5505 5536 5505 Source: Canadian Election Study, 1965-2011 (Pooled and imputed data). Note: The data only include immigrant respondents. The models also include controls for the 13 waves of the CES survey (not shown). Method: Logistic regression (Vote) and least squares regression (Feeling). Dependent variables: Which party did you vote for? (Vote); How do you feel about [party]? (Feeling) aA variable that matches the party in power in the election study and the party in power upon arrival (i.e., A dummy variable that reads 1 if the party in power at the moment of the survey is the same as the party in power on arrival and 0 otherwise). Its coefficient represents the average welcoming effect for the Conservatives (if the focal party is the Liberals) and for the Liberals (if the focal party is the Conservatives); bA dummy variable that indicates whether the focal party is in power at the moment of the survey. Its coefficient represents the fixed effect for the general effect of the focal party being in power among all immigrants in average; cThe coefficient resulting from this interaction represents the marginal effect when the focal party is the welcoming party. † significant at p < .10; ∗p < .05; ∗∗p < .01; ∗∗∗p < .001 voting behaviour.

128 6.4.2 Immigrantness Versus Race

Recall the timelines shown in Figure 6.1. These graphs show that the sharpest contrasts in voting patterns are found between ethnic voters and non-ethnic voters. Note the variation in racial minorities’ support for the Liberals and the Conservatives before and after the entrenchment of the Charter of Rights and Freedoms in 1982. Before 1982, the Liberals enjoyed a clear advantage among visible minorities. The highest support for the Liberals among ethnic voters was on the eve of the 1982 Charter entrenchment.1 But quickly after, Liberal support began to fade. At the same time, Mulroney’s Conservatives were trying to strengthen the Conservative Party’s ethnic constituency by enacting, for example, the Canadian Multiculturalism Act (Loney, 1998). The implosion of Mulroney’s party and the advent of the Reform Party as the principal vehicle of Conservative ideology in Canada makes it practically impossible to determine if Mulroney’s strategy could have translated into an enduring shift in racial minorities’ voting patterns.2 It is somewhat doubtful given the Reform Party’s positions and its members’ comments, which allowed opponents to portray the party as hostile to racial minorities. It seems reasonable to link the decline in ethnic support for the Conservatives during these years to an image problem. Previous CES studies showed that the Reform Party and the Canadian Alliance both suﬀered from being perceived as intolerant and extreme parties (Blais et al., 2002; Nevitte et al., 2000). Nevertheless, there is observable growth in immigrant and visible minorities’ support for the Conservatives between the 2008 and 2011 elections. The question is: why?

Comparing immigrants’ and visible minorities’ voting patterns side by side is not suﬃcient

1The 1980 CES data show similar results, which is not surprising given the fact that the 1980 sample was constituted exclusively of 1979 CES respondents. The 1980 CES data were not included in this analysis to avoid autocorrelation problems. 2For the 1993, 1997 and 2000 Elections, this analysis merges the information for the Progressive- Conservative Party and the Reform Party/Canadian Alliance. In December 2003, the Canadian Alliance and the Progressive-Conservative Party merged to become the current Conservative Party of Canada.

129 to determine the independent effects of each of these factors on vote choice. Especially since these two categories strongly overlap (Bilodeau and Kanji, 2010). Moreover, it is possible that immigrants and racial minorities share common sociodemographic characteristics that predispose them to vote a certain way, meaning that their behaviour is not necessarily determined by race per se. For example, the fact that immigrants tend to be more educated and come from various religious and ethnic backgrounds might better explain their voting patterns than the experience of having left their country of birth. In this case, the independent effect of being an immigrant–immigrantness–on vote choice needs to be distinguished from the voting behaviour of immigrants as a social group. To assess the independent effects of immigrantness and race on the vote for each party, we turn to regression analyses. Figure 6.2 shows the effect of immigrantness and ethnicity on the propensity to vote for the Conservatives and the Liberals from 1968 to 2011 controlling for sociodemographic factors.1

The results illustrate the importance of distinguishing between the effect of race and immigrantness. The Liberal advantage appears to be driven principally by race, not immigrantness. When race is not controlled for, the effect of being an immigrant has a significant independent effect on voting for the Liberals during the elections following the entrenchment of the Charter of Rights and Freedoms (Results shown in the Table 17 in the Appendix). The effect disappears, however, once race is taken into account. In fact, once race is taken into account, the independent effect of immigrantness on the vote is never statistically significant in any year for either party between 1968 to 2011. The results also confirm the Conservative Party’s difficulties in attracting racial minorities. As expected, these difficulties improved in 1988, when Mulroney made efforts to reach

1Summaries of the results for all the variables included in these models can be found in the Tables 17, 18 and 19 in the Appendix.

130 Figure 6.2: Independent Eﬀects of Immigrantness and Race on the Vote

Liberals Conservatives 0.15

0.10 0.10

0.05 ● 0.05 ● ● ● ● ● ● ● 0.00 ● ● ● ● 0.00 ● ● ● ● ● ● ● ● ● ● −0.05 ● ●

Effect on Vote Effect −0.05 on Vote Effect

−0.10 −0.10

−0.15

1968 1974 1979 1984 1988 1993 1997 2000 2004 2006 2008 2011 1968 1974 1979 1984 1988 1993 1997 2000 2004 2006 2008 2011 Year Year (a) Immigrantness

Liberals Conservatives

0.4 0.2

0.2 ● ● ● ● ● ● ● ● ● 0.0 ●

● ● ● ● ● ● ● ● 0.0 ● ● ●

● ● ● Effect on Vote Effect on Vote Effect −0.2

−0.2

1968 1974 1979 1984 1988 1993 1997 2000 2004 2006 2008 2011 1968 1974 1979 1984 1988 1993 1997 2000 2004 2006 2008 2011 Year Year (b) Racial minority

Source: Canadian Election Study (CES), 1965 to 2011. Method: Least-squares regression. Vertical bars indicate 95% confidence intervals, which reflect a significance level of 0.05 Dependent variable: Which party did you vote for?

131 racial minorities, and worsened in 1993 with the advent of the Reform Party. Statistically significant negative effects of race on the Conservative vote are actually just found for the years following 1993. However, the independent negative effect of race on voting for the Conservatives did not decrease in 2011 as was expected. That said, when a model is estimated to uncover the determinants of being a new Conservative voter in 20111, race does appear as a significant factor (see Table 6.2). While immigrantness does not on its own appear to be a statistically significant influence on the vote, the interaction between race and immigrantness–being an immigrant and a racial minority–is a strong predictor of being a new Conservative voter in 2011.

Is it then possible that factors other than race pushed some immigrant minorities to vote for the Conservatives in 2011? And is it also possible that some members of racial minorities were more sensitive to the Conservative Party’s appeals than others? If the Conservatives actually targeted racial minorities only in a select few ridings, it is not surprising that racial eﬀects are not present in analyses that examine the electorate as a whole. The next section examines these possibilities and takes a closer look at speciﬁc, allegedly targeted ridings.

6.4.3 Racial Minorities as Targeted Cross-Pressured Voters

How fixed is the negative relationship between racial minorities and the Conservatives? The different theories in the literature discussed above offer different answers to that question. More structural explanations, which examine the common experience shared by immigrants—or immigrantness—assume that immigrants’ relationships with parties

1New Conservative voters are those who voted for the Conservatives in 2011, but recall to have voted otherwise in 2008.

132 Table 6.2: Who Shifted? The New Conservative Voters of the 2011 Election

Shifted for the Conservatives in 2011 (1) (2) (3) (4)

Immigrant 0.25 — 0.17 -0.25 Race — 0.49† 0.45 -0.77 Immigrant X Race — — — 2.22∗∗ Controls Quebec -0.21 -0.18 -0.19 -0.20 West -0.06 -0.04 -0.03 -0.04 Maritimes 0.34 0.32 0.32 0.25 Woman 0.23 0.23 0.23 0.26† French language -0.09 -0.08 -0.10 -0.14 Other language 0.58∗ 0.60∗∗ 0.55∗ 0.54∗ Catholic 0.03 0.08 0.07 0.09 Protestant 0.07 0.11 0.10 0.10 Other religion -0.17 -0.10 -0.18 -0.30 Less than 34 years-old 0.29 0.32 0.34 0.37 More than 55 years-old -0.35∗ -0.28 -0.30† -0.25 Below High School 0.48∗ 0.53∗ 0.55∗ 0.55∗ University Degree -0.38∗ -0.38∗ -0.38∗ -0.41∗ Low Income 0.11 0.06 0.07 0.07 High Income -0.07 -0.02 -0.04 -0.02 Urban -0.01 0.00 -0.02 -0.06 (Intercept) -2.58∗∗∗ -2.58∗∗∗ -2.68∗∗∗ -2.64∗∗∗ N 2705 2685 2667 2667 Source: Canadian Election Study, 2011. † significant at p < .10; ∗p < .05; ∗∗p < .01; ∗∗∗p < .001 Method: Logistic regression. The coefficients represent effects on log odds. Full results with standard errors presented in Table 20 in the Appendix. Note: The dependent variable is a dichotomous variable comprising those who voted Conservative in 2011, but recall to have voted otherwise in 2008. emerge from their time of arrival in the country. Yet this explanation does not satisfacto- rily explain the recent increase in minority support for the Conservative Party.1 It seems

1It does not mean that such structural explanations are not useful to explain some other aspects of immigrants’ political attitudes and behaviour. As specified by White et al. (2008), different theories are relevant to different approaches to an object under study: “Some focus on such orientations as the acquisition of partisan identification or interest in politics, while others focus on such behaviors as the decision to vote or not. There is no reason to presume that the pathways to adaptation will be identical for each dimension of engagement” (White et al., 2008, 270).

133 that race matters more than immigrantness in explaining the recent increase in Conser- vative support among minorities. The Conservative Party’s past positions on minority issues had undoubtedly alienated racial minority electoral segments. But given that the Conservatives have made great eﬀorts to court the minority vote in recent years, might we expect a change in the sociodemographic proﬁle of their supporters? It is possible that the Conservatives’ history of negatively positioning themselves on minority issues prevented some racial minority voters who otherwise agree with the party to not vote for it. Figure 6.3 explores this possibility.

Figure 6.3: Immigrant and Racial Minorities’ Ideological Positions and Resistance to the Conservatives

0.4

0.3

0.2 Density

0.1

0.0

Leftist Centrist Rightist Ideological Scale (a) Distribution (b) Vote by ideology

Source: Vote Compass, 2011 Canadian Federal Election. Note: The dashed and dotdashed lines represent (i) immigrant and (ii) racial minorities, respectively. The solid line represents all the other voters. Immigrant voters are those not born in Canada. racial minorities are those who indicate that their ancestors’ origins are non-European. The left-right ideological factor scale is made up of 23 items measuring positional issue attitudes on a ﬁve-point scale (see Table 22 in the Appendix for details). The probability of voting Conservative is based on respondents’ personal evaluation of their probability of voting for the party. The question was posed as “What is the probability that you will vote for the Conservative Party?”

The two graphs presented in Figure 6.3 use VC data to suggest that racial minorities

134 might be cross-pressured voters who could represent potential for Conservative growth. These voters form a substantial targetable electoral segment. Despite their conservative attitudinal inclinations, these targetable voters had previously not voted Conservative because of the party’s positions on minority issues. Figure 6.3 (a) shows the distribution on an aggregate issue scale1 for immigrants, racial minorities, and other Canadians. This graph provides support for the idea that neither immigrant nor racial minorities sit on the left side of the political spectrum, compared to other Canadians. Despite being traditionally associated with the Liberals, many racial minorities actually hold conservative positions on key issues. The average ideological position for the three groups is very similar.2 However, the shape of the distributions shows that immigrant and racial minorities are slightly less polarized than other Canadians; a non-negligible portion of immigrant and racial minority groups is positioned at the centre-right of the political spectrum, while the mode of the distribution for other Canadians is clearly situated at the centre-left. When we look at voting patterns for the Conservatives in relation to these aggregate issue positions, Figure 6.3 (b) indicates that the Conservatives had diﬃculties with racial minorities—but not immigrants—positioned at the centre-right of the political spectrum. These descriptive statistics support the idea that the Conservatives have a strategic incentive to advertise their positional issues to these cross-pressured racial minorities. The large Canadian ethnic population might represent great potential for growth for the Conservative Party.

According to the political marketing literature, a micro-campaign targeting cross-pressured racial minorities that care about and hold positions on issues favourable to the Conserva-

1The left-right ideological factor scale is made up of 23 items measuring positional issue attitudes on a ﬁve-point scale (see Table 22 in the Appendix). 2On a scale to 0-1 (0 representing the most leftist position and 1, the most rightist), the average position for immigrants is 0.33 (s.d.=0.13); 0.32 (s.d.=0.13) for racial minorities; and 0.34 (s.d.=0.12) for other Canadians

135 tive Party could have beneﬁts for that party. As previously discussed, the Conservatives targeted ten “very ethnic” ridings in the 2011 federal election. Flanagan (2011) argues that the Conservatives began to target various ethnic groups such as Korean, Chinese, Vietnamese, and Filipino groups following the 2008 election. The political behaviour of these voters does not necessarily spring from their immigrantness per se or from their race. The Conservatives saw these communities as being potentially more receptive to conservative ideas because many of the communities’ constituents are fairly religious and have Christian backgrounds. An example of Conservative “ethnic” targeting is found in the case where, on the eve of the 2011 election, the Conservatives advertised their position on same-sex marriage in some but not all ethnic newspapers distributed in the region of Toronto. But did this Conservative targeting strategy work? A perspective attempting to answer that question by focusing on electoral strategy theory is diﬃcult to demonstrate empirically. Moreover, combining considerations of the dynamic nature of social groups, the heterogeneous distribution of individual issue salience and between-campaign party positioning complicates the developement of statistical models.

Micro-campaigning effects are expected to distort the effects of these factors in the targeted ridings. It is reasonable to assume that heterogeneous campaign efforts result in stronger independent effects on the vote of the factors primed by the micro-campaign in targeted ridings than in non-targeted ridings. For example, if the Conservatives sys- tematically targeted racial minorities—regardless of racial minorities’ positions on issues and their issue salience—in some ridings more than others, the independent effect of race on the vote should be significantly stronger in these targeted ridings. Similarly, if the Conservatives had only presented their issue positions in advertisements in targeted ridings, the effect of positional issues should be different in these ridings. As specified in the previous chapter, interactive terms can be used in statistical models to capture such

136 heterogeneous effects. In order to capture micro-campaign effects, the next model of the probability of voting for the Conservatives includes interactive terms between a dummy variable representing the targeted ridings and the different factors of interest: race and two different issue-proximity measures.

The two different measures of proximity are distinct. The first measure looks at the proximity of the respondents and the Conservatives on all issues.1 The second measure focuses on the proximity between the issue that the Conservative Party and respondents deemed most important.2 This second proximity measure is more in line with the purpose of detecting micro-campaign effects. As discussed throughout the thesis, a micro-campaign implies detecting and targeting segments of the population that are most susceptible to influence. If the Conservatives’ micro-campaign was successful in reaching the specific issue publics that it targeted, the effect of issue-public proximity should be stronger in targeted ridings.

The results show that the effect of race is not statistically different in the targeted ridings than elsewhere. Given the popular discourse about Conservatives’ targeting of “very ethnic” ridings, this result might seem surprising. But recall that the expectations about the Conservatives’ micro-campaign were that racial minorities were targeted because of their positions on issues, not because of their race per se. However, the aggregate proximity measure does not appear to have more of a significantly different effect in “very ethnic” targeted ridings than elsewhere. This last result does not mean that aggregate

1The steps taken to build this aggregate proximity measure are as follows: 1) The difference between respondents and the Conservatives is calculated for each of the 23 main VC positional issue questions; 2) These single-issue differences are then summed into an aggregate difference measure; and 3) The measure is reversed in order to transform the maximum score on the aggregate difference measure into the lowest score on the aggregate proximity measure, and vice versa. 2This second proximity measure is calculated like the aggregate proximity one, but it takes into consideration only the VC questions that relate to the issue that respondents identified as the most important to them personally.

137 Table 6.3: Targeted Ridings: Race, Proximity on All Issues, or Issue Publics?

Probability of Voting Conservative (1) (2) (3) Targeted ridings X Race −0.006 Targeted ridings X Proximity −0.002 Targeted ridings X Issue Public 0.059∗∗∗ N 483, 552 483, 552 483, 552

Source: Vote Compass, 2011 Canadian Federal Election. Method: Least squares regression. Full regression results, including statistical controls, are presented in Table 21 in the Appendix. Note: The Targeted ridings represent the “very ethnic” ridings targeted by the Conservatives in 2011: Elmwood-Transcona (MB); York Centre (ON); Richmond Hill (ON); Brampton-Springdale (ON); Eglinton-Lawrence (ON); Vancouver South (BC); Bramalea-Gore-Malton (ON); Burnaby-Douglas (BC); Newton-North Delta (BC); and Mount Royal (PQ). The probability of voting Conservative is based on respondents’ personal evaluation of their probability of voting for the party. The question was posed as “What is the probability that you will vote for the Conservative Party?” ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

issue proximity does not matter; actually, it appears as the most important determinant in the model (see results in the Table 21 in the Appendix). This effect is just not unique to targeted ridings. Finally, the most significant variable for our purpose, the issue-public proximity, has a significantly different and positive effect on the probability of voting for the Conservatives in targeted ridings, when compared to other ridings. This last result provides some support for the success of a Conservative issue-based targeting strategy in these ridings. But it does not definitively determine that the targeting strategy was successful in attracting new Conservative voters. Issue-based micro-targeting tactics like direct mail are not limited to only those voters who represent potential for party growth. Especially in close-race ridings, these tactics are also used to encourage voters that are already Conservative partisans to get out and vote (Dale and Strauss, 2009; Parry et al., 2008).

138 Figure 6.4: Riding-Level Eﬀect of Issue-Public Proximity on Being a New Conservative Voter

0.8 ●

●

●●

● ● ● ●● ●● ●●● ●●● ●●● ●● ●●●●● ●●●●● ●●● ●●● ●● ●● ●●●●● ●●●●●● ● ●●●● ●●● ●● ●●●●●● ●●●●● ●●●● ●●● ●●●● ●●●●● ●●●●●●●●● ●●●●●●● 0.4 ●●●●●●●●●●● ●●●●●● ●●● ●●● ●●● ●●●●●●●● ●●●● ●●●●● ●●●● ●●●●●● ●●●●●● ●●●●●●●● ●●●●●● ●●●● ●●●●●●● ●●●● ●●●●● ●●●●●●●●● ●●●●●●● ●● ●●●●●●● ●●●●● ●●●●●● ●●●●●●●●●●● ●●● ●● ●●●●●● ●●● ●●●●● ●●●●● ●● ●● ●● ●●●●● ●●●●●●● ● ●● ●●●●●● ● ●●● ●● Positional Issue Effect Positional ●

● ● 0.0 ● ●

Electoral Ridings

Source: Vote Compass, 2011 Canadian Federal Election. Dependent variable: Being a new Conservative voter: A dichotomous variable comprising those who voted Conservative in 2011, but recall to have voted otherwise in 2008. Method: Least squares regression (308 iterations). Note: The black lines represent the “Very ethnic” ridings targeted by the Conservatives in 2011: From right to left, Elmwood-Transcona (MB); York Centre (ON); Richmond Hill (ON); Brampton- Springdale (ON); Eglinton-Lawrence (ON); Vancouver South (BC); Bramalea-Gore-Malton (ON); Burnaby-Douglas (BC); Newton-North Delta (BC); and Mount Royal (PQ). The grey lines represent the other ridings.

139 6.5 Discussion

This chapter explored whether the behaviour of Canadian political parties can be said to deviate from the brokerage politics model, wherein positional issues are cast aside in favour of valence issues and leader images. It suggested that, contrary to what the brokerage model might expect, Canadian parties indeed use positional issues in their electoral strategies. Speciﬁcally, they seem to use these issues to leverage support among the groups to which the issues are salient, by way of micro-targeting strategies.

The difficulty of distinguishing between the effects of campaign strategies on different types of voters demonstrates the limits of currently available public opinion data. Larger datasets allow for more fine-grained, micro-level analyses. For instance, Figure 6.4 shows the independent effects of issue-public proximity on the probability of being a new Con- servative voter in 308 Canadian ridings. The independent effect of issue-public proximity is significant in nearly all ridings. The black lines represent the ten targeted ridings. Note that the Conservatives won seven of these targeted “very ethnic” ridings. And the three ridings that they failed to win—Mount Royal, Newton-North Delta, and Burnaby- Douglas—are also those where the effect of issue-public proximity is weakest of the ten targeted ridings. However, even the VC sample is not large enough to detect significant results in comparisons of campaign effects between individual ridings. Still, the results presented in this chapter offer a useful starting point for future research. More effort certainly needs to be put into uncovering the effect of modern campaigning:

The question “do campaigns matter?” has long been debated in the political science literature. We consider this a settled question: of course campaigns “matter,” but the more interesting issues are for whom and under what con-

140 ditions campaigns matter. (Hillygus and Jackman, 2003, 583)

Further research could triangulate public opinion data with analyses of parties’ micro- campaign messages. Moreover, large-sample surveys should add questions probing whether respondents were contacted by specific political parties over the course of the campaign and whether respondents know the parties’ positions on issues. Such information would render possible a more direct study of the effects of campaign tactics on voters. Ultimately, collecting larger samples would allow for more fine-grained analyses, which could in turn provide more empirical evidence for micro-campaign effects at the individual-riding level. It could also enable the study of even smaller electorate subgroups. Further analyses of VC data actually show a difference in issue attitudes between Hispanic and Asian Cana- dians (see Figure 10 in the Appendix). But even larger samples would make it possible to study more directly the effect of different issues on different racial minorities in different ridings. These research opportunities are especially relevant as they relate to current political party behaviour. As previously discussed, Flanagan (2011) argues that the Con- servatives targeted specific racial minority groups, but not others. The precise reasons behind such a decision and the various effects that this decision had on voters should be further analyzed. Conservative targeting efforts were also not limited to racial minorities. Religious minorities that are not traditional Conservative electoral segments, such as Jews, have also been in the party’s sights over the last few years (Martin, 2010). This fact further reinforces the idea that what matters to understanding campaign strategy, more than immigrantness and race, is detecting targetable communities that represent potential for electoral growth. The last chapter further explores such a possibility.

