Federations have to strike a delicate balance between unity and regional diversity. In the Canadian case, the highly decentralized nature of the federal system enables the provinces and regions to exercise considerable autonomy. In this context, policy disagreements have often been interpreted through the prism of regionalism. Some observers of Canadian politics have expressed concerns that this amplifies disagreement and could undermine the legitimacy of national policies and institutions.
The significance of regionalism in Canada has long been debated and studied. In this paper, the authors’ approach is different from that of much literature in this field: they explore whether disagreements stemming from value cleavages are made worse by Canada’s regional nature. Their analysis is based on an original survey that measures the degree to which Canadians are divided along three value inclinations: egalitarianism, traditionalism and legal rigorism (a belief that laws should apply strictly and uniformly to everyone). The survey was conducted in early 2015, in five regions: British Columbia, the three Prairie provinces, Ontario, Quebec, and the four Atlantic provinces. The samples comprised around 1,000 respondents.
The authors carried out detailed empirical analyses of 7 public spending domains, as well as 12 policy issues ranging from restrictions on religious symbols to support for oil pipelines. Their principal finding is that Canadians are first and foremost divided over values, not regions. Disagreements on policy issues exist because of Canadians’ differing values, not because of territorial fracture lines. In other words, the three value inclinations and the conflicts they generate are present in every region.
Moreover, individuals who share the same values, regardless of where they live, have similar policy preferences. For example, whether they live in Quebec or the Prairie provinces, proponents of social justice show similar levels of support for environmental spending.
Based on their findings, the authors present three implications for policy. First, to be successful, policy should be designed, framed and promoted to appeal primarily to values, not regions. Although the composition of values varies among regions, this does not prevent the formation of cross-regional alliances that could foster some level of national acceptance. Second, those who seek countrywide support for national policies will nevertheless sometimes need to adopt regionally sensitive communications strategies, notably to appeal to values that have been overlooked in the past in given regions by policy-makers. Third, provincial policy-makers may find it beneficial to exchange with their counterparts in other provinces when developing policies and strategies.
More broadly, the authors conclude that there may have been a tendency to exaggerate fears that disagreements among Canada’s regions will be harmful to policy acceptance and institutional legitimacy. Regional differences should not be ignored, but more systematic attention to value disagreements within regions seems warranted.
Disagreement over preferred policy options is a normal feature of any functioning democracy (Montpetit 2016). There is, however, a threshold beyond which disagreement becomes harmful, stifling policy development or making it difficult for important segments of the population to accept government decisions. In Canada, these sorts of entrenched policy disagreements are often interpreted through the prism of regionalism. As Simeon and Elkins (1974, 397) argued some time ago, “Canadian politics is regional politics” — an adage that alludes to differences in values and material interests across Canada. For example, Quebecers’ values of solidarity and self-preservation are sometimes thought to be incompatible with the value other Canadians accord to individual rights and freedoms.
Western Canadians also have a history of disagreement with policies of the federal government, which, they claim, have failed to reward their hard work, serving instead a class of rent seekers in central Canada. This apparent regional conflict was illustrated by the election of Stephen Harper as prime minister in 2006. Although many western Canadians celebrated the event as a reassertion of conservative values, Harper’s policies were frequently depicted in the rest of the country as distinctly un-Canadian.
If real regional value-based cleavages exist, their presence may threaten the legitimacy of federal institutions. To take one example, policy disagreement around the proposed expansion of interprovincial energy pipelines has been particularly divisive. For instance, when the federal government authorized the Kinder Morgan pipeline in the fall of 2016, the decision was welcomed with applause from the Prairie provinces but with loud contestation in British Columbia. It did not take much more for observers to interpret the conflict as a clash between British Columbia’s strong environmental values and the Prairies’ economic interests and beliefs (McSheffrey 2016). The eventual decision on the Energy East pipeline surely will generate a similar conclusion that Canada’s regions have developed such distinctive sets of values that any national policy decision will struggle to find acceptance across the country.
In this context, policy-makers who seek national support for policies face a difficult challenge, which gives rise to the following question: Are value-based disagreements in Canada made worse by the highly decentralized nature of the Canadian federation? In the affirmative, such disagreement might lead to entrenched division between regional communities, polarized by distinct sets of political values and attitudes. Alternatively, have conventional thinking and the mainstream Canadian media exaggerated differences in regional values? If that is the case, policy-makers might benefit from paying more attention to values than to regions when devising policy.
To answer these questions, we assess political disagreements over values, and the extent to which these disagreements engender irreconcilable regional divides in policy preferences. Based on nationally and regionally representative samples of the Canadian population, we show that, on a broad range of policy issues and spending priorities, Canadians are primarily divided along three value inclinations: egalitarianism, traditionalism and “legal rigorism” (a belief that laws should apply strictly and uniformly to everyone). Moreover, we find this value divide to be largely independent of regions: values divide Canadians over policy issues and spending priorities irrespective of where they reside. Canadians disagree among themselves on policy issues not so much because of the decentralized nature of the country, but because they form a pluralist society. We therefore argue that Canada exhibits, at best, a weak form of regionalism in which the composition of values from one region to the next varies, but without preventing cross-regional alliances capable of fostering some level of national acceptance.
This finding has important implications for the country’s governance. The Canadian federation was designed to enable the expression of regional differences. Now particularly decentralized, this design sometimes inspires fears that interregional disagreements make it excessively difficult to formulate meaningful national policies that receive support from coast to coast. These disagreements, it is further feared, could undermine the legitimacy of federal institutions and might even endanger the country in the longer run. Our analysis suggests that such concerns should not be exaggerated, and points to several strategies policy-makers might use to help build support for policies in a context in which values matter more than regional differences.
When individuals disagree on political issues, debates normally focus on their policy preferences or preferred electoral candidates. In other words, debates — the visible manifestation of a disagreement — occur over specific issues, such as a new tax, the need to increase government spending on education, the fitness for office of a given candidate, and so on. Individuals, however, are economical in forming their preferences — that is, they do not conduct a complete search for information and balance the pros and cons each time a new issue appears on the public agenda or each time a potential candidate for office is mentioned in the press. Instead, they rely on heuristics, which are cognitive shortcuts that enable people to form preferences rapidly and cohesively on a large number of specific issues (Barker and Hansen 2005; Lau and Redlawsk 2001).
Values are among the most important heuristics available to individuals when forming their policy preferences (Skitka and Tetlock 1993). Defined as “abstract, general conceptions about the desirable and undesirable end-states of human life” (Jacoby 2006, 706), values serve as behavioural guides. Without values, defining preferences on a wide variety of policy issues would be a demanding task, requiring massive information and computation on the part of individuals. To save energy and time, individuals rely on values, which enable them to develop preferences on a large number of issues in a consistent and economical manner. An individual who prizes egalitarian values, for example, is likely, in a debate on welfare programs, to push for more spending. Disagreements over policy preferences are often a by-product of value differences — hence the central place accorded to values in the study of political disagreement.
Among the several approaches to analyzing political values, we draw from the literature four insights that guide our empirical analysis. First, individuals can hold several values simultaneously, not all of which serve as heuristics in determining policy preferences. In fact, for a value to operate as a heuristic, it must resonate with the framing of the issue (Goren et al. 2016). Although in some cases the echo is fairly self-evident — welfare spending resonates with egalitarianism, for example — it is not always the case. Environmental policy preferences, for example, have also been related to egalitarian values, especially where issue-framing stresses the potential social inequalities associated with projects, not just their potential environmental harm (Lachapelle and Montpetit 2014; Montpetit and Lachapelle 2013).
Second, most empirical analyses of values rely on survey research. Items intended to measure values require some a priori sense of which values matter most in a society. For example, Jacoby includes questions on freedom, equality, economic security, morality, social order, individualism and patriotism in his survey instrument based on “their relevance to the philosophical foundations of the American political system along with their salience in contemporary political discourse” (2014, 759). Other scholars deduce their questions from theory — for example, cultural theory and Schwartz’s circular model have been popular in recent years (Schwartz 2012; Thompson, Ellis and Wildavsky 1990). These two theories assume that values stem from humans’ orientations toward life. Schwartz’s values, for example, stem from the inclination of individuals toward conservation versus openness to change, and self-enhancement versus self-transcendence. Individuals who are inclined toward self-transcendence, with a tad of conservatism, tend to value benevolence, for example. Cultural theory, in contrast, assumes that different kinds of values are derived from the way in which individuals are organized in social relations. For example, an individual with a strong sense of belonging to a group, but who dislikes the hierarchical organization of society, will value equality.
Third, whether drawn from a particular theory, assumption or intuition, there are important similarities in the types of values researchers measure, although the labels often differ. Measuring social justice or egalitarian values is common, as are attempts at measuring individual autonomy and freedom. Respect for authority, the conservation of traditions and control over one’s own life are also common targets of social research. More important, the questions used by the proponents of Schwartz’s model and cultural theory to build measurement scales are strikingly similar (see, for example, Goren et al. 2016; Kahan and Braman 2006). In adapting them to the Canadian context, we drew extensively from these batteries of questions.
Fourth, another measurement strategy, which Jacoby (2006, 2014) uses, is to ask survey respondents to rank their values. Jacoby prefers this strategy on the ground that values are never isolated from one another but are structured into value configurations. In fact, individuals organize the values they hold in hierarchical order, and it is the order, rather than this or that value, that matters in the formation of policy preferences. For example, an individual who values equality but who also displays some level of deference to authority might be more tolerant of the use of police force against protesters. Although we did not use Jacoby’s measurement strategy in our survey, our analysis nonetheless accounts for the relative effects of different values by including them simultaneously in regression models. This is not the same as assessing the effect of a value structure, but our method nonetheless allows us to measure the relative importance of several values for each respondent.
Studies of regionalism in Canada stem from hypothesized or observed variations in political attitudes, values and cultures from one region to the next (Anderson 2010; Berdahl and Gibbins 2014; Héroux-Legault 2016; Montpetit and Foucault 2014; Paquet 2016; Simeon and Blake 1980; Simeon and Elkins 1974; Young, Faucher and Blais 1984). Scholarly debates have frequently centred on the delimitation of regions, or on whether regions, provinces or other entities best capture territorially circumscribed political differences (see, for example, McGrane and Berdahl 2013; Cochrane and Perrella 2012).
Although the delimitation of “region” and the depth of regional divides remain up for debate, it would be hard to argue that political disagreements in Canada are not at least partially affected by the country’s territorial division. The reason is fairly straightforward: although the formation of a federation depends on the willingness of individuals to unite the population of diverse territories into a single country, a federation is also the product of a desire to preserve the cultural distinctiveness of the populations living in these territories (Elazar 1985). This latter desire has always been strong in Canada; it has, throughout the history of the country, justified decentralization and resistance to centralizing forces. As a result, Canada today is one of the most decentralized federations in the world (Rodden 2004).
To conceptualize interregional disagreements in Canada, we rely on the distinction that Gidengil et al. (1999) make between regionalism’s “composition component” and its “effect component.” The composition component stresses the distribution of values from one region to the next. Here, regionalism would signify that the same values are present in all regions, although the distribution of these values varies from one region to the next. For example, egalitarian values might be more common in Quebec than deference to authority, which in turn might be more prevalent than social conservatism. Meanwhile, on the Prairies, social conservatism might be more common than deference to authority, which in turn might be more prevalent than egalitarianism. From this compositional perspective, however, two individuals strongly committed to the same values but separated by a regional boundary nevertheless can have similar policy preferences.
In contrast, the effect component of regionalism assumes that regional divides are so deep that issues are framed differently across regions, such that values have distinct regional effects — to say nothing of the possibility that each region will have its own distinctive set of values that count for preference formation. For example, an Albertan with a strong commitment to social equity might favour oil pipelines, to the extent they are framed as sources of wealth that help sustain generous social programs. In other regions of the country, however, individuals with a similar commitment to egalitarian values might instead oppose oil pipelines, seeing them as furthering inequality in favour of wealthy oil executives.
To summarize, when the effect component is at play, interregional disagreements over national policy proposals become prominent, as even individuals who share similar values become divided along regional cleavages. Effect components are indicative of a strong form of regionalism, which might lead to difficulties in making national policies and legitimacy problems for the central government. In contrast, the composition component reflects a weaker form of regionalism. In this latter case, regional value compositions might vary, but this variation is unlikely to have a large effect on aggregate policy disagreement across regions.
The empirical analysis developed in this study draws on an original, nationally and regionally representative survey of five major Canadian regions. The survey was conducted between March 17 and April 9, 2015, using a Computer-Assisted Web Interface administered by the Léger Research Intelligence Group. Respondents were drawn from Léger’s panel of more than 400,000 Canadians, the majority (61 percent) of whom were initially recruited from random digit dialling telephone samples. Survey invitations were sent in waves, thus avoiding an overrepresentation of the categories of respondents who tend to respond faster. The survey was open 24 hours a day and could be completed in several sessions. The survey was administered to five samples of Canadians, each representative of a particular region: British Columbia (BC: 1,012 respondents), the Prairies (PR: 1,032 respondents), Ontario (ON: 997 respondents), Quebec (QC: 1,035 respondents) and the Atlantic (ATL: 1,011 respondents). All results were weighted according to gender, age and language, using the latest census data from Statistics Canada.
Some scholars argue that regionalism in Canada is complex, and question the common understanding of Canada’s regions as dividing the country from east to west (Cochrane and Perrella 2012; Henderson 2004). Several other scholars, however, argue that this common-sense conceptualization of Canadian regionalism works well, and that other ways of dividing the country — notably, by province — fail to add much in terms of making more convincing descriptive and causal inferences (Anderson 2010; Héroux-Legault 2016; McGrane and Berdahl 2013). Although we did not test all possibilities, we found that using provinces instead of regions would not give us additional leverage in explaining policy preferences. Most notably, the variation observed in the three Prairie provinces was small, and disaggregating the data by province added little to the analysis of how values influence policy support, while splitting the region into its three constituent provinces reduced sample size, statistical power and representativeness. The same applies to the Atlantic provinces, which we also analyzed as a single region.
The questions in our survey were based on previous research, although we also conducted focus groups to validate our survey instrument; where necessary, we adjusted the wording of the questions for better comprehension by respondents. Following this crucial step, we included 18 value questions in our final instrument, which we reduced to three value inclinations — egalitarianism, traditionalism and legal rigorism — following a factor analysis (explained in the appendix).
The first value inclination, egalitarianism, measures a predisposition to support a society in which individuals are equal, in terms of wealth and opportunity, irrespective of the groups to which they belong. In scholarly work, it is the value inclination that seems to have the largest impact on various political attitudes, in Canada and elsewhere (Druckman and Bolsen 2011; Kahan et al. 2012; Lachapelle, Montpetit and Gauvin 2014; Montpetit and Lachapelle 2016; Silva, Jenkins-Smith and Barke 2007). Traditionalism, the second value inclination, refers to support for some social norms perceived as important in an idealized past. An excellent indicator of this value inclination is the importance given to the traditional family (Boily 2010; Bricker and Ibbitson 2013). Lastly, legal rigorism refers to an inclination to value laws that are applied strictly and uniformly to everyone. The refusal of special rights, even to an underprivileged minority, as well as stricter punishments for offenders, is indicative of legal rigorism. This value inclination also echoes some form of deference, often thought to play some role in Canadian political culture (Nevitte 1996).
Our survey measured policy preferences with questions about 7 spending preferences and 12 issue-specific preferences. Support for more spending on health, education and science was high and generally consistent within and across regions. Support for spending was lower overall and less consistent on defence, welfare programs and the national broadcaster, the Canadian Broadcasting Corporation. Support for spending on the environment was in-between. Disagreements were more significant on precise issues, including limits on abortion, increasing immigration, unions’ right to strike, police power, religious symbols and pipeline construction. (For details, see the appendix.)
Suggesting that there might be a composition component to Canadian regionalism implies that values might be distributed differently from one region to the next. The fact that, from region to region, the same three value inclinations arise from our numerous value questions already indicates that values — while not uniformly distributed — are nevertheless supported across the country. Moreover, as our analysis shows, the fact that these three value inclinations explain a large share of the variation in policy preferences in all five regions further testifies to large overlaps in values across regions. However, the distribution of the three value inclinations varies from one region to the next, a finding that is consistent with a compositional understanding of Canadian regionalism.
We computed responses to value questions to produce factor scores that situate respondents’ value inclinations in relation to a national average of zero. A negative sign indicates that a value inclination for a particular individual (or groups of individuals living in a particular region) is below the national average, while a positive number indicates that a particular value inclination is more salient in a region relative to the national average. Figure 1 presents the distributions of the scores for all five regions, as well as for Canada as a whole.
Compared with the national average, egalitarianism is more prominent in the Atlantic region, Quebec and Ontario. Legal rigorism is relatively more prominent on the Prairies and in Atlantic Canada, and above the national average in Quebec. It is under the national average in Ontario and British Columbia. Traditionalism is relatively more pronounced on the Prairies and in British Columbia. Also indicated in figure 1, a small minority of respondents in all regions vehemently opposed egalitarianism and legal rigorism. The long tails on the negative side of the distributions of egalitarianism and legal rigorism indicate a large distance between a small number of respondents and the national average for these value inclinations. Meanwhile, the relatively fat tails on the positive side of the distributions of traditionalism indicate that conservative values are present across the country, particularly on the Prairies, which stand out as the most distinct of the five regions.
To obtain a better sense of regional value differences, we ran a series of ordinary least square (OLS) regressions on the three value inclinations. Our goal with these models was to compare the explanatory power of the socio-economic variables conventionally used to explain values (see, for example, Héroux-Legault 2016) with that of regions. If regions really matter in value formation, adding regions to conventional models that explain values should boost the explained variance (the R2). In our regression models, presented in detail in the appendix, the socio-economic variables performed as the literature would lead one to expect. While explaining some better than others, socio-economic variables accounted for between 7 and 19 percent of the observed variance in values.
To provide some context, the models suggest that the younger a respondent is, the more egalitarian and less traditionalist the respondent is likely to be. Religious individuals are, on average, more traditionalist and have a higher predisposition to legal rigorism. Men are less egalitarian than women, and education reduces the probability that the respondent will have traditionalist and legal rigorist values. Individuals who identify with the federal Liberal Party are more egalitarian and less traditionalist than those identifying with the Conservatives. Moreover, as figure 2 illustrates, including regions in our models offered limited value added relative to traditional socio-economic predictors. Education, gender, party identification and religion are much more important than region of residence in explaining the distribution of values across the country.
In short, variations in values by region are relatively weak compared with variations that arise from differences in religion, language, age, gender and the existence of partisan groups within any given region. Canada’s regions have, at best, a modest impact on value differences among Canadians.
So far we have presented evidence that Canadian regionalism is associated with a modest composition component. However, relatively small social differences can have large political effects: a small difference in votes can have a large impact on the number of seats obtained by a political party; a small difference in public opinion can have a large impact on policy change; and, crucially for the focus of our study, a small difference in values can have a large impact on policy preferences. Indeed, given the distance separating some regions of the country and the power exercised by provincial governments, it might well be that policy debates are framed differently from one region to another. In turn, these differences in framing might exacerbate the modest differences in values observed across regions, producing sharp regional differences in preferences.
To verify this possibility, we ran OLS regressions on seven spending preferences, with the three value inclinations as the main independent variables of interest. Figure 3 compares the size of the explained variance with and without the inclusion of value inclinations in our models. Relative to models that include only socio-economic predictors, introducing values into the equation offers a statistically significant boost in terms of explained variance in spending preferences across all seven domains. In other words, much more than age, gender, religion or partisan affiliation, an individual’s values explain a substantively significant part of the individual’s spending preferences.
The hypothesis of an effect component to regionalism assumes, however, an interaction between values and regions. That is, values are expected to have different effects on policy preferences across different regions. To test this possibility, we reproduced all seven regressions, each time introducing an interaction term between one of the three values and the regions. As interactions are best interpreted in graphical format (Brambor, Clark and Golder 2006), figures 4 through 7 illustrate the results for the four domains in which the variations are the largest. (On the vertical axes of figures 4 to 11, “linear prediction” indicates the level of preference for a given issue, in a given region, at various value scores.)
Figure 4 presents the results of the interaction between values and regions on welfare spending. Overall, the patterns observed across regions are strikingly similar. For instance, although egalitarianism and legal rigorism have a powerful impact on welfare spending preferences, the effect is similar across all regions. Although the effect of traditionalism is less pronounced (as shown by the slope in figure 4), the overall effect is similar from one region to the next. And although British Columbia respondents — regardless of their scores on specific values — tended to be slightly more supportive of welfare spending than were respondents in other regions, they were particularly supportive if they valued egalitarianism, and significantly less supportive if they valued legal rigorism, just like other egalitarians and legal rigorists across the country. Thus, in contrast to the strong regionalism characterized by an effect component, the effects of value inclinations are similar across regions.
Figures 5 to 7 present the results for three other spending domains: environmental spending (figure 5), defence spending (figure 6), and spending on the CBC (figure 7). It is on defence spending that we come closest to finding evidence of an effect component of regionalism. In contrast to what we observe in other regions, egalitarian values in Atlantic Canada are associated with greater support for increased defence spending. The difference, however, is small and barely statistically significant. Atlantic Canada hosts shipyards, naval bases and other important military infrastructure. In this context, egalitarians might view support for defence spending as a source of additional employment, rather than as a desire to increase the country’s capacity to wage war. This being said, legal rigorism has the same effect on preferences for defence spending in Atlantic Canada as it does in the other regions (as shown earlier, legal rigorism is a strong value inclination in the Atlantic region).
Although our survey failed to find much evidence that region matters in the process of transforming values into spending preferences, figures 4 through 7 are nonetheless interesting in that they show the importance of values in explaining preferences. Egalitarianism increases support for welfare and environmental spending and for more spending on the CBC. Support for welfare spending decreases with traditionalism, but it decreases far more with a predisposition toward legal rigorism. Legal rigorists might associate welfare spending with some form of cheating. Legal rigorists also appear particularly reticent to spend on the CBC, perhaps viewing the network as unfairly advantaged in the competition with private networks for audiences and publicity revenues. Legal rigorists also support defence spending more than do traditionalists, while traditionalists support environmental spending more than do legal rigorists.
A possible objection to these tests might stress the proximity between some of the value questions and particular categories of spending (see Goren et al. 2016). For example, it might be said that a question on whether more should be done to reduce inequality between the rich and the poor measures more or less the same thing as a question about whether or not there should be more welfare spending. To further explore the power of values in explaining policy disagreement, we also examined preferences on specific policy issues.
To further isolate the role of values in explaining variations in policy preferences, our survey examined public preferences using 12 issue-specific items. Some of these issues have been more salient in parts of the country in recent years — for example, the gun registry on the Prairies, deficits in Ontario and religious symbols in Quebec — while others, such as foreign policy, are not as easily associated with a particular region. It is hardly surprising, therefore, that a gun registry received less support on the Prairies than it did in the rest of the country, while restrictions on religious symbols obtained more support in Quebec. Crucially, however, even on issues that are particularly salient in a given region, the region’s residents disagreed. The only issue of the 12 that appears relatively consensual in all regions is the need to make workplace safety regulations more stringent.
To test the effect of values on these specific policy preferences, we ran on each of them the regression model used for spending preferences. We obtained similar results: overall, the model works well, explaining between 10 and 31 percent of the variation in policy preferences, and as with spending preferences, values explain a large percentage of the variation in specific policy preferences, compared with the socio-economic factors included in the model. The only exception is support for new highway tolls, on which the model explains a mere 3 percent of the variation, perhaps owing to the fact that this issue is less value-charged and was insufficiently salient at the time of the survey.
As for spending preferences, introducing interactive terms in models examining support for specific policies increased the explained percentage of the variation in the dependent variables by one percentage point, at best. Figures 8 to 11 present a sample of the results we obtained with these interactions and come closest to evidence of an effect component of Canadian regionalism. Overall, however, results are much more consistent with the weak form of regionalism described earlier, which is characterized more by a composition component than an effect component.
Perhaps unsurprisingly, maintaining a gun registry received stronger support among those with egalitarian predispositions, irrespective of region of residence. As indicated by the relatively steep slopes observed in the left panel of figure 8, the effect of greater egalitarian sentiment is consistently strong and positive in terms of predicting support for maintaining a gun registry across all regions. In contrast, individuals with a stronger predisposition toward traditional values were less supportive of such a registry, especially if living on the Prairies (as indicated by a mean below 3, the “neutral” category on our 1 to 5 scale). Traditionalism, in fact, has a slightly stronger effect on the preference for a gun registry on the Prairies than in other regions, as illustrated by the steeper slope. Likewise, legal rigorists on the Prairies, while still in favour of a gun registry, were less supportive than those in the rest of the country. In other words, we can see a small effect component of regionalism in Figure 8.
An effect component is clearly visible in figure 9. Although individuals scoring above the national average for egalitarianism in Quebec were less supportive of making it mandatory for governments to balance their budget, egalitarian values do not matter as much in the other regions of Canada. On this same issue, however, traditionalism and legal rigorism in Quebec have effects that are similar to those found in the other regions of the country. The effect component is thus limited to balanced budgets being particularly salient for egalitarian values in Quebec, potentially reflecting the frustration among egalitarians in the province with the austerity policy the Couillard government was still pursuing at the time of the survey.
In recent years, the debate over religious symbols has been a particularly salient issue in parts of Canada. For instance, the Parti Québécois introduced a bill in the Quebec National Assembly prohibiting religious clothing in the civil service in 2013, and ran the 2014 provincial election on a pledge to pass the bill if re-elected. The party was defeated, but the issue has not disappeared from the agenda. Despite the issue’s particular salience in the province, however, as shown in figure 10, Quebec does not distinguish itself much from the other regions, for several reasons. First, egalitarian values do not shape public opinion on religious symbols as much as one might expect, a finding that is reflected by relatively flat slopes across regions, perhaps with the exception of British Columbia. These flat slopes probably result from the tension that exists between egalitarian predispositions toward gender equality (which might lead to support for limiting religious obligations imposed only on women) on the one hand, and egalitarian sensitivity to protecting minority rights (which might lead to opposition toward limits on religious symbols based on toleration of religious differences) on the other. Second, as in other regions, the difference in support for restrictions on religious symbols between traditionalists and nontraditionalists in Quebec is not statistically significant, which suggests there is no effect component here. Third, legal rigorism increases support for restrictions on religious symbols in Quebec, just as it does in all other regions. In fact, the strongest support for restricting religious dress is not in Quebec, but among legal rigorists in Ontario, British Columbia and on the Prairies.
Figure 11 presents results for perceptions of the proposed Energy East pipeline, an issue that has divided Canadian regions in recent years. Here again, the figure presents at best limited evidence of an effect component. Stronger egalitarian predispositions are associated with significantly more negative perceptions of the Energy East pipeline, regardless of where one resides. Although this is true in all regions, it is particularly the case among egalitarians in Quebec and British Columbia, where pipeline projects have been particularly controversial. Pipelines are promoted by large oil corporations, which tend to be viewed with suspicion by individuals who value equality. What is interesting, however, is that this relationship between egalitarian values and attitudes toward pipelines also prevails on the Prairies, the region that has the most to gain from additional pipelines. In contrast to egalitarians, legal rigorists have a much more positive perception of energy pipelines. This is true on the Prairies, but also in Quebec and British Columbia, where legal rigorists might also have reacted negatively to protests against pipelines. The strongest evidence of an effect component of regionalism in figure 11, however, is in the middle panel, which plots the effect of traditional values. In fact, traditionalism appears to have distinct effects on support for energy pipelines across different regions, as illustrated by the diverging slopes in the figure. Overall, the effect of regionalism appears to be greatest among Canadians with a weaker predisposition toward traditionalist values, although this effect disappears — that is, opinions across regions converge — when comparing individuals with a greater inclination toward traditionalism.
In short, we find some instances of a modest effect component associated with regionalism. Although beyond the scope of this study, we suspect these differences to be the result of the regional framing of issues, which might resonate in different ways with the value orientations explored here. It is worth noting again that figures 8 to 11 present the most compelling evidence in support of an effect component. On the other eight specific issues, regional issue framings failed to influence the way values translate into policy preferences. Overall, then, we find substantial evidence that is consistent with a weak form of regionalism where values shape policy disagreements on large spending categories, as well as on specific policy issues, across all Canadian regions and in a similar way.
We began by stating that disagreement is an essential component of a functioning democracy, while warning that, beyond a certain threshold, policy disagreement might become harmful to policy-making capacity and legitimacy. Given the decentralized nature of the Canadian federation and the sheer size of the country, policy debates risk becoming provincialized or regionalized, pulling apart individuals who otherwise have similar values. For example, debates surrounding pipeline construction and the like are often interpreted through a regional lens as pitting some provinces against others in ways that erode the legitimacy of federal institutions such as the National Energy Board. Our survey shows that energy pipelines, along with several other controversial policy issues and spending priorities, divide Canadians with different value inclinations, irrespective of where they live. Overall, our results suggest that, although regions matter, policy disagreements are primarily the product of value conflicts that are similar from one region to the next.
The importance of regionalism in Canada has been under the microscope for a long time. Research has examined whether regions differ along policy preferences, values, economic policy choices and policy regimes. These studies often disagree on the significance of regionalism for the country. The originality of the current study rests on our comparison between regions and values as distinct factors of division. Instead of simply listing regional differences, we use values as benchmarks to measure the significance of such differences. We find that values have similar effects on policy preferences in every region. Compared to the importance of values in defining policy preferences, regions play a much smaller role. This is not to say that regions do not matter; the crucial point here is that we find a composition effect for regions — that is, regions are internally divided over similar values, while the weight given to each value varies from one region to the next. Therefore, the construction of energy pipelines, gun registration and restricting religious symbols get more or less support from region to region, but within each region these issues are controversial, dividing individuals variably inclined to different sets of values. In other words, disagreements on policy issues exist because of the plurality of values that characterizes Canadians, not because of territorial lines of fracture.
This analysis carries some practical lessons, and we conclude with three implications for policy. First, policy should be designed, framed and promoted to appeal primarily to values, not to regions. A policy that appeals to egalitarianism, legal rigorism or traditionalism might be supported in all regions, though possibly a little more in some regions than in others. In contrast, efforts to appeal to any particular region risk alienating other regions, even residents whose values resonate with the policy. For instance, a policy too obviously aimed at gaining support in Ontario risks being viewed with suspicion by Canadians in other regions, regardless of their value inclinations. A policy obviously aimed at gaining support among egalitarians, depending on how it is framed, might draw criticism from traditionalists and/or legal rigorists, but could attract supporters from every region.
Obtaining the widest support possible is crucial to policy success. In this regard, the division of Canadians along more than two value inclinations offers opportunities for framing policy in ways that garner support beyond any single value inclination. We saw, for example, that egalitarian values increase support for a gun registry, while traditional values reduce support for this policy. Closely associated with the murder of 14 women at Montreal’s École polytechnique in 1989, the registry appeals to those who believe that guns symbolize an extreme form of male domination. Conversely, in traditional circles, guns are often viewed as a means for the breadwinner to provide the family with food and physical safety. A gun registry, however, has apparently failed to resonate, one way or another, among legal rigorists. Thus, to increase the support base for a gun registry, proponents might address the issue in ways that also appeal to legal rigorists. In other words, the law-and-order frame might become prominent in a way that shows that current regulations offer Canadians — both women and men — insufficient protection.
Second, our analysis has implications for policy actors at the provincial level. As policy-making in a democracy often amounts to attending to problems in a context where disagreement is not only common but prized, provincial policy-makers and advocates have much to learn from one another. Values divide regions in similar ways, just as they induce spending and policy preferences in similar ways across the country. In such a context, provincial policy-makers and policy advocates might find it beneficial to consult their counterparts in other provinces when developing policies and strategies. Our findings should encourage provincial policy-makers and other policy actors to continue and even intensify discussions on best practices to cope with disagreement and strategies to build policy support across value differences.
Third, if policy should be designed, framed and promoted to appeal first to values, regions should not be entirely forgotten. Although the three value inclinations we found to be common heuristics operate in much the same way in all regions, there are also some instances of regionally distinct effects. Requiring governments to have balanced budgets, for example, receives weaker support among Quebec egalitarians than it does among egalitarians in the rest of the country.
Our results suggest that, while national and regional debates are usually framed in the same way across the country, promoters of national policy who seek countrywide support sometimes should adopt regionally sensitive communication strategies. For example, if building a base of support for the Energy East pipeline in Quebec and the Atlantic provinces is important, some sensitivity to these regions’ particularities might be warranted. In fact, the distribution of values in Quebec and Atlantic Canada advantages pipeline opponents. Egalitarians in these regions are likely to view the pipeline as largely a benefit to the oil industry, and while this should also be true of Alberta’s egalitarians, egalitarians are more numerous in the Atlantic provinces and Quebec. As figure 11 indicates, however, traditionalism and legal rigorism offer more opportunities to build support in Quebec than anywhere else in the country. Therefore, a regionally sensitive strategy might be to expend additional efforts to reach out to Quebec traditionalists and legal rigorists.1
Federations are designed to balance unity and regional distinctiveness. Striking the right balance is difficult, however. Excessive appeals to national unity may produce perceptions that the federal idea has been violated, while too much emphasis on decentralization feeds fears that interregional disagreements will be harmful to policy acceptance and legitimacy. Our study of public opinion in Canada, one of the most decentralized federations in the world, suggests that these latter fears may be exaggerated. Regional differences should not be ignored, but more systematic attention to value disagreements within regions seems warranted.
Our survey began with a series of 18 value questions, several of which are adaptations of questions used by cultural theory scholars (see Kahan, Jenkins-Smith and Braman 2011). In addition, several overlap with questions used to test Schwartz’s model (see Goren et al. 2016). After conducting focus groups, we also developed our own questions to tap into the particularities of Canadian culture. For each of these, respondents were asked to indicate their level of agreement on a six-point scale, from strongly disagreeing to strongly agreeing. We ran a factor analysis on the results, with a principal component specification and Varimax rotation, in order to identify latent variables in our battery of 18 questions. Latent variables here refer to related or correlated questions, which together indicate a value inclination latent in our survey (known as a factor in factor analysis).
Although each question in a set of correlated questions in a factor analysis captures a particular dimension of a value inclination, together the set of questions provides a reliable measure of the overall value inclination. Depending on the method of interpretation of factor analysis, we found between three and four latent value inclinations among our 18 questions. Following the suggestion by Goren et al. (2016) that all values might not be relevant in explaining particular political preferences, we ran some tests on each of the four factors, and found that the fourth factor did not add any explanatory power to the first three.
Our analysis thus left us with three value inclinations, which we present in table A-1. To maintain some symmetry among the three value inclinations, for each of the factors we kept only the three questions with factor loadings above 0.5. We reproduced the factor analysis at the regional level, and the nine questions always loaded onto the same three factors. In other words, the 9 questions presented in table A-1 helped us identify the same value inclinations for Canadians across the country.
Value inclinations are not organized as neatly in individuals as they are in table A-1. An individual can feel strongly that inequalities between rich and poor should be reduced but not feel as strongly about the need to reduce inequalities between men and women. In fact, an individual’s values — and by extension those of a region — can be quite complex. To capture this complexity, we predicted factor scores for each value inclination using the three components highlighted in table A-1. This weighting procedure produced factor scores that account for the relative importance attributed to each item within a given factor (that is, value orientation) for a particular individual.
We centred factor scores at zero, so the mean scores, presented in table A-2, are differences from national average scores. The regional scores vary between -2.88 and 1.19 for egalitarianism, -1.41 and 1.91 for traditionalism and -2.57 and 1.27 for legal rigorism. As shown in table A-3, the Atlantic provinces and Quebec have strikingly similar scores: both regions score higher than the national average for egalitarianism and legal rigorism, and below average for traditionalism. Although this similarity may come as a surprise, the portrait of Quebec that emerges is consistent with the importance Quebecers are known to accord to social justice. Interestingly, Ontarians are also above the Canadian average in terms of egalitarianism but below average for legal rigorism. In fact, the Atlantic provinces and Quebec share with the Prairies, but not with Ontario or British Columbia, stronger than average legal rigorist sentiment. While sharing with British Columbia an above average inclination for traditionalism, the Prairies appear the most distinct of all five regions: there, support for egalitarianism is below average, while support for both traditionalism and legal rigorism is above average.
Our survey included an additional 7 questions on spending preferences and 12 questions on preferences about specific policy issues. The 7 spending domains are health, education, welfare, the environment, science, defence and the Canadian Broadcasting Corporation. In each of these domains, respondents were asked whether they would like to see government spend much less (1), less (2), about the same (3), more (4) or much more (5). On three of these issues — health, education and science — Canadians across the country were relatively consensual: they wanted more spending. The average score for health and education was 4, and for science 3.8. The standard deviation (SD) for all three domains was small, at 0.8. With so little variation, value and regional cleavages likely add little to the understanding of these spending preferences. At 3.7, support for environmental spending was a little lower than for health and education. And with an SD of 0.9, a little more disagreement among Canadians is evident. Spending preferences on defence (3.0; SD 1.0), welfare (3.0; SD 1.1) and the CBC (2.9; SD 1.1) were more diverse. On these four domains, value and regional differences might explain relatively more of the observed disagreement.
We measured specific policy preferences on a scale ranging from 1 to 5, where 1 indicated strong opposition and 5 strong support. The question on the oil pipeline, however, was phrased somewhat differently (see table A-3) and used a scale from 0 to 10, where 0 indicated that the pipeline is not beneficial at all and 10 that it is extremely beneficial. (It is good practice in survey design to modify response scales to verify whether respondents react to the scales, rather than to questions. The oil pipeline response scale helped us to ascertain the validity of the responses.) More specific than questions on spending preferences, questions on policy preferences yielded fewer consensual responses. In fact, only our question on the necessity of tougher workplace regulations yielded a level of support comparable to that given by respondents to health and education spending, with little variation (as indicated by a relatively low standard deviation). The questions, means and standard deviations are presented in table A-3, and cover such issue areas as labour rights, health immigration, national security and policies related to energy and the environment. The issues were selected for their importance in Canada or particular regions. Although the mean support/opposition varies from one region to the next, none of the issues except work safety regulation is consensual in a region.
Studies (notably, Héroux-Legault 2016; McGrane and Berdahl 2013) have shown that gender, age, education, religion, revenue, language and party identification are the typical socio-economic factors that affect values. Our regressions sought to measure how much of the variation in value inclinations is explained once we added the respondent’s region of origin to these socio-economic factors — see table A-4 for the detailed regression results. To simplify, we measured the socio-economic factors using only continuous and dummy variables. If we replaced the dummy variable identifying, for example, the supporters of the Liberal Party by a categorical variable that included all the major political parties, the results became more specific but did not change. Although the proportion of explained variance increased, introducing categorical variables among the socio-economic factors did not change the effect of regions.
The regressions in table A-4 confirm the relevance of the socio-economic variables. Consistent with previous studies, age is negatively correlated with egalitarianism and positively correlated with traditionalism and legal rigorism (Héroux-Legault 2016). Although unrelated to egalitarianism, education is negatively correlated with traditionalism and legal rigorism. Income is negatively correlated with both egalitarianism and traditionalism but unrelated to legal rigorism. Men tended to be less egalitarian and more traditionalist than women. Men and women did not differ, however, in their inclination toward legal rigorism. Individuals most likely to attend religious services tended to have more traditional values than did less religious individuals; they also tended to score a little higher on legal rigorism. Respondents who identified with the Liberal Party were more inclined toward egalitarian values and less inclined toward traditional ones than were other respondents, especially Conservatives (not shown in table A-4). Lastly, respondents who filled out the questionnaire in French tended to value traditions less than those who filled it out in English. These socio-economic variables explain 8 percent and 7 percent, respectively, of the variation in egalitarianism and legal rigorism (columns 1 and 5 in the table), and as much as 16 percent of the variation in traditionalism (column 3).
Columns 2, 4 and 6 on table A-4 illustrate the salience of including region of residence in models that predict values: for two value inclinations — traditionalism and legal rigorism — an additional 1 percent of the variation is explained; for egalitarianism, the added value is less than 1 percent. Using British Columbia as the region of reference, we see that the Prairies are the most distinct region when it comes to value inclinations: of all the regions, only the Prairies significantly differ from British Columbia on egalitarianism. The difference between British Columbia and Quebec on traditionalism is barely statistically significant, while the difference between the Prairie provinces and British Columbia is particulary large. Similarly, Atlantic Canada is more rigorist than British Columbia, but the distinction between British Columbia and the Prairies on legal rigorism is even sharper.
Despite these regional variations in values, it is important to keep in mind that taking regions into account adds little to the explanatory power of socio-economic variables typically used to explain value differences among individuals. In other words, table A-4 suggests that disagreements over values in Canada are a socio-economic matter much more than a question of one’s region of residence.
Table A-5 presents regressions showing the effect of values on four spending priorities over which disagreements are greatest. We do not present the results for education, health and science to avoid overburdening the table; in fact, we obtained similar results for these three domains, although the wide consensus on the need to spend more leaves us with little variation to be explained. The models that include both values and socio-economic variables (models 2, 4, 6 and 8) performed well in all spending domains, explaining between 16 and 24 percent of the variation in preferences; alone, the socio-economic variables explain much less of the variation. When we removed values from the model, the explained variation in welfare spending preferences dropped from 24 percent to 5 percent, in environmental spending preferences the explained variation dropped from 19 percent to 4 percent, in defence spending preferences it dropped from 16 percent to 7 percent, and in spending on the CBC it declined from 18 percent to 6 percent.
The older the respondent, the more welfare and defence spending and the less environmental spending the individual was likely to prefer. University graduates preferred more spending in all domains except defence. Males were less likely than females to prefer environmental spending. Partisans of the Liberal Party were in favour of more welfare, environmental and CBC spending. The wealthier the respondent, the less likely the individual was to prefer spending on welfare. Francophone respondents were more likely to support Radio-Canada than anglophones were to support the CBC and less likely to support defence spending. Egalitarianism is positively associated with support for more welfare, environmental and CBC spending and with less defence spending. Traditionalism and legal rigorism have the opposite effect, although legal rigorism has a stronger effect than traditionalism on defence spending. Overall, value inclinations explain much more of the variation in spending preferences than do conventional socio-economic factors.
The authors would like to thank Scott Matthews for his comments on an earlier version of this study, as well as the two peer reviewers solicited by the Institute for Research on Public Policy. Jean-Philippe Gauvin provided excellent research assistance. The authors also acknowledge the financial support of the Social Sciences and Humanities Research Council of Canada.
Anderson, C. 2010. “Regional Heterogeneity and Policy Preferences in Canada: 1979–2006.” Regional & Federal Studies 20 (4-5): 447-68. doi:10.1080/13597566.2010.523620
Barker, D.C., and S.B. Hansen. 2005. “All Things Considered: Systematic Cognitive Processing and Electoral Decision-Making.” Journal of Politics 67 (2): 319-44. doi:10.1111/j.1468-2508.2005.00319.x
Berdahl, L., and R. Gibbins. 2014. Looking West: Regional Transformation and the Future of Canada. Toronto: University of Toronto Press.
Boily, F. 2010. Le conservatisme au Québec: retour sur une tradition oubliée. Quebec City: Presses de l’Université Laval.
Brambor, T., W.R. Clark, and M. