Culture, Constraint and Social Learning

There are two prevailing ways of thinking about the lines of ideational division in Canada. The first strategy is a broad ecological approach rooted in the concept of culture (Bell, Reference Bell1992; Horowitz, Reference Horowitz1966). The challenge from this perspective is to identify and explain aggregate-level differences between groups. The focus is on the content of opinions, that is, what people think about the issues. Lipset (Reference Lipset1986: 136–38), for example, contrasts the “achievement orientation,” “egalitarianism” and “individualism” of Americans with the “ascription,” “elitism” and “group orientation” of Canadians. Wiseman (Reference Wiseman and Dunn1996; Reference Wiseman2007), similarly, sets out to explain why Albertans are more individualistic and religious than their counterparts in the rest of the country. In these cases, the analytical vantage point is the group level. Lipset (Reference Lipset1986) says nothing about whether the group-oriented Canadians are also the elitist ones, or whether the achievement-oriented Americans are also the egalitarian ones. Wiseman (Reference Wiseman and Dunn1996; Reference Wiseman2007) does not ask whether religious Albertans are especially individualistic or whether individualistic Albertans are at all religious. In short, group-level analyses underscore the differences between groups across multiple dimensions of political thought, but they do not examine the belief systems of the individuals within those groups.

The second strategy focuses on the structure of public opinion, that is, how people organize their policy preferences about multiple issues. The unit of analysis is mixed. The core concept is “constraint” (Converse, Reference Converse and Apter1964: 252). From here, researchers search via statistical induction for patterns of relationships between variables. How these variables cluster most efficiently into separate “dimensions” of opinion is said to illuminate the underlying factors that constrain them. Laponce (Reference Laponce1981: 196), for example, discovers three dimensions of opinion: a “religious” dimension, an “economic” dimension and a “hierarchical group centrism” dimension.Footnote ³ Nevitte and colleagues (Reference Nevitte, Blais, Gidengil and Nadeau2000: 46–47) uncover similar dimensions: “cynicism,” “moral traditionalism,” “free enterprise” and “out-groups.” Cross and Young (Reference Cross and Young2002: 869) find two strong dimensions, “social tolerance” and “laissez-faire economics,” and two weak dimensions, “provincial powers” and “populism.” This approach is inductive.Footnote ⁴ Laponce (Reference Laponce1981) does not say why there are three dimensions of opinion, rather than one or two or 26 dimensions. And Cross and Young (Reference Cross and Young2002) do not explain why opinions about a distinct society clause for Quebec load more strongly on their “social tolerance” dimension than on their “provincial powers” dimension. In these cases, the objective is to simplify and describe, rather than explain, the contours of public opinion. Thus, individual opinions are allowed to vary, but only along those dimensions that most efficiently describe the structure of public opinion for the group as a whole. In short, the content of public opinion varies at the individual level, but the structure of public opinion varies at the aggregate level.

In this article, a third perspective is explored. The unit of analysis is the individual. The core concept is “social learning” (McClosky and Zaller, Reference McClosky and Zaller1984: 12). In its broadest sense, social learning is the notion that political opinions form through interactions of individual-level factors like personality, religiosity, partisanship and rationality, and social factors like family upbringing, religion, party membership and socioeconomic class (for example, Alford et al., Reference Alford, Funk and Hibbing2005). Many of these influences generate distinctive bundles of opinions for individuals and groups by affecting simultaneously more than one opinion. From this standpoint, the content and structure of opinions are congenitally entwined. A religion that proscribes homosexuality and abortion affects the content of opinions by generating right-wing opinions about each of these issues. And it affects the structure of opinions by linking these issues together as precepts in a single doctrine.

Thinking about opinions in this way, as intervening variables, reshapes the prevailing template of ideational disagreement in two respects. First, conceptualizing opinion formation as an active process means, for any issue, that a “non-opinion” is the default position. Diametrically opposing levels of exposure to a single influence do not generate opposing opinions about the same issues. Non-exposure has no effect on opinions, rather than an equal and opposite effect on the same range of opinions. Thus, opposing thoughts about precisely the same issue stem from altogether different sources, rather than from different levels of exposure to the same source.

Second, tracing the content and structure of opinions to common origins means that people who differ in the substance of their opinions are likely to differ in the organization of their opinions. How people think about a political issue is likely to affect in important respects how they situate that issue with other elements of the political universe. The homogenizing assumption inherent in the prevailing methodology of statistical induction—exploratory factor analysis of national or regional samples—does not square with the view that the structure of public opinion is as contingent and variable as the content of public opinion.

In sum, there are different ways to look at ideological disagreement. We may look at it, in a cultural sense, as differences between groups in the content of opinions about multiple issues, thereby imputing to the opinion sets of individuals the issue-by-issue positions that distinguish their country, region, province or linguistic group from other countries, regions, provinces or linguistic groups. Alternatively, we may look at ideological disagreement in a structural sense by narrowing the level of analysis and ploughing inductively for underlying patterns between variables. And finally, we may look at ideological disagreement from the standpoint of individual predispositions interacting with the social environment to produce differences between people in the origins, organization and content of their opinions. Certainly, the latter of these approaches is the most demanding theoretically insofar as it requires that researchers conjecture in advance about potential sources of opinion. It is also more rewarding theoretically because it covers simultaneously the content and structure of opinions. But does it describe the empirical world more accurately? And does it better explain the Canadian political environment? It is to these questions that the analysis now turns.

The Structure and Content of Left-Right Disagreement

Richard Johnston (Reference Johnston and Perlin1988: 59) observes that “the very idea of ideology presupposes a hierarchy of ideas.” Ideologies emerge from core ideas that people consult when they develop their opinions about issues. One of these ideas, certainly, is the prescriptive belief in human equality (Bobbio, Reference Bobbio1996). Egalitarianism underwrites an assortment of left-wing opinions for a large number of people, including opinions about economic equality, racial equality, gender equality and the equality of gays and lesbians (Noël and Thérien, Reference Noël and Théren2008; Matthews, Reference Matthews2005).

Other beliefs are also far-reaching. A belief in the divinity of scripture engenders, in many cases, right-wing opinions about abortion and homosexuality (Laponce, Reference Laponce1981). A belief in the efficiency of free-market capitalism begets right-wing opinions about tax policy, government regulation and welfare spending (Blais et al., Reference Blais, Gidengil, Nadeau and Nevitte2002; Nevitte et al., Reference Nevitte, Blais, Gidengil and Nadeau2000). And a belief in the superiority of one's own group may well breed, by extension, negative opinions about people who are different, including, typically, gays, lesbians, racial minorities, and immigrants (Laponce, Reference Laponce1981).

It is well nigh impossible, of course, to list every foundational belief, let alone the potentially creative and idiosyncratic ways that people apply them to their political environments. But the inability to see to the edges of the ideological universe does not mean that the big ideas at the centre have to be discarded. On this point, instruments tuned to the nexus of political disagreement consistently detect the telltale signs of equality, religion, capitalism and intolerance (Benoit and Laver, Reference Benoit and Laver2006; Budge et al., Reference Budge, Klingemann, Volkens, Bara and Tanenbaum2001; Klingemann et al., Reference Klingemann, Volkens, Bara, Budge and McDonald2006). These beliefs press on opinions about some of the same issues. But each belief also bears on opinions about distinctive configurations of issues. The dominant religious traditions in Western countries, for example, proscribe homosexuality and abortion but say little to nothing about tax policy and government spending initiatives. More generally, there is certainly no guarantee that the structuring effects of different underlying influences extend to cover the same range of issues. The central hypothesis to be explored is straightforward. Just as there are differences in the foundational ideas that underpin left-wing and right-wing opinions, so too are there likely to be differences between the left and right in the ways that people lump political issues together into ideologically coherent bundles of ideas about the political world.

The empirical analysis proceeds in two stages. The first part establishes the policy space of the political parties themselves. This stage of the analysis turns, first, to Benoit and Laver's survey of experts (Reference Benoit and Laver2006) about the policy positions of political parties, and, second, to content analyses of party platforms from the Comparative Manifesto Research Project (CMRP) (Budge et al., Reference Budge, Klingemann, Volkens, Bara and Tanenbaum2001; Klingemann et al., Reference Klingemann, Volkens, Bara, Budge and McDonald2006). Benoit and Laver (Reference Benoit and Laver2006) surveyed national political experts, including 104 experts from Canada, and asked each of them to locate the positions of their country's parties on a common battery of policy dimensions. The CMRP examines systematically the content of the election platforms proposed by political parties in democratic countries, including those proposed by Canadian parties in 18 federal elections between 1945 and 2000. These data are particularly useful in research designs, such as this one, where it is necessary to treat the policies of a political party as potentially different than the opinions of the party's supporters in the electorate.