141 Chapter 7

Digging Deeper: The Micro-Targeting of Jewish and Muslim Voters

The results presented in the previous chapter indicate that race matters more than immigrantness in explaining voting behaviour in Canada. Although the initial supposition might be that racial minorities’ voting behaviour is homogenously influenced by individuals having had a common racial experience, the empirical evidence seems to suggest otherwise. Specifically, more detailed analyses reveal that there is a heterogeneity in the influences driving the voting behaviour of various racial groups. The results demonstrate the need to fragment large concepts such as immigrantness and race into subcategories that allow us to examine smaller, more specific communities. Recall the distinction in issue attitudes between Hispanic and Asian Canadians. Such differences in attitudes seem much less probably attributable to racial genes than to these individuals being members

142 of distinct cultural communities. Indeed, the various cultural communities of Canada are not all alike. Just like women are concerned about issues that relate more specifically to their unique social experiences (O’Neill, 2002), so different cultural groups cultivate unique political behaviours in their constituents. This situation explains why members of different cultural communities do not care about the same issues. Identifying what matters to different communities can be useful for political parties seeking room for growth. According to Flanagan (2011), Conservatives indeed targeted fiscally conservative ethnic and religious communities during the 2011 Canadian federal election campaign. And during the same-sex marriage debate in Parliament, the Conservative party advertised its position on same-sex marriage in select ethnic newspapers.

From the perspective of political marketing, the fact of psychological attachment to a targetable group is more important than the specific characteristics that define different groups.1 Given this idea, Jewish and Muslim cultural communities can be studied in a way similar to the way groups based on country of origin have heretofore been studied.

This chapter aims to push further the ideas already developed in this thesis through an examination of the recent evolution of the Muslim and Jewish vote. It is known that the Conservative party has targeted the Jewish community in the past by making the Israeli cause a cornerstone of their foreign policy: “Though Muslims outnumbered Jews by two to one in Canada, the Jewish community was more politically impactful. Harper was aware, for example, that he stood to gain a major advantage in the Canadian media with

1This conception of a social group is similar to the one described by Lewis-Beck et al. (2008). For these authors, group identity can prompt political behaviour: “The reality of a group can be psychological, and that reality can change our behaviour. When we feel close to our union, that gives the union a separate existence, and it aﬀects our actions. [...] These psychological interpretations color our attitudes and acts. If we know that the union opposes NAFTA, we as individuals will tend to oppose it, because of our feelings for the union” (Lewis-Beck et al., 2008, 306).

143 his position” (Martin, 2010, 82).1 According to Flanagan (2011), the consequence of such a one-sided position on the Israeli issue was the alienation of the Muslim community.

It is reasonable to assume that citizens who identify with the Muslim or Jewish communities are likely to care more about the government’s position on the Israeli question than are members of other ethnic or religious groups. Members of these two religious communities are more likely to form an issue public based on that particular political issue. This does not mean that all members of those two ethnic groups care about the issue nor that all non-members do not care. Nevertheless, ethnic communities might be proﬁtable electoral segments for parties to target, especially because community members are often concentrated in speciﬁc geographic areas where they share a community space and a media sphere. The Israeli issue was used as a wedge issue to attract voters from the Jewish community who typically vote Liberal. Interestingly, the only Quebec riding that the Conservatives targeted before the 2011 election was the riding of Mount Royal, which is home to the second largest Jewish population in Canada and which is on the list of the ten “very ethnic” ridings targeted by the party. The riding with the largest Jewish population in Canada, Thornhill in Ontario, which was once a Liberal stronghold, was won by the Conservatives in 2008 and has since been retained.

A quick glance at the ideological distribution of Jewish and Muslim voters in Figure 7.1 reveals that there is a clear ideological divide between these two religious communities. While Muslims are more ambivalent on issues and are more centrist, Jews are more left- wing. However, when asked to evaluate the probability of voting for the Conservatives, the two religious groups are polarized. Notwithstanding ideology, Jewish voters are always more willing to vote for the Conservatives than are other Canadians, and Muslim voters

1Actually, according to the 2011 Canadian National Household Survey, there are 329,500 Jews and 1,053,945 Muslims in Canada.

144 Figure 7.1: Jewish and Muslim Voters’ Ideology and Propensity to Vote Conservative

0.4

0.3

0.2 Density

0.1

0.0

Leftist Centrist Rightist Ideological Scale (a) Distribution (b) Vote by ideology

Source: Vote Compass, 2011 Canadian Federal Election. Note: The dashed and dotdashed lines represent, respectively, respondents identifying as (a) Muslim and (b) Jewish. The solid line represents all the other voters. The left-right ideological factor scale is made up of 23 items measuring positional issue attitudes on a ﬁve-point scale (see Table 22 in the Appendix for details). The probability of voting Conservative is based on respondents’ personal evaluation of their probability of voting for the party. The question was posed as “What is the probability that you will vote for the Conservative Party?” are always less willing to vote for the Conservatives.

It is impossible to demonstrate conclusively that the apparent polarization of Jewish and Muslim voters on the propensity to vote Conservative is a result of the Israeli issue. Some scholars believe that this might be the case (see Srebrnik, 2014). However, this issue question has not been asked in public opinion studies large enough to allow investigation at such a micro-subgroup level. Nevertheless, it is still possible to probe the question using indirect means. If the recent positioning of the Conservatives on the Israeli issue caused some of the apparent polarization between Jewish and Muslim voters, we should be able to detect a change in support for the Conservatives among these groups in time.

145 We should expect that Jewish voters have become more supportive of the Conservatives over the years while Muslim voters have become less supportive. Despite sample size problems, diﬀerent CES waves can be used to test this possibility. Table 7.1 shows the eﬀect of being Jewish and Muslim on the vote when controlling for other sociodemographic factors. Table 7.1: The Canadian Jewish and Muslim Vote (1968-2011)

Vote choice (CES) Vote intention (VC) 1968-1988 1993-2000 2004-2006 2008-2011 2011 Liberals Jewish 1.12∗∗∗ 0.10 2.01∗∗ -0.32 -0.27∗∗∗ Muslim -0.14 0.23 12.80∗∗∗ 2.92∗∗ 0.91∗∗∗ New Democrats Jewish 0.74∗ 1.50∗∗ 0.78 1.52 -0.17∗ Muslim -0.82 0.48 11.44∗∗∗ -0.96 0.78∗∗∗ N 6941 4037 3044 3581 105 903 Source: Canadian Election Study, 1968-2011. Vote Compass, 2011 Canadian Federal Election. Method: Multinomial logistic regression (Base category: Conservatives). The coefficients represent effects on log odds. Note: The models also include a set of socio-demographic controls (not shown). † significant at p < .10; ∗p < .05; ∗∗p < .01; ∗∗∗p < .001

The small number of Jewish and Muslim voters present in each CES sample requires a pooling of many CES samples taken over time.1 Despite the lack of precise information for each election, some trends are apparent. First, Jewish and Muslim voters have contrasting political behaviour. Second, the Conservative party’s long-standing difficulty with minorities is apparent. Being a Muslim voter is a very strong predictor of not voting for the Conservatives. This result is strongest and most statistically significant during the period for which we have the most recent data.2 Jewish voters seem more volatile and seem to 1When models are estimated for every CES wave separately, the results are similar, despite the large standard errors of the coefficients for the variables representing Jews and Muslims. Linear regression models also provide similar results, which are summarized graphically in Figures 11,12, and 13 in Appendix. 2Reflecting Canadian demography at the time, very few Muslim voters were included in older CES surveys.

146 have oscillated from the Liberals to the NDP during the 1993-2000 period. Nevertheless, they exhibit a relatively consistent negative relationship with the Conservative party. It is perhaps interesting to note that the relationship between being Jewish and voting Conser- vative reversed during the 2008-2011 period. The lack of statistical significance indicates that these coefficients cannot be considered statistically different from zero, and that it is ultimately impossible to determine the direction of the relationship between Jewish voters and any of the main federal political parties during the 2008-2011 period. Although the 2008-2011 result fails to attain statistical significance, it still suggests that Jewish voters moved from opposition to the Conservative party to greater acceptance of it during that period. The small number of Jewish and Muslim respondents in the CES data and the pooling into time periods precludes the direct observation of these subgroups’ political behaviours during the 2011 election campaign. For that reason, we turn again to the larger Vote Compass data.

A voting model is estimated using VC data. This model is nearly identical to the ones run with the CES data; it is presented in Table 7.1. Its results do not contradict those emerging from CES models. One main advantage of VC data is their ability to allow for the examination of subgroups; specifically, the 2011 VC data include 2924 respondents identifying as Jewish or Muslim.1 It should thus come as no surprise that VC data can detect statistically significant effects for smaller groups better than can surveys the size of the CES. Some scholars have already shown that the validity of multivariate results estimated with VC data is comparable to the validity of the results obtained from more traditional surveys (Fournier, Gélineau and Harell, 2014). The results of the VC model support the idea of a new Jewish-Muslim polarization, yet one that favours the Conservatives. More-

1In comparison, the number of CES respondents identifying as Jewish or Muslim is much smaller: 189 (1968-1988), 167 (1993-2000), 103 (2004-2006), and 140 (2008-2011).

147 over, the relationships ﬁt the expectations for support of the various parties: all other factors held constant, Muslim voters clearly oppose the Conservatives, while Jewish voters clearly support them. The inverse relationship is found for support for the Liberals. The eﬀect of being Muslim is particularly substantive; it is the strongest sociodemographic predictor for voting for the Liberals and against the Conservatives.

Often, in conventional voting models, the effect of being a member of a particular community is obscured by the use of different variables that encompass both immigrantness and ethnicity. The effect is also obscured by the fact that ethnic respondents have religions other than Protestantism and Catholicism or speak languages other than English and French. A single variable indicating voters from ‘other religions’ would never capture the alleged widening polarisation between voters of Jewish and Muslim faiths. By definition, if there was a polarisation, the positions of Jews and Muslims would simply cancel out because the effects are in opposite directions.1 There are therefore clear advantages to considering different ethnic or religious groups independently of one another. However, it is important to avoid essentialistic conceptualizations of social groups. These groups are vehicles of socialization and their members should not be considered as having fixed characteristics. Ultimately, these conceptual challenges add to the problems of sample size that are often encountered when dealing with subgroups. Conclusions about the ef- ficacy of the Conservative targeting strategy must therefore rest on indirect evidence, in particular, on the polarization effect that the Conservatives’ foreign policy is alleged to have had on Jewish and Muslim voters.

Why is there an independent eﬀect of being a member of a religious community on the vote? One possibility is that religious communities have speciﬁc political interests that

1Of course, cancelling-out eﬀects also depend on the strengths of the relationships and the sizes of each of the social groups.

148 may or may not correspond to different parties’ political offers. Positional issues might be what link these voters to political parties. As it so happens, when measures of respondents’ positions and personal saliences on foreign policy are added to the previous regression model (see Table 23 in the Appendix), the effects of community membership are greatly reduced.1 That finding is consistent with the interpretation that positional issues matter to the voters that care about them. Such an idea constitutes the core of the issue-public hypothesis. Given the Conservative party’s general position on foreign policy, the relationship appears to be in the right direction: the greater one’s support for military involvement abroad, the more likely one is to vote Conservative. Also, the significant interaction between personal salience and one’s position on the foreign policy issue supports to some degree the idea of salience-based heterogeneity in the effect of positional issues. In future research, more precise questions that tap the Israel issue more directly are expected to provide even clearer results.

1The positional issue measure for foreign policy is a question regarding the military involvement of Canada abroad. The individual salience measure is a single question directly tapping respondents’ self-reported interest in foreign issues.

149 Chapter 8

Conclusion

The Canadian political landscape has traditionally been considered as characterized by a system of brokerage politics. In a brokerage politics system, political parties do not take strong, ideological positions on issues in order to avoid alienating portions of a highly fragmented electorate (Bickerton and Gagnon, 2004a; Carty, Young and Cross, 2000). Instead of campaigning on positional issues, parties campaign on leader images and valence issues—issues about which there is no disagreement, such as having a good economy or low crime rates (Clarke et al., 1984). Given that political parties do not pronounce themselves on issues, it is difficult for issues to impact citizens’ vote choices. The irrelevance of issues is exacerbated by citizens’ general lack of political sophistication. People are considered to be insufficiently informed about politics to have concrete issue positions upon which they can base vote choices (Converse, 1964). Although the news media place significant emphasis on issues, the empirical evidence for their relevance is uneven at best. The conventional wisdom is that positional issues do not matter much to citizens. Coupled with the idea that voters have strong partisan biases, this wisdom leaves little room for party’s positions on issues to affect electoral outcomes (Berelson, Lazarsfeld and McPhee, 1954; Campbell et al., 1960; Converse, 1964). There is some

150 empirical support for the consistency of these discouraging ﬁndings over time (Lewis- Beck et al., 2008; Miller and Shanks, 1996) and across contexts (Blais et al., 2002; Butler and Stokes, 1969; Converse and Pierce, 1986).

Despite the early recognition of the limits of citizens’ political competence, many studies since have tempered or contradicted the conclusion that citizens are largely politically unsophisticated. Some scholars contend that the measurement methodologies used by these early studies exaggerate the inconsistency of citizens’ issue attitudes (Achen, 1975; Ansolabehere, Rodden and Snyder Jr., 2008). Others suggest that although voters might not have full political information, they can still engage in valence-issue voting and handle retrospective incumbent evaluations (Fiorina, 1981; Key Jr., 1966). And not all scholars dismiss the importance of positional issues. An entire literature focusing on the positions of voters and parties on issues has developed parallel to the Columbia and Michigan school models (see Budge et al., 2001; Downs, 1957; Enelow and Hinich, 1984; Rabinowitz and Macdonald, 1989). The debate about positional issues became more nuanced when the conditions necessary for positional issues to matter were relaxed. In part, these conditions were relaxed with some scholars contending that issues could be conceptualized into different types, such as “easy issues” or “wedge issues” (Carmines and Stimson, 1980; Hillygus and Shields, 2008). The conditions were further relaxed when some authors argued that not all voters care about and are affected by the same issues (Krosnick, 1990), and that the effect of issues can differ depending on the competitive or economic context of the voter (Vavreck, 2009). Ultimately, the cumulative literature on positional issues is very fragmented. One needs to take a genuine leap of faith to answer even the simplest of questions: do positional issues matter?

This thesis has aimed to show that positional issues can matter and that political parties

151 indeed have an incentive to elaborate electoral strategies based on positional issues, and to not simply engage in brokerage-party type of behaviour. The results show that positional issues can affect voting behaviour and are thus fundamental tools in the toolboxes of political parties. Different tests suggest that previous work might have underestimated the importance of positional issues because of a lack of attention to measurement issues and to the heterogeneity of mass publics. Methodologically, the effect and stability of positional issues is substantially increased once issues are operationalized using multi-item scales and once measurement error is taken into consideration. The impact of scaling might explain in part the sharp contrast between the statistical relevance of values and that of positional issues found in previous studies. While values are usually operationalized using scales, positional issues are typically operationalized using single-items. This method of measurement might have led to the underestimation of the impact of positional issues compared to values. The effect of positional issues can also be redeemed by considering the presence of salience-based heterogeneity in the electorate. The political behaviour literature typically conceptualizes the heterogeneous effects of issues on the vote in a dichotomous way: issues can matter to politically-informed elites and cannot matter to the uninformed masses (Converse, 1964; Zaller, 1992). The analyses presented in this thesis suggest that, despite the undeniable structural effects of education, political information, and political interest on overall sophistication, political expertise might be more fragmented and widespread than has been previously considered. Indeed, the results show that voters do not need to be political Einsteins to care about, know about and have stable attitudes about a small set of issues.

There is reason to believe that the aforementioned results regarding measurement of positional issues and salience-based heterogeneity are generalizable outside the Canadian context. The results showing the eﬀects of issue scales are already a reproduction of

152 results that were initially observed in the American context by Ansolabehere, Rodden and Snyder Jr. and published in the American Political Science Review in 2008. There is no theoretical reason to believe that these Canadian and American results, as the results about salience-based heterogeneity, would not also be observed in other democracies.

The findings presented in this thesis are also consequential because of their potential impact on electoral strategy research. Indeed, the thesis validates some assumptions about public opinion underlying political marketing theory, which tends to be more in line with the behaviour of electoral practitioners. It validates the political marketing notion that political parties–contrary to what is expected of brokerage parties–take position on issues in order to attract voters (Gould, 1998; Paréand Berger, 2008). It also validates the assumption that modern political leaders are more likely to adapt their political offers to the demands and positions of potential party voters (Geer, 1996). And it gives credence to the idea that most campaign strategies are no longer built around a single campaign theme. Instead, they are increasingly based on targeting tactics aimed at specific segments of the voting public (Hillygus, 2007; Hillygus and Shields, 2008). Given that campaign strategies ultimately depend on the nature of the electorate, it is important for parties to identify which positional issues matter to whom, and when.

8.1 Macro- and Micro-Campaigns

Do the ﬁndings of this thesis regarding the relevance of positional issues and salience-based heterogeneity undermine brokerage politics in Canada? Not necessarily. Political parties’ perceived ability to deal with such consensual issues has repeatedly been shown to matter for vote choice. It would be naive to entirely contradict the basic assumptions of brokerage

153 politics. As repeated throughout the text, the findings of this thesis aim to add subtlety to existing approaches to voting and party behaviour, not undermine them. The goal is certainly not to tear down and replace well-established theories. To be clear, the existence of issue publics does not undermine the need for political parties to elaborate strategies based on valence issues or the state of the economy (Clarke, Kornberg and Scotto, 2008; Vavreck, 2009). Political parties must still elaborate such strategies. However, they can, at the same time, target precise voter segments, as Hillygus and Shields (2008) show. Macro-campaigns that focus on country-wide strategies and micro-campaigns that target specific issue publics can co-exist. Indeed, it is possible now more than ever for political parties to reach individual voters. An increasingly fragmented political space, splintered by various factors such as increasingly complex social, workplace and media environments, creates greater opportunities for political parties to communicate with specific groups. It is hard to believe that the advent of social media and other communication technologies has had no impact on electoral politics, especially given the big impact the invention public opinion polls had on politics in the 1940s (Geer, 1996; Jacobs and Shapiro, 1994). Communication changes increase the ability of parties to reach citizens and to develop more complex electoral strategies. Consequently, parties are no longer limited to nation- wide campaigning and positional issues have become fundamental elements in parties’ toolboxes.

The brokerage model was developed to account for the way that Canadian political parties focused on valence issues to avoid alienating segments of a highly fragmented electorate (Carty, 2013; Clarke et al., 1984). Discouraging potential voters from turning away from one’s party is still a good way to achieve electoral goals. However, parties can now also try to achieve their goals by elaborating messages that target precise voter segments. Because positional issues do not all have the same eﬀects on the same voters, parties

154 can target certain voters by priming certain positional issues. Such a strategy might be considered as belonging to a different type of brokerage politics, one that still implies valence issues at the macro-level, but uses positional issues at the micro-level to broker to disparate interests. Given that positional issues can be electorally profitable, analyzing them in a fine-grained manner represents a fruitful way to evaluate a party’s potential for growth. As such, positional issues are undoubtedly important elements in political parties’ strategic toolboxes.

The approach adopted by this thesis is subject to certain limits. Close inspection of the concept of heterogeneity raises some questions. For instance, how do we trace the contours of the heterogeneity of issue effects? The literature shows that divisional lines can be drawn along various cleavages such as geography (Henderson, 2004), partisanship (Ne- undorf, Stegmueller and Scotto, 2011), race (Peffley and Hurwitz, 2007), and personality traits (Gerber et al., 2010). Defining the concept of heterogeneity must be done with great care. For that reason, this thesis has restricted its analysis of heterogeneity to the heterogeneity of positional issue effects. However, the concept of heterogeneity can quickly turn into a threat to the parsimony of voting theories. It is true that public opinion has become so complex that more exploratory research on its intricacies is justified. But too deeply examining the causal mechanisms that underly heterogeneity might lead to more confusion instead of to a better understanding of voting behaviour.

This thesis argues that the concept of heterogeneity complements the framework of political marketing. However, the internal consistency of political marketing theory must be enhanced for this to be undeniably true. Political marketing theory must reconcile itself with the ﬁndings of mainstream voting theories. At this stage, political marketing has suﬀered from undertheorization. Undertheorization has rendered the precepts of

155 political marketing difficult to falsify and has affected the parsimony of the framework, thereby diminishing its potential for generalization. It has also led to a paradox in the political marketing model arising from its ambition to reconcile rational-choice and sociological models. Treating voters as unique and rational individuals while also arguing that they can be standardized into segments creates obvious operationalization problems. Ultimately, an amalgam between models that belong to different epistemological frameworks might result in the development of theoretical weaknesses instead of the creation of a new and more accurate model. That said, this thesis shows that political marketing can potentially become an integrative framework for studying modern campaign strategies.

8.2 Potential for Future Research

The evidence presented in this thesis emerges from data collected during election campaigns. Future research might consider reproducing these ﬁndings with data collected between elections. In electoral democracies, campaigns are prime moments during which parties attempt to convince citizens to vote for them. Because political strategies are most intensely employed during election periods, it is natural to look to campaigns when investigating the eﬀects of political strategies. The fact that most voting behaviour data are collected during elections likely reinforces a campaign-period research focus. However, parties do not only strategize during election campaigns. In fact, most party repositioning occurs between elections, when parties are elaborating their platforms for following elections. And voting intentions do not only shift during election campaigns. Indeed, Lazarsfeld, Berelson and Gaudet (1948) note in their study of the 1940 American election that the events occurring before the campaign changed “over twice as many votes as all the

156 events of the campaign” (Lazarsfeld, Berelson and Gaudet, 1948, 102). According to these authors, campaigns simply serve to activate existing latent political preferences.

This thesis has also considered elite- and salience-driven heterogeneity while leaving the door open to eventually testing other types of heterogeneity. It intends to pave the way for further investigation of the potential impact of issue attitudes. It also intends to pave the way for further investigation into the way that these attitudes are distributed in the electorate.

The cognition-based heterogeneity proposed by Bartle (2005) speaks to the distribution of attitudes, as it involves different ways of cognitively organizing political belief systems, but it is more complex and amorphous than the two types of heterogeneity explored in this thesis. A separate body of research can be dedicated to exploring the bases and implications of this third kind of heterogeneity. Under cognition-based heterogeneity, the assumption of the homogeneity of the population is far more relaxed. But the circumstances under which one expects to find cognition-based heterogeneity are more difficult to pin down and thus require further investigation. Cognition-based heterogeneity follows from other concepts in the literature that describe the ways that voters organize their beliefs about the political world: belief systems (Converse, 1964), conceptual maps (Axelrod, 1973), schemas (Conover and Feldman, 1984), frames (Minsky, 1975), or shared understandings (Goldberg, 2011). These concepts all endeavour to account for how beliefs are structured and to illustrate the associational nature of attitudes.

Different groups of people organize their political ideas in different ways. Belief systems influence how different stimuli prompt different behaviours or different attitudes in people. Assessing the different ways in which issue positions are organized can help us to better explain how interpretive frames are distributed in the population. It might be that

157 some issues, such as those related to the environment or immigration, are more ideologically versatile than other issues (see Inglehart, 1998). The fact that these issues are less well-defined by traditional left-right cleavages might allow them to fit into various political frames. The integration of environmental concern into the extreme-right discourse of Europe is a point in case (Olsen, 1999), as is the increase in liberally-framed criticisms of multiculturalism (Joppke, 2004). In Canada, we might expect to discover cognition- based heterogeneity on immigration issues, particularly in Quebec, where there has been significant debate about reasonable accommodation for religious minorities (see Gias- son, Brin and Sauvageau, 2010). The debate over reasonable accommodation revolves in part around two competing liberal values and interpretive frames: religious tolerance and equality between men and women. Liberal-minded voters make different arguments depending on the interpretive frame they use to reason about the debate. The consequences of such heterogeneity might be particularly relevant if one is interested in studying more abstract concepts such as ‘tolerance toward ethnic minorities’ or ‘xenophobia.’ To that end, cognition-based heterogeneity has the potential to uncover positional-issue effects confined to specific groups of voters.

Cognition-based heterogeneity implies a profound reconceptualization of the mass public as far more complex and fragmented than salience-based heterogeneity believes it to be. Bartle (2005) points out that no single equation can describe the complex processes underlying a type of heterogeneity defined by different reasoning processes. But there is support for the idea that “theory should be just as complicated as all our evidence suggest” (King, Keohane and Verba, 1994, 20). More effort is required to find methodologies better suited for this type of research object. For Lane (1962), contextual analyses that include in-depth interviews are necessary to understanding the variance in the political ideas of the public (Lane, 1962). For others, like Stimson (1975), exploratory factor

158 analyses might uncover attitude ‘structures’ beyond those accounted for by the standard elite-ideological dimension. New methodologies that have been developed to deal with the challenge of capturing heterogeneous attitudinal structures, such as Relational Class Analysis (Goldberg, 2011), also seem quite promising. The idea underlying these new methods is the simultaneous use of clustering and relational methodologies. Yet in the absence of a general theory of heterogeneity it will remain diﬃcult to study political behaviour in this way (Bartle, 2005, 657).

The rehabilitation of positional issues might also open up other research opportunities. More attention ought to be paid to contextual factors that favour positional issues mattering. Whether positional issues play different roles in federal, provincial, and municipal elections might be examined, for instance. Considering that party identification in Cana- dian municipal politics is weak, if not non-existent, it is likely that positional issues are therein important. Some argue that positional issues are indeed important in municipal politics: “Although suburban politics are routinely characterized as issueless, we find that issue positions are the strongest predictor of vote choice” (Oliver and Ha, 2007, 402). Obviously, such research is contingent on data collection at the municipal level.

This thesis demonstrates the importance of putting greater eﬀort into data collection. The assumed unimportance of positional issues leaves the research community, especially in Canada, with few issue items. The items that exist are often limited to public issues of the day and do not have consistent question wordings between survey waves. The use of multi-item issue scales would entail the inclusion in surveys of more questions tapping attitudes about the same issues. Similarly, more rigourous study of the heterogeneity of the electorate would require larger datasets that include various issue topics.

159 8.3 Normative Implications

Citizens voting according to their policy preferences is arguably good for democracy. Such a proposition is more encouraging for proponents of classical democratic theory than is a perspective that depicts citizens principally as partisan followers. Or brokerage politics that results in “elections [deciding] who shall govern, but not the substance of public policy” (Clarke et al., 1984, 172). However, there is doubt that fragmented political interests and micro-campaigning are good for democracy. Detectable heterogeneneity in the population might encourage political parties to segment the electorate and target specific groups while ignoring broader interests. From a strictly economic perspective, segmenting and targeting strategies respond more efficiently to voters’ needs and wants. Like commerce is about giving consumers what they want, democracy is about giving voters what they want. But politics differ from the commercial market in the sense that they are based on the common good. The clientelist aspect of political marketing precludes the idea of democratic deliberation and the search for consensus. This type of market-oriented politics might lead to the isolation of voters that have less strategic value to parties. It might be argued that majorities have always had more access to political parties’ ears and that this even defines democracy. Micro-campaigning tactics might have the effect of increasingly isolating voting segments that have no strategic value to political parties. If Canadian political parties were to engage in political marketing more intensively, it would be interesting to see whether the structural effect of party competition would lead to a greater fragmentation or a greater polarization of the party system.

160 Appendix: Figures

161 Figure 1: Eﬀect of issues on vote choice (Breakdown by issues)

(a) (b)

Economy Environment Foreign/US relations Economy Environment Foreign/US relations 10 10 10 ● ● ● 8 8 8 0.8 0.8 0.8 ●

6 6 6 ● ● 4 4 4 0.4 0.4 0.4

● ●

● 2 2 2 ● ● ● 0 0 0 0.0 0.0 0.0

Statistical significance (p−value) (Items) Scales Statistical significance (p−value) (Items) Scales Statistical significance (p−value) (Items) Scales (Items) Scales (Items) Scales (Items) Scales Effect on vote (|Beta coefficients|) on vote Effect (|Beta coefficients|) on vote Effect (|Beta coefficients|) on vote Effect

Law and order Minority issues Social programs Law and order Minority issues Social programs

● 10 10 10 8 8 8 0.8 0.8 0.8 6 6 6 4 4 4 162 0.4 0.4 0.4 ● ● ● ● ● 2 ● 2 2 ● 0 0 0 0.0 0.0 0.0 Statistical significance (p−value) Statistical significance (p−value) (Items) Scales Statistical significance (p−value) (Items) Scales (Items) Scales (Items) Scales (Items) Scales (Items) Scales Effect on vote (|Beta coefficients|) on vote Effect (|Beta coefficients|) on vote Effect (|Beta coefficients|) on vote Effect

Moral issues Women issues Moral issues Women issues

● ● ● ● 10 10 ● ● ● ●

● 8 8 0.8 ● 0.8 ● ● ● 6 6

● ● ● 4 4 0.4 0.4 ● ●

2 2 ●

● ● 0 0 0.0 0.0 Statistical significance (p−value) Statistical significance (p−value) (Items) Scales (Items) Scales (Items) Scales (Items) Scales Effect on vote (|Beta coefficients|) on vote Effect (|Beta coefficients|) on vote Effect

Source: Canadian Election Study (CES), 2004 and 2008. Method: Binary logistic regression (summary of 2808 models). Factor scaling. Dependent variables: Vote(Conservative), Vote(Liberal), Vote(NDP). Independent variables: Issue scales [included individually] (See Appendix for details.) Controls: 1) SES only (468 models); 2) (...) + Values (468 models); 3) (...) + Party identification (468 models); 4) (...) + Economic perception (468 models); 5) (...) + Leader evaluation (468 models); 6) (...) + Incumbent evaluation (468 models). Note: The three vertical lines in graphic (a) represent the statistical significance levels: from bottom to top, p < .001; p < .05; p < .1. Figure 2: Effect of issues, values and partisan identifications on vote choice (Breakdown by values and party identifications)

(a) (b)

Significance of issues and values, 2004 and 2008 Strength of issues, values and party identification, 2004 and 2008

● ● ● ● ● 10 1.0 ● ● ● ● ● ● ● ● ● ● ● ● 8 0.8 ● ● ● ● ● ● ● ● ● 6 0.6

● ●

● ● 4 0.4 ● ● ● ● ● Statistical significance (p−value) ● ● ● (|Beta coefficients|) on vote Effect ●

● 2

0.2 ● ● ● 0 0.0

(Items) Scales Religious Post−Mat Right Cynicism Liberal Id NDP Id Cons Id (Items) Scales Religious Post−Mat Right Cynicism Liberal Id NDP Id Cons Id

Source: Canadian Election Study (CES), 2004 and 2008. Method: Binary logistic regression (summary of 2808 models). Factor scaling. Dependent variables: Vote(Conservative), Vote(Liberal), Vote(NDP). Independent variables: (Items) (Issue items [included individually]); Scales (Issue scales [included individually]); Religious. (Religiosity); Post-mat. (Post-materialism [4-item index]); Right (Right-Left self-placement); Cynicism (Cynicism); Liberal Id (Liberal partisan); NDP Id (NDP partisan); Cons Id (Conservative partisan); Controls: 1) SES only (468 models); 2) (...) + Values (468 models); 3) (...) + Party identiﬁcation (468 models); 4) (...) + Economic perception (468 models); 5) (...) + Leader evaluation (468 models); 6) (...) + Incumbent evaluation (468 models). Note: The three vertical lines in graphic (a) represent the statistical signiﬁcance levels: from bottom to top, p < .001; p < .05; p < .1.

163 Figure 3: Evolution of voters’ most important issue

40 Economy Environment Foreign/US relations Law and order Minority issues Moral issues Other Social programs Women issues

20 % most important issue

19651968 1974 1979 1984 1988 1993 19972000 2004200620082011

Source: Canadian Election Study (CES), 1965-2011. Question: What is the most important issue to you personally in this federal election? Note: Most important issues relative to the Free-Trade Agreement debate with the United States are considered in the category Foreign/US relations. This issue was largely the most salient in the 1988 election.

164 Figure 4: The eﬀect of voters’ most important issue

(a) Average Eﬀect (b) Party

0.05 0.05

●

0.04 0.04 ●

0.03 0.03 ● ● ● ● ●

● 0.02 ● 0.02 ● ● ● ● ● ●

● ● ● Explained variance of issues Explained variance of issues Explained variance ● ● 0.01 0.01 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.00 ● ● 0.00

1965 1968 1974 1979 1984 1988 1993 1997 2000 200420062008 2011 1965 1968 1974 1979 1984 1988 1993 1997 2000 200420062008 2011

Source: Canadian Election Study (CES), 1965-2011. Method: Linear regression models are used in order to facilitate the direct comparisons between coefficients from different years. Question: What is the most important issue to you personally in this federal election? Note: The explained variance of issues is calculated from the difference in adjusted R-squared of fully-specified models including and excluding the issue variables. Solid line = Conservatives; Dashed line = Liberals; Dotted line = New Democrats.

165 Figure 5: Increase in Issue Impact? (1)

Economy

Conservatives Liberals New Democrats

0.25 ● 0.25 0.25

2008 2011 2008 2011 2008 2011

Environment

Conservatives Liberals New Democrats

0.25 0.25 0.25

● ●

● ● |Coefficient| |Coefficient| ● |Coefficient| 0.00 ● 0.00 0.00

2008 2011 2008 2011 2008 2011

Foreign/US relations

Conservatives Liberals New Democrats

0.25 0.25 0.25 ● ● ● ● ●

2008 2011 2008 2011 2008 2011

Law and order

Conservatives Liberals New Democrats

0.25 0.25 0.25

2008 2011 2008 2011 2008 2011

Source: Canadian Election Study (CES), 2004, 2008 and 2011. Note: The 2004 election is the base category.

166 Figure 6: Increase in Issue Impact? (2)

Minority issues

Conservatives Liberals New Democrats

0.25 0.25 0.25

●

2008 2011 2008 2011 2008 2011

Moral issues

Conservatives Liberals New Democrats

0.25 0.25 0.25

● |Coefficient| ● |Coefficient| ● |Coefficient| ● 0.00 ● 0.00 0.00 ●

2008 2011 2008 2011 2008 2011

Social programs

Conservatives Liberals New Democrats

0.25 0.25 0.25

2008 2011 2008 2011 2008 2011

Women issues

Conservatives Liberals New Democrats

0.25 0.25 0.25

●

● ●

2008 2011 2008 2011 2008 2011

Source: Canadian Election Study (CES), 2004, 2008 and 2011. Note: The 2004 election is the base category.

167 Figure 7: How Citizens Perceive Parties’ Stands on Minority Matters

0.6 0.6

0.4 0.4

0.2 0.2 Concern minorities about racial Concern minorities about racial 0.0 0.0 All Immigrants Visible minority All Immigrants Visible minority

(a) 1984 (b) 1993

Source: Canadian Election Study (CES), 1984 and 1993. Question wording: I’ll read a word or phrase, and I’d like you to tell me how well it ﬁts or describes each of the parties: Concerned with the interests of ethnic minority groups (1984) Does the [federal party] want to do much more, for racial minorities, somewhat more, about the same as now, somewhat less, or much less? (1993) Note: The bars represent, from left to right, Conservatives, Liberals and New Democrats. The term ‘Conservatives’ refers to the Progressive-Conservative party in 1984 and to the Progressive-Conservative party and the Reform Party in 1993.

168 Figure 8: Minorities’ Partisan Support Through Time (Three main parties)

Conservatives Liberals New Democrats

100 100 100

75 75 75

50 50 50

25 25 25 Percentage of vote Percentage of vote Percentage of vote Percentage

0 0 0

19651968 1974 1979 1984 1988 1993 19972000 2004 20082011 19651968 1974 1979 1984 1988 1993 19972000 2004 20082011 19651968 1974 1979 1984 1988 1993 19972000 2004 20082011 (a) Immigrants

Conservatives Liberals New Democrats

100 100 100

75 75 75

50 50 50

25 25 25 Percentage of vote Percentage of vote Percentage of vote Percentage

0 0 0

19651968 1974 1979 1984 1988 1993 19972000 2004 20082011 19651968 1974 1979 1984 1988 1993 19972000 2004 20082011 19651968 1974 1979 1984 1988 1993 19972000 2004 20082011 (b) Racial minority

Source: Canadian Election Study (CES), 1965 to 2011. Note: The black line represents, accordingly, (a) immigrant and (b) racial minorities. The dashed line represents all the other voters. Immigrant voters are those not born in Canada. Racial minorities are those respondents who indicate that their ancestors’ origins are non-European.

169 Figure 9: Eﬀect of Positional Issues on Issue publics in “very ethnic” targeted ridings 1 1 0.75 0.75 0.5 0.5 Pr(Voting Conservatives) Pr(Voting Conservatives) Pr(Voting 0.25 0.25 0 0

0 0.25 0.5 0.75 1 0 0.25 0.5 0.75 1 Degree of Proximity on Issues Degree of Proximity on Issues

(a) Issue Publics (b) Proximity

Source: Vote Compass, 2011 Canadian Federal Election. Note: The black line represents the “very ethnic” ridings targeted by the Conservatives in 2011: Elmwood- Transcona (MB); York Centre (ON); Richmond Hill (ON); Brampton-Springdale (ON); Eglinton-Lawrence (ON); Vancouver South (BC); Bramalea-Gore-Malton (ON); Burnaby-Douglas (BC); Newton-North Delta (BC); and Mount Royal (PQ). The grey lines represent the other ridings. The grey line represents all the other ridings.

170 Figure 10: East-Asian and Hispanic Canadians’ Ideological Positions and Resistance to Conservatives

0.4

0.3

0.2 density

0.1

0.0

Leftist Centrist Rightist Ideological Scale (a) Distribution (b) Vote by ideology

Source: Vote Compass, 2011 Canadian Federal Election. Note: The dashed and dotdashed lines represent, accordingly, respondents of (a) Hispanic and (b) East-Asian origins. The solid line represents all the other voters. The left-right ideological factor scale is made up of 23 items measuring positional issue attitudes on a ﬁve-point scale (see Table 22 for details). The probability of voting Conservative is based on respondents’ personal evaluation of their probability of voting for the party. The question was posed as “What is the probability that you will vote for the Conservative Party?”

171 Figure 11: Evolution of the eﬀect of being Jewish/Muslim on the vote for the Conserva- tives

(a) Jewish (b) Muslim Conservatives Conservatives 1.0 1.0

0.5 0.5

0.0 0.0 Pr(Vote) Pr(Vote)

−0.5 −0.5

−1.0 −1.0

1965 1968 1974 1979 1984 1988 1993 1997 2000 200420062008 2011 1965 1968 1974 1979 1984 1988 1993 1997 2000 200420062008 2011

Source: Canadian Election Study (CES), 1965-2011. Method: Ordinary least squares regression. Dependent variable: Vote(Conservative).

Figure 12: Evolution of the eﬀect of being Jewish/Muslim on the vote for the Liberals

(a) Jewish (b) Muslim Liberals Liberals 1.0 1.5

1.0

0.5

0.0 0.0 Pr(Vote) Pr(Vote)

−0.5

−1.0

−1.0 −1.5

1965 1968 1974 1979 1984 1988 1993 1997 2000 200420062008 2011 1965 1968 1974 1979 1984 1988 1993 1997 2000 200420062008 2011

Source: Canadian Election Study (CES), 1965-2011. Method: Ordinary least squares regression. Dependent variable: Vote(Liberal).

172 Figure 13: Evolution of the eﬀect of being Jewish/Muslim on the vote for the NDP

(a) Jewish (b) Muslim NDP NDP 1.0 1.0

0.5 0.5

0.0 0.0 Pr(Vote) Pr(Vote)

−0.5 −0.5

−1.0 −1.0

1965 1968 1974 1979 1984 1988 1993 1997 2000 200420062008 2011 1965 1968 1974 1979 1984 1988 1993 1997 2000 200420062008 2011

Source: Canadian Election Study (CES), 1965-2011. Method: Ordinary least squares regression. Dependent variable: Vote(NDP).

173 Appendix: Tables

Table 1: United States vs. Canada: The eﬀect of scales on explained variance

ANES 1992 ANES 1996 CES 2004 CES 2008 CES 2011 Baseline model (Pseudo-R2) 0.570 0.590 0.386 0.394 0.465

Scales (2)† + 0.090 + 0.070 + 0.013 + 0.004 + 0.004

Scales (8)† - - + 0.050 + 0.077 + 0.051

Source: American National Election Study (ANES), 1992 and 1996. Canadian Election Study (CES), 2004-2011. Method: Probit regression (Adjusted Pseudo-R2). Note: The “baseline” models include only Party ID and ideology scale as independent variables. † Ansolabehere et al. (2008) results are only based on two issue scales: Economy (13 items) and Moral (6 items). The authors also tried “adding a Foreign Policy Scale but it was never statistically or substantively signiﬁcant in 1992 or 1996” (Ansolabehere et al, 2008: 226). Contrary to these results, all eight positional issue scales are statistically or substantively signiﬁcant in Canada in at least one of the elections under study.

174 Table 2: Issue scales information (1)

Issue scales Issue items Factor loadings Economy When businesses make a lot of money, everyone 0.69 (2004) (Position: Free market) beneﬁts, including the poor. 0.62 (2008)

α = 0.59 (2004); 0.62 (2008) How much conﬁdence have in: Big business. 0.66 (2004) Eigenvalues† = 2.03 (2004); 2.11 (2008) 0.58 (2008)

Overall, free trade with the U.S. has been good for 0.64 (2004) the Canadian economy. 0.69 (2008)

International trade creates more jobs in Canada than 0.58 (2004) it destroys. 0.55 (2008)

175 And corporate taxes: should corporate taxes be 0.53 (2004) increased, decreased or kept about the same? 0.61 (2008)

If people can’t ﬁnd work in the region where they live, 0.31 (2004) they should move to where the jobs are. 0.50 (2008) Environment ... the Environment? Should the Federal government 0.79 (2004) (Position: Environmentalist) spend more, less, or about the same as now? 0.79 (2008)

α = 0.40 (2004); 0.41 (2008) Protecting the environment is more important than 0.79 (2004) Eigenvalues (1st factor) = 1.25 (2004); 1.26 (2008) creating jobs. 0.79 (2008)

Source: Canadian Election Study (CES), 2004 and 2008. † Eigenvalues for the ﬁrst factors. Table 3: Issue scales information (2)

Issue scales Issue items Factor loadings Foreign/US relations How do you feel about the United States? 0.71 (2004) (Position: More involvement, closer ties) 0.63 (2008)

α = 0.54 (2004); 0.51 (2008) Do you think Canada’s ties with the United States should 0.64 (2004) Eigenvalues = 1.74 (2004); 1.76 (2008) be much closer, somewhat closer, about the same as now, 0.71 (2008) somewhat more distant or much more distant?

... Defence? [Or ...Military] spending? Should the Federal 0.64 (2004) government spend more, less, or about the same as now? 0.60 (2008) 176 How much conﬁdence have in: The armed forces. 0.46 (2004) 0.72 (2008)

Canada should participate in peacekeeping operations 0.45 (2004) abroad even if it means putting the lives of Canadian 0.35 (2008) soldiers at risk . Law and order Do you favour or oppose the death penalty for people 0.75 (2004) (Position: Tough on crime) convicted of murder? 0.72 (2008)

α = 0.61 (2004); 0.63 (2008) What is the best way to deal with young oﬀenders 0.74 (2004) Eigenvalues = 1.68 (2004); 1.70 (2008) who commit violent crime: One, give them tougher 0.78 (2008) sentences; or Two, spend more on rehabilitating them?

We must crack down on crime, even if that means that 0.74 (2004) criminals lose their rights. 0.76 (2008)

Source: Canadian Election Study (CES), 2004 and 2008. † Eigenvalues for the ﬁrst factors. Table 4: Issue scales information (3)

Issue scales Issue items Factor loadings Minority issues Immigrants make an important contribution to this 0.70 (2004) (Position: More acceptance) country. 0.67 (2008)

α = 0.76 (2004); 0.76 (2008) Too many recent immigrants just don’t want to ﬁt into 0.67 (2004) Eigenvalues = 3.00 (2004); 2.95 (2008) Canadian society. [Reversed] 0.69 (2008)

We should look after Canadians born in this country 0.66 (2004) ﬁrst and others second. [Reverved] 0.69 (2008)

177 ... and racial minorities? Use any number from zero to 0.64 (2004) one hundred. 0.62 (2008)

Do you think Canada should admit: more immigrants, 0.62 (2004) fewer immigrants, or about the same as now? 0.63 (2008)

Which statement comes closest to your own view: 0.58 (2004) 1) If Aboriginal peoples tried harder, they could be as 0.51 (2008) well oﬀ as other Canadians: 2) Social and economic conditions make it almost impossible for most Aboriginal peoples to overcome poverty; 3) Not sure.

How do you feel about aboriginal peoples? 0.54 (2004) 0.52 (2008)

It is more diﬃcult for non-whites to be successful in 0.44 (2004) Canadian society than it is for whites. 0.49 (2008)

Source: Canadian Election Study (CES), 2004 and 2008. † Eigenvalues for the ﬁrst factors. Table 5: Issue scales information (4)

Issue scales Issue items Factor loadings Social programs The welfare state makes people less willing to look 0.62 (2004) (Position: No cut) after themselves. [Reversed] 0.65 (2008)

α = 0.56 (2004); 0.57 (2008) ... Welfare? Should the Federal government spend more, 0.61 (2004) Eigenvalues = 1.96 (2004); 1.98 (2008) less, or about the same as now? 0.65 (2008)

... Health Care? Should the Federal government spend 0.60 (2004) more, less, or about the same as now? 0.47 (2008)

How much do you think should be done to reduce 0.57 (2004) the gap between the rich and the poor in Canada: 0.64 (2008) much more, somewhat more, about the same as now,

178 somewhat less, or much less?

... Education? Should the Federal government spend 0.55 (2004) more, less, or about the same as now? 0.48 (2008)

Do you favour or oppose having some private hospitals 0.48 (2004) in Canada? 0.51 (2008)

Source: Canadian Election Study (CES), 2004 and 2008. † Eigenvalues for the ﬁrst factors. Table 6: Scaling information (5)

Issue scales Issue items Factor loadings Moral issues Do you favour or oppose same-sex marriage, or do you 0.87 (2004) (Position: Traditional) have no opinion on this? 0.88 (2008)

α = 0.80 (2004); 0.80 (2008) Gays and lesbians should be allowed to get married. 0.89 (2004) Eigenvalues = 2.57 (2004); 2.55 (2008) [Reversed] 0.91 (2008)

... and gays and lesbians? Use any number from zero 0.83 (2004) to one hundred. 0.79 (2008)

Do you think it should be: very easy for women to get 0.57 (2004) an abortion, quite easy, quite diﬃcult, or very diﬃcult? 0.57 (2008) Women’s issues The feminist movement: 1) Just tries to get equal 0.71 (2004) 179 (Position: Feminist) treatment for women; 2) Puts men down 0.73 (2008)

α = 0.62 (2004); 0.63 (2008) The feminist movement encourages women: 1) To be 0.71 (2004) Eigenvalues = 2.27 (2004); 2.26 (2008) independent and speak up for themselves; 2) To be 0.72 (2008) selﬁsh and think only of themselves

How much do you think should be done for women: 0.56 (2004) much more, somewhat more, about the same as now, 0.51 (2008) somewhat less, or much less?

The best way to protect women’s interests is to have 0.55 (2004) more women in Parliament 0.53 (2008)

... and feminists? Use any number from zero to one 0.54 (2004) hundred. 0.55 (2008)

Discrimination makes it extremely diﬃcult for women to 0.53 (2004) get jobs equal to their abilities. 0.51 (2008)

Society would be better oﬀ if more women stayed home 0.30 (2004) with their children. 0.33 (2008)

Source: Canadian Election Study (CES), 2004 and 2008. † Eigenvalues for the ﬁrst factors. Table 7: Evolution of the Eﬀect of the Economy Issue Position on Attitudes Toward Political Parties

Attitude toward political parties Conservatives Liberals New Democrats Economy (Issue position) 0.247∗∗∗ 0.128∗∗∗ −0.085∗∗∗ (0.027) (0.026) (0.026) Maritimes −0.009 0.025∗ 0.050∗∗∗ (0.014) (0.014) (0.014) West −0.007 −0.028∗∗∗ −0.023∗∗ (0.010) (0.010) (0.010) Quebec 0.012 −0.048∗∗∗ −0.019 (0.018) (0.018) (0.018) French −0.044∗∗ −0.004 0.020 (0.018) (0.018) (0.018) Other language 0.040∗∗ 0.011 −0.016 (0.016) (0.016) (0.016) Catholic 0.060∗∗∗ −0.007 −0.020∗ (0.012) (0.012) (0.012) Protestant 0.071∗∗∗ −0.002 −0.038∗∗∗ (0.012) (0.011) (0.011) Other religion 0.010 −0.006 −0.001 (0.020) (0.019) (0.020) Immigrant −0.036∗∗∗ −0.024∗ 0.012 (0.014) (0.014) (0.014) Woman −0.006 0.001 0.028∗∗∗ (0.008) (0.008) (0.008) Less than 34 y-o 0.022∗ 0.031∗∗ 0.039∗∗∗ (0.013) (0.012) (0.013) More than 55 y-o −0.003 −0.010 −0.014 (0.009) (0.009) (0.009) Less than high school −0.005 −0.038∗∗∗ −0.022 (0.014) (0.014) (0.014) University degree −0.029∗∗∗ 0.051∗∗∗ 0.044∗∗∗ (0.009) (0.009) (0.009) Urban −0.013 0.014 0.014 (0.010) (0.009) (0.009) Low income −0.021∗ −0.017 −0.022∗∗ (0.011) (0.011) (0.011) High income 0.013 0.006 −0.032∗∗∗ (0.010) (0.010) (0.010) Liberal partisan −0.063∗∗∗ 0.228∗∗∗ 0.004 (0.011) (0.011) (0.011) NDP partisan −0.114∗∗∗ 0.001 0.267∗∗∗ (0.015) (0.015) (0.014) Conservative partisan 0.256∗∗∗ −0.110∗∗∗ −0.099∗∗∗ (0.011) (0.011) (0.011) 2008 0.044∗∗ 0.078∗∗∗ 0.054∗∗∗ (0.020) (0.020) (0.020) 2011 −0.150∗∗∗ 0.012 0.135∗∗∗ (0.026) (0.025) (0.025) 2008 X Economy −0.016 −0.129∗∗∗ 0.009 (0.045) (0.044) (0.045) 2011 X Economy 0.230∗∗∗ −0.068 −0.129∗∗∗ (0.051) (0.050) (0.050) constant 0.328∗∗∗ 0.398∗∗∗ 0.472∗∗∗ (0.018) (0.018) (0.018) N 2, 963 2, 980 2, 911 R2 0.376 0.274 0.255 Adjusted R2 0.371 0.268 0.248 Source: Canadian Election Study (CES), 2004, 2008 and 2011 (Issue Scales). ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

180 Table 8: Evolution of the Eﬀect of the Environment Issue Position on Attitudes Toward Political Parties

Attitude toward political parties Conservatives Liberals New Democrats Environment (Issue position) −0.073∗∗ 0.050∗ 0.138∗∗∗ (0.029) (0.028) (0.028) Maritimes −0.005 0.018 0.028∗∗ (0.014) (0.013) (0.013) West −0.005 −0.030∗∗∗ −0.032∗∗∗ (0.010) (0.010) (0.010) Quebec 0.002 −0.059∗∗∗ −0.016 (0.017) (0.016) (0.017) French −0.016 0.008 0.015 (0.017) (0.016) (0.017) Other language 0.044∗∗∗ 0.016 −0.016 (0.015) (0.015) (0.015) Catholic 0.060∗∗∗ 0.009 −0.011 (0.012) (0.011) (0.012) Protestant 0.084∗∗∗ 0.005 −0.031∗∗∗ (0.011) (0.011) (0.011) Other religion 0.013 −0.011 −0.005 (0.019) (0.018) (0.018) Immigrant −0.021 −0.019 −0.002 (0.013) (0.013) (0.013) Woman −0.018∗∗ 0.001 0.034∗∗∗ (0.008) (0.008) (0.008) Less than 34 y-o 0.029∗∗ 0.037∗∗∗ 0.033∗∗∗ (0.013) (0.012) (0.012) More than 55 y-o −0.004 −0.006 −0.022∗∗ (0.009) (0.009) (0.009) Less than high school −0.002 −0.036∗∗∗ −0.019 (0.014) (0.013) (0.013) University degree −0.018∗∗ 0.047∗∗∗ 0.040∗∗∗ (0.009) (0.009) (0.009) Urban −0.012 0.016∗ 0.011 (0.009) (0.009) (0.009) Low income −0.023∗∗ −0.021∗∗ −0.016 (0.011) (0.010) (0.011) High income 0.027∗∗∗ 0.007 −0.036∗∗∗ (0.010) (0.009) (0.009) Liberal partisan −0.053∗∗∗ 0.238∗∗∗ 0.005 (0.011) (0.010) (0.010) NDP partisan −0.143∗∗∗ −0.001 0.271∗∗∗ (0.015) (0.014) (0.014) Conservative partisan 0.286∗∗∗ −0.093∗∗∗ −0.098∗∗∗ (0.011) (0.010) (0.011) 2008 0.031 −0.060∗ 0.095∗∗∗ (0.034) (0.032) (0.033) 2011 0.080∗∗∗ −0.030 0.035 (0.028) (0.027) (0.027) 2008 X Environment 0.003 0.134∗∗∗ −0.049 (0.051) (0.048) (0.049) 2011 X Environment −0.145∗∗∗ 0.041 0.049 (0.042) (0.040) (0.040) constant 0.448∗∗∗ 0.403∗∗∗ 0.356∗∗∗ (0.025) (0.023) (0.024) N 3, 255 3, 275 3, 200 R2 0.345 0.286 0.256 Adjusted R2 0.340 0.281 0.250 Source: Canadian Election Study (CES), 2004, 2008 and 2011 (Issue Scales). ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

181 Table 9: Evolution of the Eﬀect of the Foreign/US Relations Issue Position on Attitudes Toward Political Parties

Attitude toward political parties Conservatives Liberals New Democrats Foreign/US relations (Issue position) 0.394∗∗∗ 0.149∗∗∗ 0.059 (0.039) (0.038) (0.039) Maritimes −0.004 0.023∗ 0.038∗∗∗ (0.014) (0.013) (0.013) West 0.002 −0.032∗∗∗ −0.032∗∗∗ (0.010) (0.009) (0.010) Quebec 0.025 −0.052∗∗∗ −0.016 (0.017) (0.016) (0.017) French −0.0004 0.008 0.015 (0.017) (0.016) (0.017) Other language 0.060∗∗∗ 0.021 −0.018 (0.015) (0.015) (0.015) Catholic 0.044∗∗∗ 0.001 −0.013 (0.012) (0.011) (0.012) Protestant 0.062∗∗∗ 0.003 −0.033∗∗∗ (0.011) (0.011) (0.011) Other religion 0.007 −0.009 −0.005 (0.019) (0.018) (0.018) Immigrant −0.016 −0.020 −0.001 (0.013) (0.013) (0.013) Woman 0.001 0.006 0.034∗∗∗ (0.008) (0.008) (0.008) Less than 34 y-o 0.037∗∗∗ 0.036∗∗∗ 0.034∗∗∗ (0.012) (0.012) (0.012) More than 55 y-o −0.005 −0.009 −0.018∗∗ (0.009) (0.009) (0.009) Less than high school 0.004 −0.036∗∗∗ −0.031∗∗ (0.013) (0.013) (0.013) University degree −0.013 0.055∗∗∗ 0.044∗∗∗ (0.009) (0.008) (0.009) Urban −0.013 0.020∗∗ 0.016∗ (0.009) (0.009) (0.009) Low income −0.023∗∗ −0.016 −0.018∗ (0.011) (0.010) (0.011) High income 0.018∗ 0.009 −0.036∗∗∗ (0.009) (0.009) (0.009) Liberal partisan −0.063∗∗∗ 0.232∗∗∗ 0.012 (0.010) (0.010) (0.010) NDP partisan −0.137∗∗∗ −0.001 0.276∗∗∗ (0.014) (0.014) (0.014) Conservative partisan 0.260∗∗∗ −0.103∗∗∗ −0.107∗∗∗ (0.011) (0.011) (0.011) 2008 0.060 0.148∗∗∗ 0.169∗∗∗ (0.037) (0.036) (0.037) 2011 −0.117∗∗∗ 0.108∗∗∗ 0.167∗∗∗ (0.033) (0.033) (0.033) 2008 X Foreign/US relations −0.035 −0.218∗∗∗ −0.195∗∗∗ (0.063) (0.062) (0.063) 2011 X Foreign/US relations 0.180∗∗∗ −0.203∗∗∗ −0.186∗∗∗ (0.056) (0.055) (0.056) constant 0.176∗∗∗ 0.347∗∗∗ 0.409∗∗∗ (0.027) (0.026) (0.027) N 3, 290 3, 305 3, 234 R2 0.383 0.279 0.248 Adjusted R2 0.379 0.274 0.242 Source: Canadian Election Study (CES), 2004, 2008 and 2011 (Issue Scales). ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

182 Table 10: Evolution of the Eﬀect of Law and Order Issue Position on Attitudes Toward Political Parties

Attitude toward political parties Conservatives Liberals New Democrats Law and order (Issue position) 0.087∗∗∗ −0.055∗∗∗ −0.083∗∗∗ (0.020) (0.018) (0.019) Maritimes −0.001 0.017 0.030∗∗ (0.014) (0.013) (0.013) West −0.004 −0.029∗∗∗ −0.032∗∗∗ (0.010) (0.009) (0.010) Quebec 0.008 −0.053∗∗∗ −0.012 (0.017) (0.016) (0.017) French −0.028 0.002 0.011 (0.017) (0.016) (0.017) Other language 0.037∗∗ 0.021 −0.015 (0.016) (0.015) (0.015) Catholic 0.061∗∗∗ 0.012 −0.009 (0.012) (0.011) (0.011) Protestant 0.078∗∗∗ 0.011 −0.029∗∗∗ (0.011) (0.011) (0.011) Other religion 0.010 −0.003 −0.007 (0.019) (0.018) (0.018) Immigrant −0.024∗ −0.019 −0.0003 (0.014) (0.013) (0.013) Woman −0.017∗∗ −0.001 0.032∗∗∗ (0.008) (0.008) (0.008) Less than 34 y-o 0.030∗∗ 0.035∗∗∗ 0.030∗∗ (0.013) (0.012) (0.012) More than 55 y-o −0.003 −0.012 −0.029∗∗∗ (0.009) (0.009) (0.009) Less than high school −0.005 −0.035∗∗∗ −0.024∗ (0.014) (0.013) (0.013) University degree −0.002 0.037∗∗∗ 0.028∗∗∗ (0.009) (0.009) (0.009) Urban −0.008 0.015∗ 0.006 (0.009) (0.009) (0.009) Low income −0.024∗∗ −0.015 −0.020∗ (0.011) (0.010) (0.011) High income 0.025∗∗∗ 0.003 −0.035∗∗∗ (0.010) (0.009) (0.009) Liberal partisan −0.044∗∗∗ 0.235∗∗∗ 0.002 (0.010) (0.010) (0.010) NDP partisan −0.137∗∗∗ 0.001 0.273∗∗∗ (0.014) (0.014) (0.014) Conservative partisan 0.287∗∗∗ −0.080∗∗∗ −0.098∗∗∗ (0.011) (0.010) (0.011) 2008 0.032 0.063∗∗∗ 0.061∗∗∗ (0.020) (0.019) (0.019) 2011 −0.052∗∗∗ 0.048∗∗∗ 0.093∗∗∗ (0.018) (0.016) (0.017) 2008 X Law and order 0.013 −0.070∗∗ 0.009 (0.032) (0.030) (0.030) 2011 X Law and order 0.067∗∗ −0.108∗∗∗ −0.055∗∗ (0.028) (0.026) (0.026) constant 0.349∗∗∗ 0.462∗∗∗ 0.497∗∗∗ (0.019) (0.018) (0.018) N 3, 308 3, 328 3, 250 R2 0.346 0.296 0.261 Adjusted R2 0.341 0.291 0.255 Source: Canadian Election Study (CES), 2004, 2008 and 2011 (Issue Scales). ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

183 Table 11: Evolution of the Eﬀect of Minority Issues Position on Attitudes Toward Political Parties

Attitude toward political parties Conservatives Liberals New Democrats Minority issues (Issue position) −0.078∗∗ 0.170∗∗∗ 0.256∗∗∗ (0.037) (0.035) (0.035) Maritimes −0.009 0.021 0.035∗∗ (0.015) (0.014) (0.014) West −0.005 −0.029∗∗∗ −0.027∗∗∗ (0.010) (0.010) (0.010) Quebec −0.003 −0.052∗∗∗ −0.012 (0.018) (0.017) (0.017) French −0.017 0.011 0.020 (0.018) (0.017) (0.018) Other language 0.049∗∗∗ 0.016 −0.013 (0.016) (0.015) (0.016) Catholic 0.059∗∗∗ 0.007 −0.011 (0.013) (0.012) (0.012) Protestant 0.083∗∗∗ 0.011 −0.029∗∗ (0.012) (0.011) (0.011) Other religion 0.015 −0.001 −0.012 (0.020) (0.019) (0.019) Immigrant −0.020 −0.024∗ −0.006 (0.014) (0.013) (0.013) Woman −0.021∗∗ −0.001 0.030∗∗∗ (0.008) (0.008) (0.008) Less than 34 y-o 0.025∗ 0.037∗∗∗ 0.036∗∗∗ (0.013) (0.012) (0.013) More than 55 y-o −0.005 −0.015 −0.028∗∗∗ (0.010) (0.009) (0.009) Less than high school 0.001 −0.027∗∗ −0.015 (0.014) (0.014) (0.014) University degree −0.009 0.033∗∗∗ 0.022∗∗ (0.010) (0.009) (0.009) Urban −0.011 0.016∗ 0.004 (0.010) (0.009) (0.009) Low income −0.028∗∗ −0.013 −0.017 (0.012) (0.011) (0.011) High income 0.024∗∗ 0.010 −0.038∗∗∗ (0.010) (0.009) (0.009) Liberal partisan −0.048∗∗∗ 0.232∗∗∗ −0.002 (0.011) (0.010) (0.010) NDP partisan −0.132∗∗∗ −0.009 0.264∗∗∗ (0.015) (0.014) (0.014) Conservative partisan 0.294∗∗∗ −0.085∗∗∗ −0.099∗∗∗ (0.012) (0.011) (0.011) 2008 0.081∗∗ 0.028 0.098∗∗∗ (0.039) (0.036) (0.037) 2011 0.080∗∗ −0.061∗ 0.056∗ (0.033) (0.031) (0.032) 2008 X Minority issues −0.083 −0.010 −0.063 (0.063) (0.060) (0.060) 2011 X Minority issues −0.155∗∗∗ 0.094∗ 0.010 (0.055) (0.052) (0.052) constant 0.444∗∗∗ 0.340∗∗∗ 0.311∗∗∗ (0.026) (0.024) (0.024) N 3, 043 3, 062 2, 989 R2 0.341 0.288 0.267 Adjusted R2 0.335 0.282 0.261 Source: Canadian Election Study (CES), 2004, 2008 and 2011 (Issue Scales). ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

184 Table 12: Evolution of the Eﬀect of Moral Issues Position on Attitudes Toward Political Parties

Attitude toward political parties Conservatives Liberals New Democrats Moral issues (Issue position) −0.301∗∗∗ −0.033 0.200∗∗∗ (0.036) (0.034) (0.034) Maritimes 0.006 0.022 0.028∗∗ (0.014) (0.014) (0.013) West −0.017∗ −0.026∗∗∗ −0.021∗∗ (0.010) (0.010) (0.010) Quebec −0.015 −0.050∗∗∗ −0.005 (0.017) (0.017) (0.017) French −0.022 0.002 0.016 (0.018) (0.017) (0.017) Other language 0.038∗∗ 0.011 −0.015 (0.016) (0.015) (0.015) Catholic 0.079∗∗∗ 0.002 −0.022∗ (0.012) (0.012) (0.012) Protestant 0.085∗∗∗ 0.002 −0.037∗∗∗ (0.011) (0.011) (0.011) Other religion 0.027 −0.007 −0.012 (0.019) (0.019) (0.019) Immigrant −0.026∗ −0.022∗ −0.004 (0.014) (0.013) (0.013) Woman −0.006 0.002 0.019∗∗ (0.008) (0.008) (0.008) Less than 34 y-o 0.023∗ 0.037∗∗∗ 0.037∗∗∗ (0.013) (0.012) (0.013) More than 55 y-o 0.003 −0.006 −0.026∗∗∗ (0.009) (0.009) (0.009) Less than high school 0.004 −0.035∗∗∗ −0.020 (0.014) (0.013) (0.013) University degree −0.015∗ 0.055∗∗∗ 0.040∗∗∗ (0.009) (0.009) (0.009) Urban −0.006 0.016∗ 0.015∗ (0.009) (0.009) (0.009) Low income −0.016 −0.013 −0.025∗∗ (0.011) (0.011) (0.011) High income 0.010 0.008 −0.028∗∗∗ (0.010) (0.009) (0.009) Liberal partisan −0.056∗∗∗ 0.238∗∗∗ 0.003 (0.010) (0.010) (0.010) NDP partisan −0.118∗∗∗ −0.004 0.252∗∗∗ (0.015) (0.014) (0.014) Conservative partisan 0.260∗∗∗ −0.095∗∗∗ −0.085∗∗∗ (0.011) (0.011) (0.011) 2008 0.099∗∗ −0.058 0.042 (0.042) (0.041) (0.041) 2011 0.034 −0.068∗∗ 0.012 (0.035) (0.034) (0.034) 2008 X Moral issues −0.092 0.129∗∗ 0.032 (0.060) (0.059) (0.058) 2011 X Moral issues −0.056 0.101∗∗ 0.078 (0.050) (0.049) (0.048) constant 0.595∗∗∗ 0.450∗∗∗ 0.315∗∗∗ (0.028) (0.027) (0.027) N 3, 150 3, 164 3, 097 R2 0.372 0.275 0.272 Adjusted R2 0.367 0.269 0.266 Source: Canadian Election Study (CES), 2004, 2008 and 2011 (Issue Scales). ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

185 Table 13: Evolution of the Eﬀect of Social Programs Issue Position on Attitudes Toward Political Parties

Attitude toward political parties Conservatives Liberals New Democrats Social programs (Issue position) 0.083∗∗∗ −0.037∗∗ −0.136∗∗∗ (0.018) (0.017) (0.017) Maritimes −0.005 0.036∗∗∗ 0.018∗ (0.011) (0.011) (0.011) West 0.002 −0.024∗∗∗ −0.029∗∗∗ (0.008) (0.008) (0.008) Quebec −0.008 −0.015 −0.018 (0.013) (0.012) (0.012) French −0.028∗∗ −0.022∗ −0.0002 (0.013) (0.012) (0.012) Other language 0.039∗∗∗ 0.019 −0.018 (0.013) (0.012) (0.012) Catholic 0.045∗∗∗ 0.005 0.003 (0.010) (0.009) (0.009) Protestant 0.061∗∗∗ 0.007 −0.010 (0.009) (0.009) (0.009) Other religion 0.003 0.009 −0.007 (0.015) (0.014) (0.014) Immigrant −0.031∗∗∗ 0.003 0.018∗ (0.011) (0.011) (0.011) Woman −0.008 −0.003 0.020∗∗∗ (0.007) (0.006) (0.006) Less than 34 y-o 0.018∗∗ 0.023∗∗∗ 0.039∗∗∗ (0.009) (0.008) (0.008) More than 55 y-o −0.009 0.005 −0.011 (0.008) (0.007) (0.007) Less than high school −0.023∗∗ −0.031∗∗∗ −0.012 (0.011) (0.010) (0.010) University degree −0.014∗∗ 0.055∗∗∗ 0.038∗∗∗ (0.007) (0.007) (0.007) Urban −0.007 0.015∗∗ 0.009 (0.007) (0.007) (0.007) Low income −0.026∗∗∗ −0.008 −0.011 (0.009) (0.008) (0.008) High income 0.023∗∗∗ 0.007 −0.030∗∗∗ (0.008) (0.007) (0.007) Liberal partisan −0.050∗∗∗ 0.225∗∗∗ −0.012 (0.008) (0.008) (0.008) NDP partisan −0.142∗∗∗ −0.013 0.256∗∗∗ (0.012) (0.011) (0.011) Conservative partisan 0.274∗∗∗ −0.079∗∗∗ −0.089∗∗∗ (0.009) (0.009) (0.009) 2008 0.029∗∗ 0.053∗∗∗ 0.061∗∗∗ (0.015) (0.014) (0.014) 2011 0.001 0.014 0.057∗∗∗ (0.012) (0.011) (0.011) 2008 X Social programs 0.025 −0.027 0.003 (0.029) (0.028) (0.028) 2011 X Social programs 0.003 −0.068∗∗∗ −0.023 (0.025) (0.023) (0.023) constant 0.387∗∗∗ 0.428∗∗∗ 0.492∗∗∗ (0.014) (0.013) (0.013) N 5, 235 5, 289 5, 149 R2 0.308 0.252 0.255 Adjusted R2 0.304 0.248 0.251 Source: Canadian Election Study (CES), 2004, 2008 and 2011 (Issue Scales). ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

186 Table 14: Evolution of the Eﬀect of Women Issues Position on Attitudes Toward Political Parties

Attitude toward political parties Conservatives Liberals New Democrats Women issues (Issue position) −0.107∗∗∗ 0.067∗ 0.216∗∗∗ (0.037) (0.035) (0.035) Maritimes 0.007 0.017 0.031∗∗ (0.014) (0.013) (0.013) West 0.004 −0.028∗∗∗ −0.030∗∗∗ (0.010) (0.010) (0.010) Quebec −0.003 −0.058∗∗∗ −0.012 (0.018) (0.017) (0.017) French −0.015 0.007 0.014 (0.018) (0.017) (0.017) Other language 0.040∗∗ 0.013 −0.019 (0.016) (0.015) (0.015) Catholic 0.068∗∗∗ 0.006 −0.012 (0.012) (0.012) (0.011) Protestant 0.080∗∗∗ 0.007 −0.024∗∗ (0.012) (0.011) (0.011) Other religion 0.009 −0.007 −0.003 (0.020) (0.019) (0.019) Immigrant −0.019 −0.016 0.010 (0.014) (0.013) (0.013) Woman −0.011 −0.004 0.014∗ (0.008) (0.008) (0.008) Less than 34 y-o 0.024∗ 0.038∗∗∗ 0.042∗∗∗ (0.013) (0.012) (0.012) More than 55 y-o 0.003 −0.008 −0.021∗∗ (0.009) (0.009) (0.009) Less than high school 0.00004 −0.037∗∗∗ −0.035∗∗∗ (0.014) (0.013) (0.013) University degree −0.020∗∗ 0.055∗∗∗ 0.038∗∗∗ (0.009) (0.009) (0.009) Urban −0.004 0.014 0.013 (0.009) (0.009) (0.009) Low income −0.023∗∗ −0.010 −0.015 (0.011) (0.011) (0.011) High income 0.025∗∗∗ 0.012 −0.033∗∗∗ (0.010) (0.009) (0.009) Liberal partisan −0.056∗∗∗ 0.232∗∗∗ 0.002 (0.011) (0.010) (0.010) NDP partisan −0.136∗∗∗ −0.00003 0.263∗∗∗ (0.015) (0.014) (0.014) Conservative partisan 0.284∗∗∗ −0.095∗∗∗ −0.099∗∗∗ (0.011) (0.011) (0.011) 2008 0.083∗ 0.014 0.057 (0.045) (0.043) (0.043) 2011 0.082∗∗ −0.026 0.022 (0.040) (0.038) (0.038) 2008 X Women issues −0.080 0.017 0.007 (0.067) (0.063) (0.063) 2011 X Women issues −0.149∗∗ 0.041 0.080 (0.060) (0.057) (0.057) constant 0.461∗∗∗ 0.388∗∗∗ 0.307∗∗∗ (0.029) (0.027) (0.027) N 3, 176 3, 192 3, 121 R2 0.338 0.272 0.273 Adjusted R2 0.333 0.266 0.267 Source: Canadian Election Study (CES), 2004, 2008 and 2011 (Issue Scales). ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

187 Table 15: Accessibility and Attitude Stability on the Same-Sex Marriage Issue

Attitude Stability Response Time −0.610∗∗∗ (0.098) University degree 0.083∗∗∗ (0.022) Less than high school −0.011 (0.027) Interested 0.222∗∗∗ (0.039) Knowledgeable 0.138∗∗∗ (0.029) constant 0.105∗∗∗ (0.029) N 1, 820 R2 0.085 Adjusted R2 0.083 Source: Canadian Election Study (CES), 2004, 2008 and 2011. ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

188 Table 16: The Eﬀect of Issue Salience: Full Regression Results

Moral issues Environment Interactions Position X Salience 0.39*** 0.19** (0.08) (0.07) Focal variables Position 0.38*** 0.21*** (0.02) (0.03) Salience -0.12 0.33*** (0.13) (0.06) Controls Age under 35 0.12* 0.22** (0.05) (0.07) Age 55+ 0.02 -0.14* (0.05) (0.06) High income 0.15*** -0.10 (0.04) (0.06) Low income -0.10* 0.05 (0.05) (0.06) University -0.15*** 0.18** (0.04) (0.06) Less than high school -0.07 -0.12 (0.08) (0.08) Interested 0.48*** 0.26 (0.12) (0.14) Knowledgeable 0.05 0.14 (0.06) (0.09) Adjusted R2 0.15 0.15 N 2376 1564 Source: Canadian Election Study (CES), 2004 and 2008 (Pooled). Method: Least Squares Regression. Log odds (Robust standard errors). Dependent variables: Attitude(Conservative, Green). Factor scales. *p < .10. **p < .05. ***p < .01.

189 Table 17: The Eﬀect of Immigrantness and Ethnicity on the Liberal Vote

1968-1979 1984-2000 2004-2006 2008-2011

Immigrant 0.10 — 0.08 0.16∗ — 0.04 0.07 — -0.08 0.02 — -0.04 Visible ethnicity — 0.43† 0.41† — 0.86∗∗∗ 0.84∗∗∗ — 0.68∗∗∗ 0.70∗∗∗ — 0.37∗ 0.39∗ Controls Quebec 0.17† 0.16 0.16 -0.46∗∗∗ -0.47∗∗∗ -0.47∗∗∗ -0.17 -0.14 -0.16 -0.65∗∗∗ -0.63∗∗∗ -0.61∗∗∗ West -0.55∗∗∗ -0.56∗∗∗ -0.56∗∗∗ -0.86∗∗∗ -0.85∗∗∗ -0.85∗∗∗ -0.83∗∗∗ -0.82∗∗∗ -0.81∗∗∗ -0.96∗∗∗ -0.95∗∗∗ -0.95∗∗∗ Maritimes 0.11 0.08 0.09 -0.09 -0.07 -0.07 0.31∗ 0.28 0.28 0.69∗∗∗ 0.69∗∗∗ 0.70∗∗∗ Woman 0.17∗∗∗ 0.17∗∗∗ 0.17∗∗∗ 0.13∗∗ 0.15∗∗ 0.14∗∗ 0.11 0.10 0.10 0.00 0.00 0.00 French language -0.21∗ -0.21∗ -0.20† -0.40∗∗∗ -0.41∗∗∗ -0.41∗∗∗ -0.76∗∗∗ -0.82∗∗∗ -0.73∗∗∗ -0.19 -0.22 -0.20 Other language 0.12∗ 0.13∗ 0.14∗ 0.26∗∗ 0.22∗∗ 0.21∗∗ 0.29∗ 0.22 0.26 0.03 -0.07 -0.02 Catholic 0.72∗∗∗ 0.72∗∗∗ 0.73∗∗∗ 0.51∗∗∗ 0.53∗∗∗ 0.53∗∗∗ 0.48∗∗∗ 0.48∗∗∗ 0.48∗∗∗ 0.00 -0.01 -0.02 Protestant -0.17 -0.17 -0.16 -0.18∗ -0.16∗ -0.16∗ -0.04 -0.00 -0.01 -0.30∗ -0.30∗ -0.30∗ Other religion 0.38∗ 0.41∗ 0.41∗ 0.28∗ 0.11 0.12 0.89∗∗∗ 0.80∗∗∗ 0.81∗∗∗ 0.14 0.05 0.03 Less than 34 years-old -0.20∗∗∗ -0.21∗∗∗ -0.20∗∗∗ -0.09 -0.08 -0.08 -0.14 -0.19 -0.19† -0.19 -0.17 -0.20 More than 55 years-old -0.04 -0.04 -0.04 0.25∗∗∗ 0.29∗∗∗ 0.28∗∗∗ 0.32∗∗∗ 0.32∗∗∗ 0.33∗∗∗ 0.35∗∗∗ 0.38∗∗∗ 0.36∗∗∗ 190 Less than high school 0.04 0.04 0.05 -0.06 -0.08 -0.08 -0.10 -0.10 -0.14 -0.13 -0.13 -0.18 University Degree 0.01 -0.01 0.01 0.08 0.06 0.06 0.17† 0.18∗ 0.17† 0.42∗∗∗ 0.40∗∗∗ 0.42∗∗∗ Low Income -0.15∗ -0.13∗ -0.13∗ 0.08 0.05 0.06 0.28∗∗ 0.27∗∗ 0.28∗∗ 0.06 0.07 High Income 0.24∗∗ 0.24∗∗ 0.24∗∗ 0.10† 0.10 0.10 0.42∗∗∗ 0.42∗∗∗ 0.42∗∗∗ 0.28∗∗ 0.31∗∗ 0.29∗∗ Urban 0.33∗∗∗ 0.34∗∗∗ 0.33∗∗∗ 0.04 0.03 0.03 0.09 0.05 0.04 0.46∗∗∗ 0.44∗∗∗ 0.44∗∗∗ (Intercept) -0.35∗ -0.33∗ -0.37∗ -0.70∗∗∗ -0.72∗∗∗ -0.73∗∗∗ -1.04∗∗∗ -1.05∗∗∗ -1.05∗∗∗ -1.56∗∗∗ -1.59∗∗∗ -1.56∗∗∗ N 7104 7017 6999 10124 9901 9876 3131 3091 3044 3615 3661 3581 AIC 9329.7 9212 9190.4 12070 11740 11715 3704.5 3627.5 3590.9 3549.6 3564.1 3513.6 Source: Canadian Election Study, 1965-2011. † signiﬁcant at p < .10; ∗p < .05; ∗∗p < .01; ∗∗∗p < .001 Table 18: The Eﬀect of Immigrantness and Ethnicity on the Conservative Vote

1968-1979 1984-2000 2004-2006 2008-2011

Immigrant -0.21∗∗ — -0.19∗ -0.10 — -0.04 -0.28∗ — -0.11 -0.13 — -0.08 Visible ethnicity — -0.56† -0.52 — -0.42∗∗∗ -0.41∗∗∗ — -0.80∗∗∗ -0.78∗∗∗ — -0.43∗∗ -0.42∗ Controls Quebec -0.39∗∗∗ -0.39∗∗ -0.38∗∗ -0.14† -0.17∗ -0.16∗ -0.84∗∗∗ -0.87∗∗∗ -0.87∗∗∗ -0.77∗∗∗ -0.79∗∗∗ -0.79∗∗∗ West 0.23∗∗∗ 0.23∗∗∗ 0.24∗∗∗ 0.46∗∗∗ 0.45∗∗∗ 0.45∗∗∗ 0.58∗∗∗ 0.60∗∗∗ 0.59∗∗∗ 0.39∗∗∗ 0.39∗∗∗ 0.39∗∗∗ Maritimes 0.44∗∗∗ 0.49∗∗∗ 0.46∗∗∗ 0.08 0.04 0.04 -0.24 -0.23 -0.24 -0.47∗∗∗ -0.49∗∗∗ -0.50∗∗∗ Woman -0.04 -0.02 -0.03 -0.26∗∗∗ -0.27∗∗∗ -0.27∗∗∗ -0.28∗∗∗ -0.26∗∗ -0.27∗∗ -0.22∗∗ -0.22∗∗ -0.22∗∗ French language -0.11 -0.08 -0.12 -0.08 -0.10 -0.09 -0.16 -0.11 -0.14 -0.65∗∗∗ -0.64∗∗∗ -0.65∗∗∗ Other language 0.05 0.05 0.05 -0.12 -0.10 -0.09 0.22 0.19 0.25 0.19 0.18 0.24† Catholic -0.28 -0.29† -0.30† -0.04 -0.06 -0.07 0.33∗∗ 0.37∗∗ 0.37∗∗ 0.79∗∗∗ 0.79∗∗∗ 0.81∗∗∗ Protestant 0.66∗∗∗ 0.66∗∗∗ 0.64∗∗∗ 0.58∗∗∗ 0.57∗∗∗ 0.57∗∗∗ 0.83∗∗∗ 0.84∗∗∗ 0.85∗∗∗ 1.01∗∗∗ 1.01∗∗∗ 1.02∗∗∗ Other religion 0.03 0.01 0.05 0.00 0.06 0.06 0.04 0.17 0.17 0.58∗∗∗ 0.72∗∗∗ 0.70∗∗∗ Less than 34 years-old -0.04 -0.03 -0.01 -0.08 -0.09 -0.09† -0.17 -0.15 -0.14 -0.05 -0.01 -0.02 More than 55 years-old 0.26∗∗∗ 0.25∗∗∗ 0.26∗∗∗ -0.05 -0.06 -0.05 0.03 0.02 0.02 -0.01 -0.02 -0.02 191 Less than high school -0.19∗∗ -0.19∗∗ -0.20∗∗ 0.08 0.07 0.06 0.24∗ 0.23 0.25 0.34∗∗ 0.37∗∗ 0.35∗∗ University Degree -0.04 -0.02 -0.03 -0.24∗∗∗ -0.23∗∗∗ -0.23∗∗∗ -0.36∗∗∗ -0.36∗∗∗ -0.36∗∗∗ -0.40∗∗∗ -0.39∗∗∗ -0.40∗∗∗ Low Income 0.20∗∗ 0.19∗∗ 0.20∗∗ -0.22∗∗∗ -0.20∗∗∗ -0.20∗∗∗ -0.20∗ -0.16 -0.19† -0.30∗ -0.31∗∗ -0.31∗∗ High Income 0.19∗ 0.20∗ 0.19∗ 0.28∗∗∗ 0.28∗∗∗ 0.28∗∗∗ -0.03 0.00 -0.01 0.08 0.05 0.07 Urban -0.29∗∗∗ -0.31∗∗∗ -0.30∗∗∗ 0.00 0.00 0.00 -0.17† -0.12 -0.13 -0.23∗∗ -0.20∗ -0.21∗ (Intercept) -1.05∗∗∗ -1.11∗∗∗ -1.03∗∗∗ -0.49∗∗∗ -0.47∗∗∗ -0.47∗∗∗ -0.75∗∗∗ -0.82∗∗∗ -0.79∗∗∗ -0.59∗∗∗ -0.59∗∗∗ -0.58∗∗∗ N 7104 7017 6999 10124 9901 9876 3131 3091 3044 3615 3661 3581 AIC 7950.6 7868.7 7840.1 13058 12695 12675 3679.7 3612.7 3567.5 4389.3 4413.2 4340.8 Source: Canadian Election Study, 1965-2011. † signiﬁcant at p < .10; ∗p < .05; ∗∗p < .01; ∗∗∗p < .001 Table 19: The Eﬀect of Immigrantness and Ethnicity on the NDP Vote

1968-1979 1984-2000 2004-2006 2008-2011

Immigrant -0.02 — -0.02 0.02 — 0.06 0.55∗∗∗ — 0.53∗∗ 0.17 — 0.11 Visible ethnicity — -0.56† -0.01 — -0.37∗ -0.39∗ — 0.03 -0.09 — 0.37∗ 0.38∗ Controls Quebec -0.90∗∗∗ -0.88∗∗∗ -0.88∗∗∗ -0.65∗∗∗ -0.64∗∗∗ -0.63∗∗∗ -0.93∗∗∗ -0.94∗∗∗ -0.90∗∗∗ 0.49∗∗ 0.47∗∗ 0.50∗∗ West 0.13 0.12 0.12 0.32∗∗∗ 0.32∗∗∗ 0.32∗∗∗ 0.20† 0.16 0.17 0.31∗∗ 0.32∗∗ 0.32∗∗ Maritimes -1.04∗∗∗ -1.04∗∗∗ -1.05∗∗∗ 0.13 0.14 0.15 0.12 0.14 0.17 -0.01 0.04 0.02 Woman -0.32∗∗∗ -0.32∗∗∗ -0.33∗∗∗ 0.18∗∗ 0.17∗∗ 0.17∗∗ 0.21∗ 0.21∗ 0.22∗ 0.22∗∗ 0.22∗∗ 0.22∗∗ French language -0.06 -0.08 -0.07 -0.18 -0.22† -0.21† -0.26 -0.29 -0.29 -0.11 -0.16 -0.09 Other language -0.18 0.00 0.01 -0.19† -0.15 -0.18 -0.69∗∗∗ -0.42∗ -0.70∗∗∗ -0.22 -0.16 -0.27† Catholic -0.77∗∗∗ -0.78∗∗∗ -0.78∗∗∗ -0.63∗∗∗ -0.61∗∗∗ -0.61∗∗∗ -0.72∗∗∗ -0.76∗∗∗ -0.73∗∗∗ -0.14 -0.14 -0.16 Protestant -0.76∗∗∗ -0.78∗∗∗ -0.77∗∗∗ -0.61∗∗∗ -0.63∗∗∗ -0.63∗∗∗ -0.80∗∗∗ -0.84∗∗∗ -0.82∗∗∗ -0.40∗∗∗ -0.41∗∗∗ -0.42∗∗∗ Other religion -0.53∗∗ -0.56∗∗ -0.54∗∗ -0.26† -0.18 -0.18 -0.98∗∗∗ -1.02∗∗∗ -1.04∗∗∗ -0.20 -0.31† -0.31† Less than 34 years-old 0.20∗ 0.22∗ 0.21∗ 0.14∗ 0.12† 0.12† 0.20 0.22† 0.21† -0.03 -0.05 -0.05 More than 55 years-old -0.11 -0.12 -0.11 -0.17∗ -0.20∗ -0.20∗ -0.29∗ -0.24∗ -0.30∗ -0.10∗ -0.18∗ -0.18∗ 192 Less than high school 0.10 0.11 0.11 0.10 0.10 0.10 0.14 0.16 0.19 -0.05 -0.05 -0.03 University Degree 0.01 0.02 0.00 0.23∗∗ 0.24∗∗ 0.24∗∗ 0.21† 0.23∗ 0.20† 0.01 0.02 0.01 Low Income -0.21∗ -0.22∗ -0.23∗ 0.06 0.07 0.07 0.09 0.05 0.08 0.05 0.07 0.07 High Income -0.63∗∗∗ -0.63∗∗∗ -0.63∗∗∗ -0.50∗∗∗ -0.51∗∗∗ -0.51∗∗∗ -0.41∗∗ -0.46∗∗∗ -0.44∗∗∗ -0.26∗∗ -0.25∗∗ -0.24∗∗ Urban 0.36∗∗∗ 0.36∗∗∗ 0.36∗∗∗ 0.23∗∗∗ 0.24∗∗∗ 0.24∗∗∗ 0.24∗ 0.27∗∗ 0.26∗ -0.10 -0.07 -0.11 (Intercept) -0.91∗∗∗ -0.90∗∗∗ -0.91∗∗∗ -1.53∗∗∗ -1.51∗∗∗ -1.52∗∗∗ -1.04∗∗∗ -0.97∗∗∗ -1.02∗∗∗ -0.87∗∗∗ -0.90∗∗∗ -0.88∗∗∗ N 7104 7017 6999 10124 9901 9876 3131 3091 3044 3615 3661 3581 AIC 5011.3 4965.7 4853.5 7522.3 7324.3 7321.8 2783 2721.3 2701.5 4198.7 4232 4151 Source: Canadian Election Study, 1965-2011. † signiﬁcant at p < .10; ∗p < .05; ∗∗p < .01; ∗∗∗p < .001 Table 20: Who Shifted? The New Conservative Voters in the 2011 Election

Shifted for the Conservatives in 2011

Immigrant*Race — — — 2.22∗∗ — — — (0.72) Immigrant 0.25 — 0.17 -0.25 (0.25) — (0.26) (0.31) Race — 0.49† 0.45 -0.77 — (0.28) (0.30) (0.60) Controls Quebec -0.21 -0.18 -0.19 -0.20 (0.30) (0.29) (0.30) (0.30) West -0.06 -0.04 -0.03 -0.04 (0.20) (0.20) (0.20) (0.20) Maritimes 0.34 0.32 0.32 0.25 (0.24) (0.24) (0.24) (0.24) Woman 0.23 0.23 0.23 0.26† (0.15) (0.16) (0.16) (0.16) French language -0.09 -0.08 -0.10 -0.14 (0.30) (0.30) (0.31) (0.31) Other language 0.58∗ 0.60∗∗ 0.55∗ 0.54∗ (0.26) (0.23) (0.26) (0.34) Catholic 0.03 0.08 0.07 0.09 (0.23) (0.23) (0.24) (0.24) Protestant 0.07 0.11 0.10 0.10 (0.22) (0.23) (0.23) (0.23) Other religion -0.17 -0.10 -0.18 -0.30 (0.33) (0.33) (0.33) (0.34) Less than 34 years-old 0.29 0.32 0.34 0.37 (0.23) (0.23) (0.23) (0.24) More than 55 years-old -0.35∗ -0.28 -0.30† -0.25 (0,17) (0.18) (0.18) (0.18) Less than high school 0.48∗ 0.53∗ 0.55∗ 0.55∗ (0.22) (0.22) (0.22) (0.22) University Degree -0.38∗ -0.38∗ -0.38∗ -0.41∗ (0.18) (0.18) (0.18) (0.19) Low Income 0.11 0.06 0.07 0.07 (0.25) (0.25) (0.25) (0.25) High Income -0.07 -0.02 -0.04 -0.02 (0.18) (0.18) (0.18) (0.18) Urban -0.01 0.00 -0.02 -0.06 (0.17) (0.18) (0.18) (0.18) (Intercept) -2.58∗∗∗ -2.58∗∗∗ -2.68∗∗∗ -2.64∗∗∗ (0.28) (0.28) (0.28) (0.28) AIC 1406.2 1393.1 1380.4 1370.9 N 2705 2685 2667 2667 Source: Canadian Election Study, 1965-2011. † significant at p < .10; ∗p < .05; ∗∗p < .01; ∗∗∗p < .001 Method: Logistic regression. The coefficients represent effects on log odds. Note: The dependent variable is a dichotomous variable considering those voters who voted Conservatives in 2011, but recall to have voted otherwise in 2008.

193 Table 21: Targeted Ridings: Race, Proximity on All Issues, or Targeted Issue Publics?

Probability to vote for the Conservatives (1) (2) (3) (4) (5) (6) Focal variables

Targeted ridings X Race −0.006 (0.010) Targeted ridings X Proximity −0.002 (0.022) Targeted ridings X Issue Public 0.059∗∗∗ (0.017) Controls

French −0.118∗∗∗ −0.118∗∗∗ −0.027∗∗∗ −0.027∗∗∗ −0.118∗∗∗ −0.118∗∗∗ (0.003) (0.003) (0.002) (0.002) (0.003) (0.003) Quebec −0.067∗∗∗ −0.067∗∗∗ −0.030∗∗∗ −0.030∗∗∗ −0.067∗∗∗ −0.067∗∗∗ (0.003) (0.003) (0.002) (0.002) (0.002) (0.002) Maritimes −0.019∗∗∗ −0.019∗∗∗ −0.015∗∗∗ −0.015∗∗∗ −0.022∗∗∗ −0.022∗∗∗ (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) West 0.035∗∗∗ 0.035∗∗∗ 0.010∗∗∗ 0.010∗∗∗ 0.039∗∗∗ 0.039∗∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Catholic 0.071∗∗∗ 0.071∗∗∗ 0.015∗∗∗ 0.015∗∗∗ 0.065∗∗∗ 0.065∗∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Protestant 0.103∗∗∗ 0.103∗∗∗ 0.036∗∗∗ 0.036∗∗∗ 0.096∗∗∗ 0.096∗∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Other religion 0.001 0.001 −0.002 −0.002 0.0005 0.0005 (0.002) (0.002) (0.001) (0.001) (0.002) (0.002) Immigrant −0.020∗∗∗ −0.020∗∗∗ −0.018∗∗∗ −0.018∗∗∗ −0.024∗∗∗ −0.024∗∗∗ (0.002) (0.002) (0.001) (0.001) (0.002) (0.002) Woman −0.104∗∗∗ −0.104∗∗∗ −0.010∗∗∗ −0.010∗∗∗ −0.088∗∗∗ −0.088∗∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Less than 30 y-o −0.014∗∗∗ −0.014∗∗∗ 0.056∗∗∗ 0.056∗∗∗ −0.004∗∗∗ −0.004∗∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) More than 65 y-o 0.022∗∗∗ 0.022∗∗∗ −0.012∗∗∗ −0.012∗∗∗ 0.014∗∗∗ 0.014∗∗∗ (0.002) (0.002) (0.001) (0.001) (0.002) (0.002) Less than high school 0.033∗∗∗ 0.033∗∗∗ −0.013∗∗∗ −0.013∗∗∗ 0.025∗∗∗ 0.025∗∗∗ (0.003) (0.003) (0.002) (0.002) (0.003) (0.003) University degree −0.052∗∗∗ −0.052∗∗∗ 0.013∗∗∗ 0.013∗∗∗ −0.047∗∗∗ −0.047∗∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Graduate degree −0.103∗∗∗ −0.103∗∗∗ 0.003∗∗∗ 0.003∗∗∗ −0.095∗∗∗ −0.095∗∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Urban −0.039∗∗∗ −0.039∗∗∗ −0.001 −0.001 −0.038∗∗∗ −0.038∗∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Low income −0.032∗∗∗ −0.032∗∗∗ −0.016∗∗∗ −0.016∗∗∗ −0.030∗∗∗ −0.030∗∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) High income 0.046∗∗∗ 0.046∗∗∗ 0.018∗∗∗ 0.018∗∗∗ 0.039∗∗∗ 0.039∗∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Targeted ridings 0.029∗∗∗ 0.030∗∗∗ 0.013∗∗∗ 0.014 0.028∗∗∗ 0.002 (0.004) (0.004) (0.003) (0.010) (0.004) (0.008) Race 0.013∗∗∗ 0.013∗∗∗ −0.004∗∗∗ −0.004∗∗∗ 0.009∗∗∗ 0.009∗∗∗ (0.002) (0.002) (0.001) (0.001) (0.002) (0.002) Proximity on all issues 1.991∗∗∗ 1.991∗∗∗ (0.003) (0.003) Issue Public 0.423∗∗∗ 0.422∗∗∗ (0.003) (0.003) constant 0.436∗∗∗ 0.436∗∗∗ −0.530∗∗∗ −0.530∗∗∗ 0.246∗∗∗ 0.246∗∗∗ (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) N 483, 552 483, 552 483, 552 483, 552 483, 552 483, 552 R2 0.103 0.103 0.468 0.468 0.152 0.152 Adjusted R2 0.103 0.103 0.468 0.468 0.152 0.152 Source: Vote Compass, 2011 Canadian Federal Election. ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

194 Table 22: Left-right scale information

Issue items Factor loadings

All Canadian troops should be pulled out,of Afghanistan immediately [Reversed] 0.38 Canada should increase its military presence in the Arctic 0.31 How much should the government spend on the military? 0.56 When there is an economic problem, government,spending usually makes it worse 0.30 The federal budget deficit should be reduced, even if it leads to fewer public services 0.50 Canada should seek closer economic relations with the USA 0.39 The environmental damage caused by the Alberta oil sands industry is exaggerated 0.66 Canada should adopt a carbon tax [Reversed] 0.63 195 Environmental regulation should be stricter, even if it leads to consumers having to pay higher prices [Reversed] 0.61 How much of a role should the private sector have in health care? 0.51 The government should fund daycare instead of giving,money directly to parents [Reversed] 0.49 It should be easier to qualify for Employment Insurance [Reversed] 0.39 How many new immigrants should Canada admit? [Reversed] 0.34 How much should be done to accommodate religious minorities in Canada? [Reversed] 0.30 Violent young offenders should be sentenced as adults 0.52 The long gun registry should be scrapped 0.58 Possession of marijuana should be a criminal offence 0.40 The government should make it easier for a woman to get an abortion [Reversed] 0.44 Marriage should only be between a man and a woman 0.57 If they so wish, terminally ill patients should be able to end their own lives with medical assistance 0.27 How much should wealthier people pay in taxes? [Reversed] 0.47 How much tax should corporations pay? [Reversed] 0.56

Source: Vote Compass, 2011 Canadian Federal Election. The scale is built using the factor scores for the first factor of principal factors factor analyses. Eigenvalue for the first factor = 5.87. Table 23: Jewish and Muslim Effect on Conservative Vote: Religious or Issue-Public?

Vote for Conservatives (1) (2) (3) (4) Maritimes −0.278∗∗∗ −0.236∗∗∗ −0.237∗∗∗ −0.225∗∗∗ (0.027) (0.029) (0.029) (0.040) Quebec −0.404∗∗∗ −0.338∗∗∗ −0.338∗∗∗ −0.303∗∗∗ (0.042) (0.043) (0.043) (0.060) West 0.264∗∗∗ 0.286∗∗∗ 0.286∗∗∗ 0.267∗∗∗ (0.017) (0.018) (0.018) (0.024) French −0.683∗∗∗ −0.468∗∗∗ −0.468∗∗∗ −0.442∗∗∗ (0.044) (0.046) (0.046) (0.063) Immigrant −0.130∗∗∗ −0.002 −0.002 0.074∗∗ (0.021) (0.023) (0.023) (0.032) Woman −0.643∗∗∗ −0.517∗∗∗ −0.517∗∗∗ −0.460∗∗∗ (0.016) (0.017) (0.017) (0.024) Less than 30 y-o −0.291∗∗∗ −0.287∗∗∗ −0.287∗∗∗ −0.233∗∗∗ (0.027) (0.028) (0.028) (0.035) More than 65 y-o 0.277∗∗∗ 0.342∗∗∗ 0.342∗∗∗ 0.380∗∗∗ (0.018) (0.020) (0.020) (0.030) Less than high school 0.310∗∗∗ 0.407∗∗∗ 0.406∗∗∗ 0.406∗∗∗ (0.039) (0.043) (0.043) (0.065) University degree −0.706∗∗∗ −0.781∗∗∗ −0.781∗∗∗ −0.709∗∗∗ (0.016) (0.017) (0.017) (0.023) Urban −0.300∗∗∗ −0.309∗∗∗ −0.309∗∗∗ −0.291∗∗∗ (0.016) (0.017) (0.017) (0.023) Low income −0.265∗∗∗ −0.158∗∗∗ −0.157∗∗∗ −0.177∗∗∗ (0.023) (0.024) (0.024) (0.035) High income 0.258∗∗∗ 0.205∗∗∗ 0.205∗∗∗ 0.212∗∗∗ (0.016) (0.017) (0.017) (0.024) Politically Interested −0.202∗∗∗ −0.334∗∗∗ −0.334∗∗∗ −0.442∗∗∗ (0.029) (0.031) (0.031) (0.042) Jewish 0.218∗∗∗ 0.203∗∗∗ 0.199 0.351∗ (0.067) (0.071) (0.150) (0.203) Muslim −0.907∗∗∗ −0.686∗∗∗ 0.077 0.373 (0.128) (0.135) (0.280) (0.415) Troops out of Afghanistan −0.529∗∗∗ −0.528∗∗∗ −0.419∗∗∗ (0.006) (0.006) (0.016) Salience of foreign aﬀairs issue 1.208∗∗∗ (0.059) Jewish X Troops out 0.002 −0.054 (0.052) (0.072) Muslim X Troops out −0.279∗∗∗ −0.379∗∗∗ (0.096) (0.138) Salience X Troops out −0.089∗∗∗ (0.018) N 105, 903 104, 520 104, 520 58, 427 Log likelihood −55, 116.880 −49, 987.610 −49, 982.960 −26, 401.950 AIC 110, 267.800 100, 011.200 100, 005.900 52, 847.910 Source: Vote Compass, 2011 Canadian Federal Election. ∗p < .1; ∗∗p < .05; ∗∗∗p < .01

196 References

Aarts, Kees and Holli A. Semetko. 2003. “The Divided Electorate: Media Use and Political Involvement.” Journal of Politics 65(3):759–784.

Abrajano, Marisa A., R. Michael Alvarez and Jonathan Nagler. 2008. “The Hispanic Vote in the 2004 Presidential Election: Insecurity and Moral Concern.” Journal of Politics 70(2):368–382.

Abramowitz, Alan I. 1995. “It’s Abortion, Stupid: Policy Voting in the 1992 Presidential Election.” Journal of Politics 57:176–186.

Abramson, Paul R. 1983. Political Attitudes in America: Formation and Change. San Francisco: W. H. Freeman and Company.

Abramson, Paul R. and Ronald Inglehart. 1995. Value Change in Global Perspective. Ann Arbor: University of Michigan Press.

Achen, Christopher. 1975. “Mass Political Attitudes and the Survey Response.” American Political Science Review 69:1218–1231.

Achen, Christopher. 2005. “Let’s Put Garbage-Can Regressions and Garbage-Can Probits Where They Belong.” Conﬂict Management and Peace Science 22:327–339.

Albouy, Serge. 1994. Marketing et communication politique. Paris: Editions´ L’Harmattan.

Althaus, Scott L. 2003. Collective Preferences in Democratic Politics: Opinion Surveys and the Will of the People. Cambridge: Cambridge University Press.

Alvarez, Michael, Mark Franklin, Lorenzo de Sio, Alexander Trechsel and Alexander Treschel. 2011. “Weighting and Matching as Methods for Combining Disparate Survey

197 REFERENCES

Data: The European Election Study 2009 and EU Proﬁler.” Presented at the 2011 Midwest Political Science Association Annual Meeting.

Alvarez, R. Michael. 1998. Information and elections. Ann Arbor: University of Michigan Press.

Alvarez, R. Michael, Asa Hopkins and Betsy Sinclair. 2010. “Mobilizing Pasadena Democrats: Measuring The Eﬀect of Partisan Campaign Contacts.” Journal of Pol- itics 72(1):31–44.

Alvarez, R. Michael and John Brehm. 2002. Hard Choices, Easy Answers. Princeton: Princeton University Press.

Andrew, Blake, Antonia Maioni and Stuart Soroka. 2006. “Just When You Thought It Was Out, Policy Is Pulled Back In.” Policy Options/Options politiques pp. 74–79.

Ansolabehere, Stephen. 2000. “Valence Politics and Equilibrium in Spatial Election Mod- els.” Public Choice 103(3/4):327–336.

Ansolabehere, Stephen, Jonathan Rodden and James M. Snyder Jr. 2008. “The Strength of Issues: Using Multiple Measures to Gauge Preference Stability, Ideological Con- straint, and Issue Voting.” American Political Science Review 102(2):215–232.

Ansolabehere, Stephen, Roy L. Behr and Shanto Iyengar. 1993. The media game: Amer- ican politics in the television age. New-York: Macmillan.

Ansolabehere, Stephen and Shanto Iyengar. 1995. Going Negative: How political advertisements shrink and polarize the electorate. New-York: Free Press.

Axelrod, Robert. 1973. “Schema Theory: An Information Processing Model of Perception and Cognition.” American Political Science Review 67(4):1248–1266.

Baines, Paul R. 1999. Voter Segmentation and Candidate Positioning. In Handbook of Political Marketing, ed. Bruce I. Newman. Thousand Oaks: Sage Publications pp. 403– 420.

Baines, Paul R. and John Egan. 2001. “Marketing and Political Campaigning: Mutually Exclusive or Exclusively Mutual?” Qualitative Market Research: An International Journal 4(1):25–33.

198 REFERENCES

Baldassarri, Delia and Amir Goldberg. 2010. “Political Belief Networks: Socio-Cognitive Heterogeneity in American Public Opinion.” Working Papers.

Bartels, Larry M. 2002. “Beyond the Running Tally: Partisan Bias in Political Percep- tions.” Political Behavior 24(2):117–150.

Bartels, Larry M. 2003. Democracy with attitudes. In Electoral Democracy, ed. Michael B. MacKuen and George Rabinowitz. Ann Arbor: University of Michigan Press pp. 194– 230.

Bartels, Larry M. 2006. “What’s the Matter with What’s the Matter with Kansas?” Quartely Journal of Political Science 1:201–226.

Bartle, John. 2005. “Homogeneous Models and Heterogeneous Voters.” Political Studies 53:653–675.

Bartle, John and Dylan Griﬃths. 2002. Social-Psychological, Economic and Marketing Models of Voting Behaviour Compared. In The Idea of Political Marketing, ed. Stephen C. M. Henneberg and Nicholas J. O’Shaughnessy. Westport, CT: Praeger pp. 19–37.

Bastien, Frédérick. 2013. Tout le monde en regarde! Quebec: Presses de l’Université Laval.

Baum, Matthew A. 2002. “Sex, lies and war: How soft news brings foreign policy to the inattentive public.” American Political Science Review 96(1):91–109.

Baum, Matthew A. and Sam Kernell. 1999. “Has Cable Ended the Golden Age of Presi- dential Television?” American Political Science Review 93:99–114.

B´elanger, Eric´ and Bonnie M. Meguid. 2008. “Issue salience, issue ownership, and issue- based vote choice.” Electoral Studies 27(3):477–491.

Bélanger, Eric´ and Richard Nadeau. 2009. Le comportement électoral des Québécois. Montreal: Presses de l’Universitéde Montréal.

Bell, Edward, Julie Aitken Schermer and Philip A. Vernon. 2009. “The Origins of Political Attitudes and Behaviours: An Analysis Using Twins.” Canadian Journal of Political Science 42(4):855–879.

199 REFERENCES

Bennett, W. Lance. 1977. “The Growth of Knowledge in Mass Belief Studies: An Epis- temological Critique.” American Journal of Political Science 21(3):465–500.

Bennett, W. Lance. 2012. “The Personalization of Politics: Political Identity, Social Me- dia, and Changing Patterns of Participation.” The ANNALS of the American Academy of Political and Social Science 644(20).

Berelson, Bernard R., Paul F. Lazarsfeld and William N. McPhee. 1954. Voting: A Study of Opinion Formation in a Presidential Campaign. Chicago: University of Chicago Press.

Bernier, Robert. 1991. Gérer la victoire? Organisation, communication, stratégie. Montréal:GaëtanMorin Editeur.´

Bernier, Robert. 2001. Un sièclede propagande? Information. Communication. Marketing gouvernemental. Sainte-Foy: Presses de l’Universitédu Québec.

Bickerton, James and Alain-G. Gagnon. 2004a. Political Parties and Electoral Politics. In Canadian politics, ed. James Bickerton and Alain-G. Gagnon. Toronto: University of Toronto Press pp. 239–262.

Bickerton, James and Alain Gagnon. 2004b. Canadian politics. Fourth ed. Toronto: University of Toronto Press.

Bilodeau, Antoine and Mebs Kanji. 2010. The New Immigrant Voter, 1965-2004: The Emergence of a New Liberal Partisan? In Voting Behaviour in Canada, ed. Cameron D. Anderson and Laura B. Stephenson. Vancouver: UBC Press pp. 65–85.

Bittner, Amanda and Royce Koop, eds. 2013. Parties, Elections, and the Future of Canadian Politics. Vancouver: UBC Press.

Black, Jerome, Richard Niemi and Bingham G. Powell. 1987. “Age and Resistance to Political Learning in a New Environment: The Case of Canadian Immigrants.” Com- parative Politics 20:73–84.

Blais, Andr´e.2000. To Vote Or Not To Vote? The Merits and Limits of Rational Choice Theory. Pittsburgh: University of Pittsburgh Press.

200 REFERENCES

Blais, Andr´e.2005. “Accounting for the Electoral Success of the Liberal Party in Canada.” Canadian Journal of Political Science 38(4):821–840.

Blais, Andr´e,Elisabeth Gidengil, Richard Nadeau and Neil Nevitte. 2002. Anatomy of a Liberal Victory. Peterborough: Broadview Press.

Blake, Donald E. 1982. “The Consistency of Inconsistency: Party Identiﬁcation in Federal and Provincial Politics.” Canadian Journal of Political Science 15:681–710.

Blidook, Kelly. 2008. “Media, Public Opinion and Health Care in Canada: How the Media Aﬀect “The Way Things Are”.” Canadian Journal of Political Science 41(2):355–374.

Bourassa, Henri. 1902. The French-Canadian in the British Empire. London: John Murray.

Bowler, Shaun and David Farrell, eds. 1992. Electoral strategies and political marketing. New-York: St-Martin’s Press.

Brady, Henry E. and Paul M. Sniderman. 1985. “Attitude Attribution: A Group Basis for Political Reasoning.” American Political Science Review 79(4):1061–1078.

Bricker, Darrell and John Ibbitson. 2013. The Big Shift. Toronto: HarperCollins.

Brown, Stephen. 1993. “Postmodern Marketing?” European Journal of Marketing 27(4):19–34.

Budge, Ian and Dennis J. Farlie. 1983. Explaining and Predicting Elections: Issue Eﬀects and Party Strategies in Twenty-Three Democracies. Boston: Allen and Unwin.

Budge, Ian, Hans-Dieter Klingemann, Andrea Volkens, Judith Bara and Eric Tanenbaum, eds. 2001. Mapping Policy Preferences: Estimates for Parties, Electors, and Govern- ments, 1945-1998. Oxford: Oxford University Press.

Budge, Ian, Ivor Crewe and Dennis J. Farlie. 1976. Party identiﬁcation and beyond. New York: John Wiley.

Burnett, John J. 2007. Nonproﬁt Marketing Best Practices. Wiley.

Butler, David and Donald Stokes. 1969. Political Change in Britain: Forces Shaping Electoral Choice. New-York: St-Martin’s Press.

201 REFERENCES

Butler, David and Donald Stokes. 1975. Political Change in Britain: The Evolution of Electoral Choice. New-York: St-Martin’s Press.

Campbell, Angus, Philip E. Converse, Waren Miller and Donald Stokes. 1960. The Amer- ican Voter. New-York: Wiley.

Carmines, Edward G. and James A. Stimson. 1980. “The Two Faces of Issue Voting.” American Political Science Review (74):78–91.

Carmines, Edward G. and James A. Stimson. 1989. Issue Evolution: Race and the Trans- formation of American Politics. Princeton: Princeton University Press.

Carty, R. Kenneth. 2013. Has Brokerage Politics Ended? Canadian Parties in the New Century. In Parties, Elections, and the Future of Canadian Politics, ed. Amanda Bittner and Royce Koop. Vancouver: UBC Press pp. 10–23.

Carty, R. Kenneth, Lisa Young and William P. Cross. 2000. Rebuilding Canadian Party Politics. Vancouver: UBC Press.

Carty, Roland Kenneth. 2002. “The Politics of Tecumseh Corners: Canadian Political Parties as Franchise Organization.” Canadian Journal of Political Science 35(4):723– 745.

Clarke, Harold D., Allan Kornberg and Thomas J. Scotto. 2008. Making Political Choices: Canada and the United States. Toronto: University of Toronto Press.

Clarke, Harold D., Allan Kornberg, Thomas J. Scotto, Jason Reiﬂer, David Sanders, Marianne C. Stewart and Paul Whiteley. 2010. “Yes we can! Valence politics and electoral choice in America, 2008.” Electoral Studies pp. 1–12.

Clarke, Harold D., Allan Kornberg, Thomas Scotto and Joe Twyman. 2006. “Flawless campaign, fragile victory: Voting in Canada’s 2006 federal election.” PS: Political Sci- ence & Politics 39(4):815–819.

Clarke, Harold D. and Allan L. McCutcheon. 2009. “The Dynamics of Party Identiﬁcation Reconsidered.” Public Opinion Quarterly 73(4):704–728.

Clarke, Harold D., David Sanders, Marianne C. Stewart and Paul Whiteley. 2009. “The American voter’s British cousin.” Electoral Studies 28:632–641.

202 REFERENCES

Clarke, Harold D., Jane Jenson, Lawrence Leduc and Jon H. Pammett. 1984. Absent Mandate: the politics of discontent in Canada. First ed. Toronto: Gage.

Clarke, Harold D., Jane Jenson, Lawrence Leduc and Jon H. Pammett. 1996. Absent Mandate: Canadian Electoral Politics in an Era of Restructuring. Third ed. Toronto: Gage.

Clarke, Harold D., Thomas J. Scotto and Allan Kornberg. 2010. “Valence politics and economic crisis: Electoral choice in Canada 2008.” Electoral Studies pp. 1–12.

Clarkson, Stephen. 2005. The Big Red Machine: how the Liberal Party dominates Cana- dian politics. Vancouver: UBC Press.

Conover, Pamela Johnston and Stanley Feldman. 1984. “How People Organize the Polit- ical World: a Schematic Model.” American Journal of Political Science 28(1):95–126.

Converse, Philip E. 1962. “Information ﬂow and the stability of partisan attitudes.” Public Opinion Quarterly 26(4):578–599.

Converse, Philip E. 1964. The Nature of Belief Systems in Mass Publics. In Ideology and Discontent, ed. David Apter. New-York: Free Press.

Converse, Philip E. 1966. The concept of a normal vote. In Elections and the Political Order, ed. Angus Campbell. John Wiley pp. 9–39.

Converse, Philip E. 1970. Attitudes and Non-Attitudes: Continuation of a Dialogue. In The Quatitative Analysis of Social Problems, ed. Edward R. Tufte. Addison-Wesley pp. 168–189.

Converse, Philip E. 1975. Public opinion and voting behavior. In Handbook of Politi- cal Science, ed. Fred I. Greenstein and Nelson W. Polsby. Reading, Massachusetts: Addison-Wesley pp. 75–100.

Converse, Philip E. 2007. Perspectives on Mass Belief Systems and Communication. In The Oxford Handbook of Political Behavior, ed. Russell J. Dalton and Hans-Dieter Klingemann. Oxford: Oxford University Press pp. 144–158.

Converse, Philip E. and Gregory B. Markus. 1979. “Plus ¸cachange...: The New CPS Election Study Panel.” American Political Science Review 73(1):32–49.

203 REFERENCES

Converse, Philip E. and Roy Pierce. 1986. Political Representation in France. Cambridge: Harvard University Press.

Couper, Mick P. 2008. Designing Eﬀective Web Surveys. New-York: Cambridge University Press.

Dale, Allison and Aaron Strauss. 2009. “Don’t Forget to Vote: Text Message Reminders as a Mobilization Tool.” American Journal of Political Science 53(4):787–804.

Dalton, Russel J. 2007. “Partisan mobilization, cognitive mobilization and the changing American electorate.” Electoral Studies 26:274–286.

Dalton, Russel and Martin Wattenberg. 2001. Parties without Partisans: Political Change in Advanced Industrial Democracies. New-York: Oxford University Press.

Dalton, Russell J. 1984. “Cognitive Mobilization and Partisan Dealignment in Advanced Industrial Democracies.” Journal of Politics 46:264–284.

Dalton, Russell J. 1988. Citizen Politics in Western Democracies: Public Opinion and Political Parties in the United States, Great Britain, West Germany, and France. Chatham: Chatham.

Dalton, Russell J. 1999. Political Support in Advanced Industrial Democracies. In Critical citizens: global support for democratic government, ed. Pippa Norris. Oxford: Oxford University Press pp. 57–77.

Damore, David F. 2004. “The dynamics of issue ownership in presidential campaigns.” Political Research Quartely 57(3):391–397.

Davidov, Eldad, Peter Schmidt and Shalom H. Schwartz. 2008. “Bringing Values Back In: The Adequacy of the European Social Survey to Measure Values in 20 Countries.” Public Opinion Quarterly 72(3):420–445.

Dawes, Christopher T. and James H. Fowler. 2009. “Partisanship, Voting, and the Dopamine D2 Receptor Gene.” Journal of Politics 71(3):1157–1171. de la Garza, Rodolfo O. and Jeronimo Cortina. 2007. “Are Latinos Republicans But Just Don’t Know It? The Latino Vote in the 2000 and 2004 Presidential Elections.” American Politics Research 35(2):202–223.

204 REFERENCES de la Garza, Rodolfo O. and Louis DeSipio, eds. 2004. Muted Voices: Latinos and the 2000 Elections. Lanham: Rowman and Littleﬁeld. de Souza, Mike. 2011. “Jason Kenney apologizes, staﬀer quits over fundraising letter.” The Gazette.

DellaVigna, Stefano and Ethan Kaplan. 2007. “The Fox News Eﬀect: Media Bias and Voting.” The Quarterly Journal of Economics 122(3):1187–1234.

Delli Carpini, Michael X. and Scoot Keeter. 1993. “Measuring political knowledge: Putting ﬁrst things ﬁrst.” American Journal of Political Science 37(4):1179–1206.

Delli Carpini, Michael X. and Scott Keeter. 1996. What Americans Know About Politics and Why it Matters. New Haven: Yale University Press.

DeSipio, Louis. 1996. Counting on the Latino Vote: Latinos as a New Electorate. Char- lottesville: The University Press of Virginia.

Downs, Anthony. 1957. An Economic Theory of Democracy. New-York: Harper.

Druckman, James N., Lawrence R. Jacobs and Eric Ostermeier. 2004. “Candidate Strate- gies to Prime Issues and Image.” Journal of Politics 66(4):1205–1227.

Dufresne, Yannick, Gregory Eady, Jennifer Hove, Peter Loewen and Clifton van der Lin- den. 2012. “Media Campaign Events in the 2011 Canadian Federal Election: Examining Campaign Eﬀects with Big Data.” Paper presented at the Annual Meeting of the Cana- dian Political Association.

Egan, John. 1999. “Political Marketing: Lessons from the Mainstream.” Journal of Mar- keting Management 15:495–503.

Ellis, Faron and Keith Archer. 1997. Reform at the Crossroads. In The Canadian Federal Election of 1997, ed. Alan Frizzell and Jon H. Pammett. Toronto: Dundurn pp. 111– 134.

Enelow, James M. and Melvin J. Hinich. 1984. The spatial theory of voting: an introduction. Cambridge: Cambridge University Press.

Entman, Robert M. 2004. Projections of Power. Chicago: University of Chicago Press.

205 REFERENCES

Epstein, Lee and Jeﬀrey A. Segal. 2000. “Measuring Issue Salience.” American Journal of Political Science 44(1):66–83.

Erbring, Lutz, Edie Goldenberg and Arthur Miller. 1980. “Front-page news and real-world cues: A new look at agenda-setting.” American Journal of Political Science 24:16–49.

Erikson, Robert S., Michael B. MacKuen and James A. Stimson. 2001. The Macro Polity. New-York: Cambridge University Press.

Farrell, David M. and Paul Webb. 2001. Political Parties as Campaign Organizations. In Parties without Partisans, ed. Russel Dalton and Martin Wattenberg. New-York: Oxford University Press pp. 102–128.

Feldman, Stanley. 1988. “Structure and Consistency in Public Opinion: the Role of Core Beliefs and Values.” American Journal of Political Science 32(2):416–440.

Festinger, Leon. 1957. A theory of cognitive dissonance. New York: Harper & Row.

Fiorina, Morris P. 1981. Retrospective Voting in American National Elections. New Haven: Yale University Press.

Flanagan, Tom. 2011. “The Emerging Conservative Coalition.” Policy Options/Options politiques .

Fournier, Patrick, Fran¸coisG´elineauand Allison Harell. 2014. “Are Voting Advice Ap- plications Useful for the Study of Elections?” Working Paper.

Franklin, Charles H. 1992. “Measurement and the Dynamics of Party Identiﬁcation.” Political Behavior 14(3):297–309.

Franklin, Charles H. and John E. Jackson. 1983. “The Dynamics of Party Identiﬁcation.” American Political Science Review 77(4):957–973.

Franklin, Mark N., Thomas T. Mackie and Henry Valen, eds. 1992. Electoral Change: Responses to Evolving Social and Attitudinal Structures in Western Countries. Cam- bridge: Cambridge University Press.

Friesen, Joe and Julian Sher. 2011. “How Courting the Immigrant Vote Paid Oﬀ for the Tories.” The Globe and Mail.

206 REFERENCES

Gamson, William A. 1996. Media Discourse as a Framing Resource. In The Psychology of Political Communication, ed. Ann N. Crigler. Ann Arbor: University of Michigan Press pp. 111–132.

Gamson, William A. and Andre Modigliani. 1987. The changing culture of aﬃrmative action. In Research in Political Sociology, ed. Richard D. Braungart. Greenwich, CT: JAI pp. 137–177.

Garson, David G. 2010. “Scales and Standard Measures.”. URL: http://faculty.chass.ncsu.edu/garson/pa765/statnote.htm

Geer, John G. 1996. From Tea Leaves to Opinion Polls: A Theory of Democratic Lead- ership. New-York: Columbia University Press.

Geer, John G. 2006. In Defense of Negativity: Attack Ads in Presidential Campaigns. Chicago: University of Chicago Press.

Gelman, Andrew, David Park, Boris Shor, Joseph Bafumi and Jeronimo Cortina. 2008. Red State, Blue State, Rich State, Poor State: Why Americans Vote the Way They Do. Princeton: Princeton University Press.

Gelman, Andrew and Gary King. 1993. “Why Are American Presidential Polls So Variable When Election Outcomes Are So Predictable?” British Journal of Political Science 23:403–451.

Gerber, Alan and Donald P. Green. 1998. “Rational Learning and Partisan Attitudes.” American Journal of Political Science 42(3):794–818.

Gerber, Alan S., Gregory A. Huber, David Doherty, Conor M. Dowling and Shang E. Ha. 2010. “Personality and Political Attitudes: Relationships across Issue Domains and Political Contexts.” American Political Science Review 104(1):111–133.

Gerring, John. 1997. “Ideology: A Deﬁnitional Analysis.” Political Research Quarterly 50(4):957–994.

Giasson, Thierry, Colette Brin and Marie-Michele Sauvageau. 2010. “Le Bon, la Brute et le Raciste. Analyse de la couverture m´ediatiquede l’opinion publique pendant la

207 REFERENCES

“crise” des accommodements raisonnables au Qu´ebec.” Canadian Journal of Political Science 43(2):379–406.

Gidengil, Elisabeth. 1992. “Canada Votes: A Quarter Century of Canadian National Election Studies.” Canadian Journal of Political Science 25(2):219–248.

Gidengil, Elisabeth, Neil Nevitte, Andr´eBlais, Joanna Everitt and Patrick Fournier. 2012. Dominance and Decline: Making Sense of Recent Canadian Elections. Toronto: University of Toronto Press.

Gimpel, James G. 1999. Separate Destinations: Migration, Immigration, and the Politics of Places. Ann Arbor: University of Michigan Press.

Ginsberg, Benjamin I. 1986. The Captive Public: How Mass Opinion Promotes State Power. New-York: Basic Books.

Goldberg, Amir. 2011. “Mapping Shared Understandings Using Relational Class Analy- sis: The Case of the Cultural Omnivore Reexamined.” American Journal of Sociology 116(5):1397–1436.

Goodyear-Grant, Elizabeth. 2013. Women Voters, Candidates, and Legislators: A Gender Perspective on Recent Party and Electoral Politics. In Parties, Elections, and the Future of Canadian Politics, ed. Amanda Bittner and Royce Koop. Vancouver: UBC Press pp. 119–139.

Goren, Paul. 1997. “Political Expertise and Issue Voting in Presidential Elections.” Po- litical Research Quarterly 50:387–412.

Goren, Paul, Christopher M. Federico and Miki Caul Kittilson. 2009. “Source Cues, Par- tisan Identities, and Political Value Expression.” American Journal of Political Science 53(4):805–820.

Gould, Philip. 1998. The Unﬁnished Revolution: How the Modernisers Saved the Labour Party. London: Abacus.

Green, Donald and Jack Citrin. 1994. “Measurement Error and the Structure of Atti- tudes: Are Positive and Negative Judgments Opposites?” American Journal of Political Science 38(1):256–281.

208 REFERENCES

Green, Donald P., Bradley Palmquist and Eric Schickler. 2002. Partisan Hearts and Minds: Political Parties and the Social Identities of Voters. New Haven: Yale University Press.

Green, Donald P. and Ian Shapiro. 1996. Pathologies of Rational Choice Theory: A Critique of Applications in Political Science. New Haven: Yale University Press.

Green, Jane. 2007. “When Voters and Parties Agree: Valence Issues and Party Compe- tition.” Political Studies 55(3):629–655.

Greenspon, Edward. 2001. Covering Campaign 2000. In The Canadian General Election of 2000, ed. Jon H. Pammett and Christopher Dornan. Toronto: Dundurn pp. 165–190.

Gr¨onroos, Christian. 1994. “From the marketing mix to relationship marketing.” Man- agement Decision 32(2):4–20.

Haller, Max. 2002. “Theory and Method in the Comparative Study of Values: Critique and Alternative to Inglehart.” European Sociological Review 18(2):139–158.

Harrell, Allison. 2013. Revisiting the ‘Ethnic’ Vote: Liberal Allegiance and Vote Choice among Racialized Minorities. In Parties, Elections, and the Future of Canadian Politics, ed. Amanda Bittner and Royce Koop. Vancouver: UBC Press pp. 140–160.

Hatemi, Peter K., John R. Hibbing, Sarah E. Medland, Matthew C. Keller, John R. Alford, Kevin B. Smith, Nicholas G. Martin and Lindon J. Eaves. 2010. “Not by Twins Alone: Using the Extended Family Design to Investigate Genetic Inﬂuence on Political Beliefs.” American Journal of Political Science 54(3):798–814.

Heath, Anthony, Stephen Fisher and Shawna Smith. 2006. “The Globalization of Public Opinion Research.” Annual Review of Political Science 8:297–333.

H´ebert, Chantal. 2007. French Kiss: Stephen Harper’s Blind Date with Quebec. Toronto: Alfred A. Knopf.

Henderson, Ailsa. 2004. “Regional Political Cultures in Canada.” Canadian Journal of Political Science 37(3):595–615.

209 REFERENCES

Henneberg, Stephen C. M. 2002. Understanding Political Marketing. In The Idea of Political Marketing, ed. Nicholas J. O’Shaughnessy and Stephen C. M. Henneberg. Westport, CT: Praeger.

Henneberg, Stephen C. M. and Nicholas J. O’Shaughnessy. 2009. “Political Relation- ship Marketing: some macro/micro thoughts.” Journal of Marketing Management 25(1/2):5–29.

Herrnson, Paul S. 2009. “The Roles of Party Organizations, Party-Connected Committes, and Party Allies in Elections.” Journal of Politics 71(4):1207–1224.

Highton, Benjamin. 2009. “Revisiting the Relationship between Educational Attainment and Political Sophistication.” Journal of Politics 71(4):1564–1576.

Hillygus, D. Sunshine. 2007. “Campaign microtargeting and presidential voting in 2004.” Presented at the 2007 Annual Meetings of American Political Science Association.

Hillygus, D. Sunshine and Simon Jackman. 2003. “Voter Decision Making in Election 2000: Campaign Eﬀects, Partisan Activation, and the Clinton Legacy.” American Journal of Political Science 47(4):583–596.

Hillygus, D. Sunshine and Todd G. Shields. 2008. The Persuadable Voter: Wedge Issues in Political Campaigns. Princeton: Princeton University Press.

Hinich, Melvin J. and Michael Munger. 1994. Ideology and the Theory of Political Choice. Ann Arbor: University of Michigan Press.

Hochschild, Jennifer L. 1981. What’s Fair? American Beliefs About Distributive Justice. Cambridge: Harvard University Press.

Holbrook, Thomas M. 1996. Do campaigns matter? Thousand Oaks: Sage Publications.

Holcombe, Randall G. 1989. “The median voter model in public choice theory.” Public Choice 61(2):115–125.

Holian, David B. 2004. “He’s stealing my issues! Clinton’s crime rhetoric and the dynamics of issue ownership.” Political Behavior 26(2):95–124.

210 REFERENCES

Hooghe, Liesbet, Gary Marks and Carole J. Wilson. 2002. “Does left/right structure party positions on European integration?” Comparative Political Studies 35(8):965–989.

Huckfeldt, Robert and John Sprague. 1995. Citizens, Politics, and Social Communication: Information and Inﬂuence in an Election Campaign. New-York: Cambridge University Press.

Inglehart, Ronald. 1971. “The Silent Revolution in Europe: Intergenerational Change in Post-Industrial Societies.” American Political Science Review 65(4):991–1017.

Inglehart, Ronald. 1977. The Silent Revolution: Changing Values and Political Styles. Princeton: Princeton University Press.

Inglehart, Ronald. 1990. Culture Shift in Advanced Industrial Society. Princeton: Prince- ton University Press.

Inglehart, Ronald. 1998. Modernization and Postmodernization. Princeton: Princeton University Press.

Inglehart, Ronald and Paul R. Abramson. 1994. “Economic Security and Value Change.” American Political Science Review 88(2):336–354.

Iyengar, Shanto. 1991. Is Anyone Responsible? How Television Frames Political Issues. Chicago: University of Chicago Press.

Iyengar, Shanto and Donald R. Kinder. 1987. News That Matters: Television and Amer- ican Opinion. Chicago: University of Chicago Press.

Iyengar, Shanto, Kyu S. Hahn, Jon A. Krosnick and John Walker. 2008. “Selective Exposure to Campaign Communication: The Role of Anticipated Agreement and Issue Public Membership.” Journal of Politics 70(1):186–200.

Jackson, John E. 1975. “Issues, Party Choices, and Presidential Votes.” American Journal of Political Science 19(2):161–185.

Jackson, John E. 1983. “The Systematic Beliefs of the Mass Public: Estimating Policy Preferences with Survey Data.” Journal of Politics 45:840–865.

211 REFERENCES

Jacobs, Lawrence R. and Robert Shapiro. 1994. “Issues, Candidate Image, and Prim- ing: The Use of Private Polls in Kennedy’s 1960 Presidential Campaigns.” American Political Science Review 88(3):527–540.

Johansen, Helene P.M. 2005. “Political Marketing: More Than Persuasive Techniques, an Organizational Perspective.” Journal of Political Marketing 4(4):85–105.

Johnston, Richard. 1992. “Party Identiﬁcation Measures in the Anglo-American Democra- cies: A National Survey Experiment.” American Journal of Political Science 36(2):542– 559.

Johnston, Richard. 2006. “Party Identiﬁcation: Unmoved Mover or Sum of Preferences?” Annual Review of Political Science 9:329–351.

Johnston, Richard, Andr´eBlais, Henry E. Brady and Jean Cr`ete.1992. Letting People Decide: Dynamics of a Canadian Election. Montreal: McGill-Queen’s University Press.

Joppke, Christian. 2004. “The retreat of multiculturalism in the liberal state: theory and policy.” British Journal of Sociology 55(2):237–257.

Just, Marion R. 1996. Crosstalk: Citizens, candidates, and the media in a presidential campaign. Chicago: University of Chicago Press.

Kahle, Lynn R., Sharon E. Beatty and Pamela Homer. 1986. “Alternative Measurement Approaches to Consumer Values: The List of Values (LOV) and Values and Life Style (VALS).” The Journal of Consumer Research 13(3):405–409.

Katz, Elihu. 1957. “The Two-Step Flow of Communication: An Up-To-Date Report on an Hypothesis.” Political Opinion Quarterly 21(1):61–78.

Katz, Elihu and Paul F. Lazarsfeld. 1955. Personal Inﬂuence: The Part Played by People in the Flow of Mass Communications. New-York: Free Press.

Keith, Robert J. 1960. “The Marketing Revolution.” Journal of Marketing pp. 35–38.

Key Jr., V. O. 1966. The Responsible Electorate. Cambridge: Belknap Press.

Key, V. O. Jr. 1961. Public Opinion and American Democracy. New-York: Alfred A. Knopf.

212 REFERENCES

Kiewiet, Roderick D. and Douglas Rivers. 1985. The economic basis of Reagan’s appeal. In The new direction in American politics, ed. John E. Chubb and Paul E. Peterson. Washington DC: Brookings Institution.

Kim, Jae-On and Charles W. Mueller. 1978. Factor analysis: Statistical methods and practical issues. Beverly Hills: Sage Publications.

Kinder, Donald R. and D. Roderick Kiewiet. 1979. “Economic discontent and political behavior: the role of personal grievances and collective economic judgments in congressional voting.” American Journal of Political Science 23(3):495–527.

King, Gary. 1991. ““Truth” is stranger than prediction, more questionable than causal inference.” American Journal of Political Science 35(4):1047–1053.

King, Gary. 2011. “Ensuring the Data-Rich Future of the Social Sciences.” Science 331.

King, Gary, Robert O. Keohane and Sidney Verba. 1994. Designing Social Inquiry: Sci- entiﬁc Inference in Qualitative Research. Princeton: Princeton University Press.

Klingemann, Hans-Dieter. 1979a. The Background of Ideological Conceptualizations. In Political Action, ed. Samuel H. Barnes and Max Kaase. Beverly Hills: Sage Publica- tions.

Klingemann, Hans-Dieter. 1979b. Measuring ideological conceptualizations. In Political Action, ed. Samuel H. Barnes and Max Kaase. Beverly Hills: Sage Publications.

Kotler, Philip. 1975. Strategic Marketing for Non-Proﬁt Organization. Englewood Cliﬀs: Prentice Hall.

Kroh, Martin. 2009. “The preadult origins of postmaterialism: A longitudinal sibling study.” European Journal of Political Research 48(5):598–621.

Krosnick, Jon A. 1990. “Government Policy and Citizen Passion: A Study of Issue Publics in Contemporary America.” Political Behavior 12:59–92.

Krosnick, Jon A. 1991. “Response Strategies for Coping with the Cognitive Demands of Attitude Measures in Surveys.” Applied Cognitive Psychology 5:213–236.

213 REFERENCES

Kuklinski, James H. and Buddy Peyton. 2007. Belief Systems and Political Decision Making. In The Oxford Handbook of Political Behavior, ed. Russell J. Dalton and Hans- Dieter Klingemann. Oxford: Oxford University Press pp. 45–64.

Kuklinski, James H. and Darrell M. West. 1981. “Economic expectations and voting behavior in United States House and Senate Elections.” American Political Science Review 29(2):436–447.

Kuklinski, James H. and Gary M. Segura. 1995. “Endogeneity, Exogeneity, Time, and Space in Political Representation: A Review Article.” Legislative Studies Quarterly 20(1):3–21.

Kuklinski, James H. and Paul J. Quirk. 2000. Reconsidering the Rational Public: Cog- nition, Heuristics, and Mass Opinion. In Elements of Reason: Understanding and Ex- panding the Limits of Political Rationality, ed. Arthur Lupia, Mathew D. McCubbins and Samuel L. Popkin. London: Cambridge University Press pp. 153–182.

Kwak, Nojin. 1999. “Revisiting the Knowledge Gap Hypothesis: Education, Motivation, and Media Use.” Communication Research 26(4):385–413.

Ladd, Everett Carll. 1995. “The 1994 Congressional Elections: The Postindustrial Re- alignment Continues.” Political Science Quarterly 110(1):1–23.

Lane, Robert E. 1962. Political Ideology: Why the American Common Man Believes What He Does. New-York: Free Press.

Laufer, Romain and Catherine Paradeise. 1990. Marketing Democracy: Public Opinion and Media Formation in Democratic Societies. New-Bruswick and London: Transaction Publ.

Laycock, David H. 2002. The New Right and Democracy in Canada: Understanding Reform and the Canadian Alliance. Oxford: Oxford University Press.

Layman, Geoﬀrey C., Thomas M. Carsey, John C. Green, Richard Herrera and Ros- alyn Cooperman. 2010. “Activists and Conﬂict Extension in American Party Politics.” American Political Science Review 104(2):324–346.

214 REFERENCES

Lazarsfeld, Paul N., Bernard R. Berelson and Helen Gaudet. 1948. The People’s Choice: How the Voter Makes Up Their Mind in a Presidential Campaign. New-York: Columbia University Press.

Leduc, Lawrence, Harold D. Clarke, Jane Jenson and Jon H. Pammett. 1984. “Partisan Instability in Canada: Evidence from a New Panel Study.” American Political Science Review 78(2):470–484.

Leege, David C., Kenneth D. Wald, Brian S. Krueger and Paul D. Mueller. 2002. The Politics of Cultural Diﬀerences: Social Change and Voter Mobilization Strategies in the Post-New Deal Period. Princeton: Princeton University Press.

Lees-Marshment, Jennifer. 2001. “The Marriage of Politics and Marketing.” Political Studies 49(4):692–713.

Lees-Marshment, Jennifer. 2004. The Political Marketing Revolution: transforming the government of the UK. Manchester: Manchester University Press.

Lees-Marshment, Jennifer. 2006. “Political Marketing Theory and Practice: A Reply to Ormrod’s Critique of the Lees-Marshment Market-Oriented Party Model.” Politics 26(2):119–125.

Leighley, Jan E. 2001. Strength in Numbers? Princeton: Princeton University Press.

Lewis-Beck, Michael. 1988. Economics and Elections: The Major Western Democracies. Ann Arbor: University of Michigan Press.

Lewis-Beck, Michael and Martin Paldam. 2000. “Economic voting: an introduction.” Electoral Studies 19:113–121.

Lewis-Beck, Michael and Mary Stegmaier. 2009. “American voter to economic voter: Evolution of an idea.” Electoral Studies 28:625–631.

Lewis-Beck, Michael, William G. Jacoby, Helmut Norpoth and Herbert F. Weisberg. 2008. The American Voter Revisited. Ann Arbor: University of Michigan Press.

Lipset, Seymour Martin and Stein Rokkan. 1967. Cleavage Structures, Party Systems, and Voter Alignments. In Party Systems and Voter Alignments, ed. Seymour Martin Lipset and Stein Rokkan. New-York: Free Press pp. 1–64.

215 REFERENCES

Lock, Andrew and Phil Harris. 1996. “Political marketing - vive la diﬀ´erence.” European Journal of Marketing 30(10/11):14–24.

Loney, Martin. 1998. The Pursuit of Division: Race, Gender, and Preferential Hiring in Canada. Montreal: McGill-Queen’s University Press.

Lupia, Arthur. 1994. “Shortcuts Versus Encyclopedias: Information and Voting Behavior in California Insurance Reform Elections.” American Political Science Review 88:63–76.

Lupia, Arthur. 1998. The Democratic Dilemma: Can Citizens Learn What they Need to Know? New-York: Cambridge University Press.

Luskin, Robert C. 1987. “Measuring political sophistication.” American Journal of Po- litical Science 31(4):856–899.

Luskin, Robert C. 1990. “Explaining Political Sophistication.” Political Behavior 12(4):331–361.

Luskin, Robert C., John P. McIver and Edward G. Carmines. 1989. “Issues and the Transmission of Partisanship.” American Journal of Political Science 33(2):440–458.

Lusztig, Michael and J. Matthew Wilson. 2008. “A New Right? Moral Isssues and Partisan Change in Canada.” Social Science Quarterly 86(1):109–128.

MacKuen, Michael B., Robert S. Erikson and James A. Stimson. 1989. “Macropartisan- ship.” American Political Science Review 83(4):1125–1142.

Mallory, James. 1954. Social Credit and the Federal Power in Canada. Toronto: University of Toronto Press.

Marbut, Robert G. 2004. ¿Un Nuevo Dia? Republican Outreach to the Latino Community in the 2000 Campaign. In Muted Voices: Latinos and the 2000 Elections, ed. Rodolfo O. de la Garza and Louis DeSipio. Lanham: Rowman and Littleﬁeld pp. 61–83.

Margolis, Michael. 1977. “From Confusion to Confusion: Issues and the American Voter (1956-1972).” American Political Science Review 71(1):31–43.

Marland, Alex, Thierry Giasson and Jennifer Lees-Marshment. 2012. Political Marketing in Canada. Vancouver: UBC Press.

216 REFERENCES

Martin, Lawrence. 2010. Harperland: The Politics of Control. The Viking Press.

Maslow, Abraham H. 1954. Motivation and Personality. New York: Harper.

Maxwell, Scott. E. and Harold D. Delaney. 2004. Designing Experiments and Analyzing Data: A Model Comparison Perspective. Second ed. Mahwah, N.J.: Lawrence Erlbaum Associates.

McAllister, Ian. 2002. Calculating or capricious? The new politics of late deciding voters. In Do Political Campaigns Matter? Campaign eﬀects in elections and referendums, ed. David M. Farrell and R¨udigerSchmitt-Beck. ECPR Studies in European Political Science London: Routledge pp. 22–40.

McClosky, Herbert and John Zaller. 1984. The American Ethos: Public attitudes toward capitalism and democracy. Cambridge: Harvard University Press.

McCombs, Maxwell E. and Donald L. Shaw. 1972. “The agenda-setting function of mass media.” Public Opinion Quarterly 36:176–187.

McGinnis, Joe. 1968. The Selling of the President 1968. London: Andre Deutsch Limited.

McIver, John P. and Edward G. Carmines. 1981. Unidimensional scaling. Vol. no. 07-024 of Sage university papers series. Quantitative applications in the social sciences Beverly Hills: Sage Publications.

McLaughlin, Eugene, John Mungie and Gordon Hughes. 2001. “The Permanent Rev- olution: New Labour, New Public Management and the Modernization of Criminal Justice.” Criminology and Criminal Justice 1(3):301–318.

McLeish, Barry J. 1995. Successful Marketing Strategies For Nonproﬁt Organizations. Wiley Nonproﬁt Law, Finance and Management Series Wiley.

McLeod, Jack M., Lee B. Becker and James E. Byrnes. 1974. “Another look at the agenda-setting function of the press.” Communication Research 1:131–166.

Meisel, John. 1975. Working papers on Canadian politics. Montreal: McGill-Queen’s University Press.

217 REFERENCES

Mendelsohn, Matthew. 1996. “The Media and Interpersonal Communications: The Prim- ing of Issues, Leaders, and Party Identiﬁcation.” Journal of Politics 58:112–125.

Merrill, Samuel and Bernard Grofman. 1999. A Uniﬁed Theory of Voting: Directional and Proximity Spatial Models. Cambridge: Cambridge University Press.

Miller, Arthur H. and Warren E. Miller. 1975. “Issues, Candidates and Partisan Divisions in the 1972 American Presidential Election.” British Journal of Political Science 5:393– 434.

Miller, Joanne and David A. Peterson. 2004. “Theoretical and Empirical Implications of Attitude Strength.” Journal of Politics 66(3):847–867.

Miller, Warren E. 1991. “Party Identiﬁcation Realignment and Party Voting: Back to Basics.” American Political Science Review 85:557–568.

Miller, Warren and Merrill Shanks. 1996. The New American Voter. Cambridge: Harvard University Press.

Minsky, Arthur. 1975. A Framework for Representing Knowledge. In The Psychology of Computer Vision. New York: McGraw-Hill.

Mutz, Diana C. 1994. “Contextualizing personal experience: the role of mass media.” Journal of Politics 56(3):689–714.

Mutz, Diana C. 1998. Impersonal Inﬂuence: How Perceptions of Mass Collectives Aﬀect Political Attitudes. New-York: Cambridge University Press.

Nadeau, Richard, Eric´ B´elanger and Fran¸coisP´etry. 2010. “Issue-Based Strategies in Election Campaigns: The Case of Healthcare in the 2000 Canadian Federal Election.” Political Communication 27(4):367–388.

Nadeau, Richard and Michael S. Lewis-Beck. 2001. “National Economic Voting in U.S. Presidential Elections.” Journal of Politics 63:159–181.

Nadeau, Richard, Neil Nevitte, Elisabeth Gidengil and Andr´eBlais. 2008. “Election Cam- paigns as Information Campaigns: Who Learns What and Does it Matter?” Political Communication (25):229–248.

218 REFERENCES

Neuman, W. Russell. 1986. The Paradox of Mass Politics: Knowledge and Opinion in the American Electorate. Cambridge: Harvard University Press.

Neundorf, Anja, Daniel Stegmueller and Thomas J. Scotto. 2011. “The Individual Level Dynamics of Bounded Partisanship.” Public Opinion Quarterly 75(3):458–482.

Nevitte, Neil. 1996. The Decline of Deference: Canadian Value Change in Cross-National Perspective. Peterborough: Broadview Press.

Nevitte, Neil, Andr´eBlais, Elisabeth Gidengil and Richard Nadeau. 2000. Unsteady State: The 1997 Canadian Federal Election. Toronto: Oxford University Press.

Newman, Bruce I. 1994. The Marketing of the President: Political Marketing as Campaign Strategy. Thousand Oaks: Sage Publications.

Newman, Bruce I. 1999a. The Mass Marketing of Politics: Democracy in the Age of Manufactured Images. Thousand Oaks: Sage Publications.

Newman, Bruce I., ed. 1999b. Handbook of Political Marketing. Thousand Oaks: Sage Publications.

Nie, Norman H., Jane Junn and Kenneth Stehlik-Barry. 1996. Education and democratic citizenship in America. Chicago: University of Chicago 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.

Norris, Pippa, John Curtice, David Sanders, Margaret Scammel and Holli Semetko. 1999. On Message: Communicating the Campaign. London: Sage Publications.

O’Cass, Aron. 1996. “Political marketing and the marketing concept.” European Journal of Marketing 30(10/11):37–53.

Oliver, J. Eric and Shang E. Ha. 2007. “Vote Choice in Suburban Elections.” American Political Science Review 3:393–408.

Olsen, Jonathan. 1999. Nature and nationalism: Right-wing ecology and the politics of identity in contemporary Germany. New York: Palgrave Macmillan.

219 REFERENCES

O’Neill, Brenda. 2002. Sugar and Spice? Political culture and the political behaviour of Canadian women. In Citizen Politics: Research and Theory in Canadian Political Behaviour, ed. Joanna Everitt and Brenda O’Neill. Toronto: OUP.

O’Shaughnessy, Nicholas J. 1990. The Phenomenon of Political Marketing. Basingstoke: Macmillan.

O’Shaughnessy, Nicholas J. 2002. The Marketing of Political Marketing. In The Idea of Political Marketing, ed. Nicholas J. O’Shaughnessy and Stephen C. M. Henneberg. Westport, CT: Praeger pp. 209–220.

Page, Benjamin I. and Calvin C. Jones. 1979. “Reciprocal Eﬀects of Policy Preferences, Party Loyalties and the Vote.” American Political Science Review 73(4):1071–1089.

Page, Benjamin I. and Richard A. Brody. 1972. “Policy Voting and the Electoral Process: The Vietnam War Issue.” American Political Science Review 66(3):979–995.

Page, Benjamin I. and Robert Shapiro. 1992. The Rational Public: Fifty Years of Trends in American’s Policy Preference. Chicago: University of Chicago Press.

Par´e,Daniel J. and Flavia Berger. 2008. “Political Marketing Canadian Style?” Canadian Journal of Communication 33:39–63.

Parry, Janine, Jay Barth, Martha Kropf and E. Terrence Jones. 2008. “Mobilizing the Seldom Voter: Campaign Contact and Eﬀects in High-Proﬁle Elections.” Political Be- havior 30(1):97–113.

Payton, Laura. 2012. “Ethnic riding targeting key to Conservatives’ 2011 victory: More than 30 ’very ethnic’ target ridings led to majority government in 2011.” CBC News.

Peﬄey, Mark and Jon Hurwitz. 2007. “Persuasion and Resistance: Race and the Death Penalty in America.” American Journal of Political Science 51(4):996–1012.

Peﬄey, Mark and Robert Rohrschneider. 2007. Elite Beliefs and the Theory of Democratic Elitism. In Oxford Handbook of Political Behavior, ed. Russell J. Dalton and Hans- Dieter Klingemann. New York: Oxford University Press pp. 65–79.

220 REFERENCES

Perloﬀ, Richard M. 1999. Elite, Popular, and Merchandised Politics: Historical Origins of Presidential Campaign Marketing. In Handbook of Political Marketing, ed. Bruce I. Newman. Thousand Oaks: Sage Publications pp. 19–40.

Petrocik, John R. 1996. “Issue Ownership in Presidential Elections, with a 1980 Case Study.” American Journal of Political Science 40(3):825–850.

Philpot, Tasha S. 2008. Race, Republicans, and the Return of the Party of Lincoln. Ann Arbor: Michigan University Press.

Pomper, Gerald M. 1996. Alive! The Political Parties after the 1980-1992 Presidential Elections. In American Presidential Election: Process, Policy, and Political Change, ed. Harvey L. Schantz. Albany: State University of New York Press pp. 135–156.

Popkin, Samuel L. 1991. The Reasoning Voter: Communication and Persuasion in Pres- idential Campaigns. 2nd ed. Chicago: University of Chicago Press.

Powell, Jr., G. Bingham and Guy D. Whitten. 1993. “A Cross-National Analysis of Eco- nomic Voting: Taking Account of the Political Context.” American Journal of Political Science 37(2):391–414.

Prior, Markus. 2007. Post-Broadcast Democracy: How Media Choice Increases Inequality in Political Involvement and Polarizes Elections. New-York: Cambridge University Press.

Rabinowitz, George, James W. Prothro and William Jacoby. 1982. “Salience as a factor in the impact of issues on candidate evaluation.” Journal of Politics 44:41–63.

Rabinowitz, George and Stuart Elaine Macdonald. 1989. “A Directional Theory of Issue Voting.” American Political Science Review 83(1):93–121.

RePass, David E. 1971. “Issue Salience and Party Choice.” American Political Science Review 65(2):389–400.

Riker, William H. 1962. The Theory of Political Coalitions. New Haven: Yale University Press.

Rivers, Douglas. 1988. “Heterogeneity in Models of Electoral Choice.” American Journal of Political Science 32:737–757.

221 REFERENCES

Rokeach, Milton. 1968. “The Role of Values in Public Opinion Research.” Public Opinion Quarterly 32(4):547–559.

Roy, Jason. 2009. “Voter Heterogeneity: Informational Diﬀerences and Voting.” Canadian Journal of Political Science 42(1):117–137.

Sandy, Carson J., Samuel D. Gosling and John Durant. 2013. “Predicting Consumer Behavior and Media Preferences: The Comparative Validity of Personality Traits and Demographic Variables.” Psychology and Marketing 30(11):937–949.

Saris, Willem E. and Paul M. Sniderman, eds. 2004. Studies in Public Opinion: Attitudes, Nonattitudes, Measurement Error and Change. Princeton: Princeton University Press.

S¨arlvik,Bo and Ivor Crewe. 1983. Decade of Dealignment: The Conservative Victory of 1979 and Electoral Trends in the 1970s. Cambridge: Cambridge University Press.

Sartori, Giovanni. 1970. “Concept Misformation in Comparative Politics.” American Political Science Review 64(4):1033–1053.

Savigny, Heather. 2004. “Political Marketing: A Rational Choice?” Journal of Political Marketing 3(1):21–38.

Scammell, Margaret. 1999. “Political Marketing: Lessons for Political Science.” Political Studies 47(4):718–739.

Schantz, Harvey L. 1996a. Sectionalism in Presidential Elections. In American Presiden- tial Election: Process, Policy, and Political Change, ed. Harvey L. Schantz. Albany: State University of New York Press pp. 93–134.

Schantz, Harvey L., ed. 1996b. American Presidential Election: Process, Policy, and Political Change. Albany: State University of New York Press.

Schattschneider, Elmer E. 1960. The Semisovereign People: A Realist’s View of Democ- racy in America. New-York: Holt, Rinehart, and Winston.

Scheufele, Dietram A. 2000. “Agenda-setting, priming, and framing revisited: another look at cognitive eﬀects of political communication.” Mass Communication & Society 3(2/3):297–316.

222 REFERENCES

Schlesinger, Joseph A. 1991. Political Parties and the Winning of Oﬃce. Ann Arbor: University of Michigan Press.

Schmitt-Beck, Rüdigerand David M. Farrell. 2002. Studying political campaigns and their effects. In Do Political Campaigns Matter? Campaign effects in elections and referendums, ed. David M. Farrell and RüdigerSchmitt-Beck. ECPR Studies in European Political Science London: Routledge pp. 1–21.

Schmitt, Neal. 1996. “Uses and Abuses of Coeﬃcient Alpha.” Psychological Assessment 8(4):350–353.

Schrodt, Philip A. 2010. “Seven Deadly Sins of Contemporary Quantitative Political Analysis.” Paper presented at the Annual Meetings of the American Political Science.

Siegfried, Andr´e.1907. The Race Question in Canada. London: E. Nash.

Sniderman, Paul M., Richard A. Brody and Phillip E. Tetlock. 1991. Reasoning and Choice: Explorations in Political Psychology. New-York: Cambridge University Press.

Soroka, Stuart. 2003. Agenda-setting dynamics in Canada. Vancouver: UBC Press.

Srebrnik, Henry. 2014. “Harper, Israel and Canada’s Jewish Community.” The Guardian. (January 8):A7.

Steger, Wayne P., Sean Q. Kelly and J. Mark Wrighton. 2006. “Campaigns and Marketing in Political Science Context.” Journal of Political Marketing 5(1/2):1–10.

Stewart, Ian. 2002. Vanishing Points: Three Paradoxes of Political Culture Research. In Citizen Politics: Research and Theory in Canadian Political Behaviour, ed. Joanna Everitt and Brenda O’Neill. Toronto: OUP pp. 21–39.

Stimson, James A. 1975. “Belief Systems: Constraint, Complexity, and the 1972 Elec- tion.” American Journal of Political Science 20:393–417.

Stimson, James A. 1999. Public Opinion In America: Moods, Cycles, And Swings. Second ed. Boulder: Westview Press.

Stimson, James A. 2004. Tides of Consent: How Public Opinion Shapes American Poli- tics. Cambridge: Cambridge University Press.

223 REFERENCES

Stokes, Donald E. 1963. “Spatial Models of Party Competition.” American Political Science Review 57:368–377.

Stokes, Donald E. 1992. Valence Politics. In Electoral Politics, ed. Dennis Kavanagh. Clarendon Press pp. 141–164.

Strøm, Kaare. 1990. “A Behavioral Theory of Competitive Political Parties.” American Journal of Political Science 34(2):565–598.

Stroud, Natalie Jomini. 2008a. “Media Eﬀects, Selective Exposure, and Fahrenheit 9/11.” Political Communication 24:415–432.

Stroud, Natalie Jomini. 2008b. “Media Use and Political Predispositions: Revisiting the Concept of Selective Exposure.” Political Behavior 30(3):341–366.

Taber, Charles S. and Milton Lodge. 2006. “Motivated Skepticism in the Evaluation of Political Beliefs.” American Journal of Political Science 50(3):755–769.

Taylor, Marylee C. 1983. “The Black-and-White Model of Attitude Stability: A Latent Class Examination of Opinion and Nonopinion in the American Public.” American Journal of Sociology 89(2):373–401.

Tichenor, Daniel J. 2002. Dividing Lines: The politics of immigration control in America. Princeton: Princeton University Press.

Tourangeau, Roger, Lance J. Rips and Kenneth Rasinski. 2000. The Psychology of Survey Response. Cambridge: Cambridge University Press.

Uhlaner, Carole J. 1989. “Rational turnout: the neglected role of groups.” American Journal of Political Science 33:390–442. van der Brug, Wouter. 2004. “Issue ownership and party choice.” Electoral Studies 23(2):209–233. van der Brug, Wouter, Cees van der Eijk and Mark N. Franklin. 2007. The Economy and the Vote: Economic Conditions and Elections in Fifteen Countries. Cambridge: Cambridge University Press.

224 REFERENCES van Deth, Jan W. and Elinor Scarborough, eds. 1995. The Impact of Values. Oxford: Oxford University Press.

Vavreck, Lynn. 2009. The Message Matters: The Economy and Presidential Campaigns. Princeton: Princeton University Press.

Wang, Wei, David Rothschild, Sharad Goel and Andrew Gelman. 2014. “Forecasting Elections with Non-Representative Polls.” International Journal of Forecasting Forth- coming.

Watt, Jaime. 2006. “Rebuilding the Mulroney Coalition - Welcome the “New Harper Conservatives”.” Policy Options/Options politiques pp. 41–46.

Wellhofer, E. Spencer. 1990. “Contradictions in Market Models of Politics: the Case of Party Strategies and Voter Linkages.” European Journal of Political Research 18(1):9– 28.

Wells, Paul. 2006. Right Side Up: The Fall of Paul Martin and the Rise of Stephen Harper’s New Conservatism. Toronto: McClelland & Stewart.

West, Darrell M. 2013. Air Wars: Television Advertising and Social Media in Election Campaigns, 1952-2012. Washington DC: CQ Press.

White, Jon and Leslie de Chernatony. 2002. “New Labour: A study of the creation, development and demise of a political brand.” Journal of Political Marketing 1(2/3):45– 52.

White, Stephen, Neil Nevitte, Andr´eBlais, Elisabeth Gidengil and Patrick Fournier. 2008. “The Political Resocialization of Immigrants: Resistance or Lifelong Learning?” Political Research Quarterly 61(2):268–281.

Wiseman, Nelson. 1986. “The Use, Misuse, and Abuse of the National Election Studies.” Journal of Canadian Studies 21:21.

Wiseman, Nelson. 2007. In Search of Canadian Political Culture. Vancouver: UBC Press.

Wlezien, Christopher. 2005. “On the salience of political issues: The problem with the ‘most important problem’.” Electoral Studies 24:555–579.

225 REFERENCES

Wong, Janelle S. 2000. “The Eﬀect of Age and Political Exposure on the Development of Party Identiﬁcation Among Asian American and Latino Immigrants in the United States.” Political Behavior 22(4):341–371.

Worcester, Robert M. and Paul R. Baines. 2006. Voter Research and Market Positioning: Triangulation and Its Implication for Policy Development. In Winning Elections With Political Marketing, ed. Philip Davies and Bruce I. Newman. Binghamton: Harworth Press pp. 11–32.

Zaller, John. 1990. “Political awareness, elite opinion leadership, and the mass survey response.” Social Cognition 8:125–153.

Zaller, John. 1992. The Nature and Origins of Mass Opinion. New-York: Cambridge University Press.

Zaller, John R. 1998. “Monica Lewinsky’s Contribution to Political Science.” PS: Political Science & Politics 31(2):182–189.

226