The second stage of the analysis turns to data from the Study of Canadian Political Party Members (SCPPM). The SCPPM is a regionally stratified random mail-back survey of 3872 card-carrying members of five Canadian political parties: Liberal (L), Progressive Conservative (PC), Canadian Alliance (CA), New Democrat (NDP), and Bloc Quebecois (BQ).Footnote ⁵ The survey, conducted in 2000, provides a snapshot of partisan opinions at a moment in Canadian history when the traditional brokerage parties, the Liberals and the PCs, were flanked on both the left and right by two procrustean ideological parties, the NDP and Canadian Alliance. The SCPPM asks respondents for their opinions about “economic issues,” like wealth redistribution and private health care, and “social issues,” like equal rights, cultural minorities and immigration.Footnote ⁶ These data make it possible to examine the ways that activists on the left and the right bundle their opinions about left/right political issues. Taken together, the empirical analyses converge on the question whether leftists and rightists organize their political preferences in different ways.

Ideology and Canadian Politics

There are discernable lines of ideological division that crisscross the Canadian party landscape. Benoit and Laver (Reference Benoit and Laver2006) asked national experts about the left/right positions of the political parties in their country across a number of policy dimensions. The experts were asked to pinpoint the location of each party on scales that range from extreme left-wing scores of 1 to extreme right-wing scores of 20. According to the Canadian experts, the NDP and the Canadian Alliance were 15 points apart on the social dimension, 13 points apart on the economic dimension, and 9 points apart on the immigration dimension. Indeed, there was a 13-point spread between the NDP and the Alliance on the overall scale of left/right ideology. The size of this ideological gap is impressive. It is greater than the distance between, for example, the Labour and Conservative parties in Britain (5.5. points), or between the Democratic and National parties in Australia (7.1 points), or between the Democrats and Republicans (9.5 points) in the United States. Indeed, in Austria, the Greens and the far-right Freedom Party are separated by 12 points. And 14 points separates the Communist Party in France from the far-right National Front. These findings stand as a forthright challenge to the conventional wisdom that party politics in Canada is quintessentially non-ideological. Even so, it is difficult to compare cross-nationally the judgments of national experts about the left/right positions of the political parties in their respective countries. These finding leave open an important line of questioning: do Canadian parties emit particularly strong ideological signals? Or is it that weak ideological signals stand out more clearly against the ostensibly pragmatic background of Canadian political history?

Evidence from the CMRP addresses precisely this question. The CMRP gathers evidence about the left/right positions of political parties through systematic content analyses of party election platforms. The content analyses rely on a single set of coding criteria to classify line by line the policy content of party manifestos. The results are quantified as the percentage of sentences that each platform devotes to different kinds of arguments. A score of 8 on the “free enterprise” dimension indicates that 8 per cent of the sentences in an election platform support free-market capitalism. A score of 15 on the “welfare state expansion” dimension suggests that 15 per cent of a party platform promotes the expansion of social welfare programs. Taken together, there are 57 policy categories into which content may be coded; 26 of these categories can be linked together in a single scale of left/right.Footnote ⁷ In this way, the CMRP facilitates direct cross-national and cross-time comparisons of the left/right positioning of a host of political parties, including Canadian ones.

Figure 1 summarizes in left/right space the results of the CMRP's content analyses of party manifestos in Canada. The x-axis represents time; the y-axis represents the left/right continuum. Thus, the points in the graph track across time the left/right location of party platforms in each Canadian federal election between 1945 and 2000. From this vantage point, the results in Figure 1 reflect the same broad outline of ideological disagreement that Benoit and Laver pick up in their 2002 survey of Canadian political pundits: the New Democratic party (NDP) is to the left, the Reform-Alliance party (REF) is to the right and the Liberal (L) and Progressive Conservative parties (PC) are to the centre-left and centre-right, respectively. But there is another finding that deserves attention. Notice the cross-time trajectory of Canada's two traditional governing parties, the Liberals (L) and Progressive Conservatives (PC). Until the late 1970s, the Liberals and Conservatives oscillate in left/right space like two pragmatic brokerage parties chasing through time the fleeting concerns of non-ideological voters. Their left/right positions are essentially interchangeable. Indeed, the supposed rightists, the Conservatives, are to the left of the Liberals in 8 of the 12 elections between 1945 and 1979.

Figure 1 Left-Right Position of Party Election Platforms in Canada, 1945–2000

Data Sources: Budge et al., Reference Budge, Klingemann, Volkens, Bara and Tanenbaum2001; Klingemann et al., Reference Klingemann, Volkens, Bara, Budge and McDonald2006

Things changed, however, in the 1980s. As the data in Figure 1 show, the ideological gap between the left and the right widens first during the 1980s, and it widens again in 1993 as the populist Reform party supplants the Progressive Conservatives as the dominant force on the Canadian right. Indeed, the election platforms of the NDP and the Reform party stretched the left/right continuum in Canada to an extent—about 60 points, on average—that exceeds even the polarizing effects of fringe parties in most European countries. Since 1990, for example, the platforms of left-wing and right-wing parties are separated by no more than 70 points in France, 60 points in Germany, 50 points in the Netherlands, 40 points in Britain, 35 points in Australia, and 25 points in the United States. In short, Canada's major political parties were divided ideologically in the latter decades of the twentieth century to an extent that they had not been divided previously. And the new ideological divide is wide by cross-national standards.

Taken together, the evidence suggests that Canada's governing parties used to orbit each other at the non-ideological midpoint of the left/right continuum. That system flew apart in the 1980s. Content analyses of party manifestos and surveys of political experts now detect powerful signals of ideological disagreement in the Canadian case. At first glance, these findings are at odds with the traditional concept of brokerage politics. But that conclusion skirts a conceptual question of whether ideological and brokerage politics are inherently incompatible. The short answer is no. Office-seeking politicians need ideologues in order to win elections; policy-seeking ideologues need sympathetic politicians in power. To be sure, ideology limits the room for brokerage and vice versa (Cross and Young, Reference Cross and Young2004). Policy-seeking activists are unlikely to abandon the mainstays of their ideological agenda in order to win elections. Thus, groups of activists will not fit together in a coalition if they hold irreconcilable and non-negotiable policy preferences on the same issues. But this is consistent with any other kind of brokerage. And people with non-negotiable preferences about altogether different issues may be able to work together quite effectively. For this reason, the structure of policy opinions among left-wing and right-wing activists may well bear in important respects on the political prospects of left-wing and right-wing parties, especially in the ideologically charged atmosphere of the 1990s and beyond.

In short, left/right ideology plays a key role in Canadian politics. Understanding the dynamic of that key role, however, requires an analysis of the ways that leftists and rightists organize their opinions about policy into bundles. It is important to consider, then, whether leftists and rightists structure their opinions about the political world in different ways. The central expectation is that they do.

The Asymmetrical Universes of Left and the Right in Canada

The expectations outlined above posit that leftists and rightists organize their political environments in distinctive ways. More specifically, leftists bundle coherently their economic and social opinions because a pervasive left-wing idea, equality, affects both sets of opinions. The ideological environment is different for those on the right. There is nothing about free-market materialism that begets right-wing opinions about social morality. And there is little about religion that engenders right-wing opinions about taxation and social welfare. Indeed, neither free-market support nor religion is likely to generate right-wing opinions about immigration and racial minorities. When it comes to economic and social issues, in effect, the overall hypothesis is that there is one left and multiple rights.

The Survey of Canadian Political Party Members (SCPPM) is useful for testing these expectations for a number of reasons. First, these data were collected from identical mail-back surveys administered to random samples of respondents drawn from the membership lists of all five political parties represented in Parliament (Cross and Young, Reference Cross and Young2002: 865). Thus, precisely the same questions are used to gauge the opinions of members from different political parties—including the NDP and the Canadian Alliance, the far left and right of Canadian politics, respectively. Second, the survey asks these respondents for their opinions about economic and social issues that are at the core of left/right disagreement. Respondents are asked to weigh spending on social programs against tax-cutting and deficit reduction (SOCPROGS). And they are asked for their opinions about wealth redistribution (INCEQUAL), employment insurance (EI), health care user fees (USERFEES), private versus public sector job creation (PRIVSECTOR), raising tuition (TUITION), private health care (HEALTHCARE), equal rights (EQUALRIGHTS), new lifestyles (NEWLIFESTYLE), special treatment of minorities (MINORITIES), immigration (IMMIGRATION), and bilingualism (BILINGUALISM). With these data it becomes possible to examine directly the ways that left-wing and right-wing activists structure their answers to identical batteries of questions about a wide range of economic and social policy issues.

This stage of the analysis turns to factor analysis to compare the answers of NDP and Alliance members to the battery of social and economic policy questions in the SCPPM. There are two broad families of factor analysis. The most common manifestation of factor analysis in Canadian public opinion literature is drawn from the body of closely related statistical techniques that are known, collectively, as exploratory factor analysis (EFA). These statistical tools enable researchers to reduce the number of variables in their analyses by identifying how co-variation between multiple observed variables can be accounted for by smaller numbers of unobserved, underlying “factors.” Variables that load on a common factor are often combined together to form aggregate indices. In this respect, EFA is a highly effective data reduction strategy.

There are a few reasons why EFA cannot effectively test theoretical expectations about underlying patterns between variables. Not the least of these reasons stems from the so-called “rotational problem.” A factor solution can be transformed, or “rotated,” to another factor solution that fits the empirical data equally well (Harman, Reference Harman1976: 27–28). Thus, EFA is appropriate for data reduction, where researchers can accept whichever factor solution involves the fewest number of factors. But EFA is not generally appropriate for comparing how the same models effectively explain the structure of opinions across groups among the individuals within those groups.

The following analysis therefore turns to a second family of statistical tools known as confirmatory factor analysis (CFA) (see, for example, Albright and Park, Reference Albright and Park2009). CFA requires that researchers specify in advance their theoretical expectations about whether, and how, underlying latent factors give rise to patterns between multiple observed variables. A theoretical model is built and tested for how well it fits the empirical evidence. CFA is particularly suited for testing different theoretical expectations against the same body of evidence and for testing the same theoretical expectations against different bodies of evidence. Both these aspects are helpful in the current case. Indeed, this stage of the analysis tests two CFA models against the answers of party members to the battery of social and economic policy questions in the SCPPM.Footnote ⁸

The first model, a straightforward single-dimensional model, posits that policy opinions are linked together by a single underlying dimension. This model supposes, first, that the people with left-leaning opinions about welfare spending also hold left-leaning opinions about social morality and immigration. And it supposes, second, that the people who hold right-wing opinions about welfare spending also hold right-wing opinions about social morality and immigration. The two-dimensional model, by contrast, divides these policy questions into two separate categories. The first category covers social issues like immigration, newer lifestyles and bilingualism, and the second category asks about such economic issues as wealth redistribution, employment insurance and health care user fees. In effect, the first model proposes that people think about economic and social issues together, and the second model suggests that people assign social and economic issues to different spheres of opinion. The theoretical conjecture about left/right differences in the organization of policy opinions generate the expectation that the single dimensional model will work considerably better when applied to the opinion structure of NDP members than to Alliance members, and that the two-dimensional model will make more sense in the case of Alliance members than in the case of NDP members.

The results of testing the one-dimensional model against the policy opinions of NDP and Alliance members are summarized in Figure 2. The standardized factor loadings, λ, appear in the middle of the inside arrows in the diagrams, and the summary statistics of model fit are underneath each figure. The first finding in Figure 2 is that the one-dimensional model describes more effectively the opinion structure of NDP members than it does the opinion structure of Alliance members. Notice the factor loadings for the model on the left-hand side of the figure. These results indicate that all but two of these questions load for NDP members at .5 or higher on the single underlying dimension proposed in model 1. Even the two exceptions, INCEQUAL (.44) and MINORITIES (.44), are relatively well connected to the underlying dimension.Footnote ⁹ But notice as well that the three strongest loadings for NDP members are EQUALRIGHTS (.74), IMMIGRATION (.71) and SOCPROGS (.66). For Canadians as a whole, these variables have been found to represent three altogether separate values dimensions: social morality (EQUALRIGHTS), out-groups (IMMIGRATION) and economics (SOCPROGS) (Laponce, Reference Laponce1981; Nevitte et al., Reference Nevitte, Blais, Gidengil and Nadeau2000). In the case of NDP members, by contrast, these variables load together atop a single dimension. For NDP members, it seems, there is an intimate connection between their opinions about social and economic issues.

Figure 2 The Opinion Structure of NDP and Alliance Members in One Dimension

Data Source: Lisa Young and William Cross, 2000.

The findings are quite different when it comes to Alliance members. These results are summarized on the right-hand side of Figure 2. Notice, first, that there is clear evidence of a pattern in the policy opinions of Alliance members. Their opinions about HEALTHCARE (.61), TUITION (.59), USERFEES (.58), INCEQUAL (.56), PRIVSECTOR (.56), EI (.50) and SOCPROGS (.46) are all bound together by their common connection to a single underlying dimension. The dimension is decidedly economic. Indeed, the second finding is that the social morality of Alliance members is weakly if at all connected to their economic opinions. The views of Alliance members about EQUALRIGHTS (.37), BILINGUALISM (.34) and MINORITIES (.31) are only weakly connected to the single underlying dimension in Figure 2. And their opinions about NEWLIFESTYLE (.18) and IMMIGRATION (−.09) are not at all connected to this underlying dimension. These findings are consistent with the hypothesis that Alliance members organize their opinions about economic issues around the concept of the “free market.” As it turns out, however, Alliance members do not bundle together their economic opinions with their social opinions.

To this point, the results highlight fundamental differences in the patterns of opinions among NDP members, on the one hand, and Alliance members on the other. Figure 3 illustrates the contours of these differences more precisely. In this case, the model posits a two-dimensional picture of left/right opinion structure. The first dimension captures opinions about economic issues, and the second dimension brings together opinions about social issues. In this light, the first finding is that two-dimensional model fits quite well the opinion structure of NDP and Alliance members. Note how economic opinions cluster together for NDP and Alliance supporters, and so, for the most part, do social opinions. But notice, first, that the correlation between the economic and social factors is considerably higher in the case of NDP members (.73) than in the case of Alliance members (.35). Notice also that IMMIGRATION (.76) loads strongly on the social dimension for NDP supporters, but not at all (.09) for Alliance supporters. In short, the evidence suggests again that social and economic opinions are more tightly intertwined among NDP members than among Alliance members. Indeed, opinions about immigration are a central feature of the social dimension for NDP members, but they do not fit at all on the social dimension for Alliance members.

Figure 3 The Opinion Structure of NDP and Alliance Members in Two Dimensions

Notes:(1) Results are standardized solutions from confirmatory factor analysis, using robust diagonally weighted least squares estimation (Jöreskog, Reference Jöreskog1990). (2) Missing data imputed using LISREL's expectation maximum (EM) algorithm for multiple imputation; (3) Number of observations = 606 for NDP; 1036 for Alliance.

Data Source: Lisa Young and William Cross, 2000.

Taken together, the results of the confirmatory factor analyses indicate that there are distinctive patterns of opinion among NDP and Alliance members. These findings are consistent with the view that the origins, content and structure of opinions are fundamentally different on the left than they are on the right. Indeed, Figure 4 juxtaposes the left/right positions of Canadian party platforms in the Comparative Manifesto Research Project (CMRP), with the level of co-variation between the economic and social factors for party members in the SCPPM. The left/right positions of party platforms are plotted along the x-axis. The y-axis represents the correlation between opinions on the economic and social dimensions for respondents in the SCPPM. These latter results are derived from replicating for Bloc Quebecois, Liberal, and PC members the same two-dimensional CFA model outlined above. Notice how the relationship between the economic and social dimension weakens as party positioning moves from the left to the right. Indeed, there is just a slight deviation from this trend for the two parties at the centre, the Liberal and PC parties. Even so, the overall trend is straightforward. Economic and social opinions are bound together more tightly by activists on the left than they are on the right.

Figure 4 The Connection between the Economic and Social Factors among Canadian Political Party Members, by Left/Right Position of Party Manifestos

Notes: (1) Results for party members are standardized solutions from confirmatory factor analysis, using robust diagonally weighted least squares estimation (Jöreskog, Reference Jöreskog1990). (2) Missing data for party members imputed using LISREL's expectation maximum (EM) algorithm for multiple imputation; (3) Number of observations = 606 for NDP; 406 for BQ; 892 for Liberal; 875 for PC; 1036 for Alliance.

Data Sources: Lisa Young and William Cross, 2000; Klingemann et al., Reference Klingemann, Volkens, Bara, Budge and McDonald2006

The evidence presented here is consistent with the argument that a deep-seated commitment to equality brings together the social and economic viewpoints of egalitarians into a coherent bundle of left-wing opinions (Bobbio, Reference Bobbio1996; Noël and Thérien, Reference Noël and Théren2008). The pattern of opinions among Canadian party members, particularly NDP members, is consistent with this argument. But while a commitment to equality shapes the social and economic opinions of leftists, it is not a commitment to inequality that shapes the social and economic opinions of rightists. Rather, right-wing opinions stem largely from free-market support and religion. Yet, while free-market support shapes the opinions of economic conservatives, the concept of the free-market does not extend to the realm of social opinions. And while religion brings together the social opinions of religious conservatives, it does not affect their opinions about economic issues. Moreover, neither religion nor free-market support bears in any way on opinions about immigrants. Consequently, the patterns of left/right disagreement are asymmetrical. The concept of equality reaches across social, economic and immigration dimensions. Religion and free-market support do not.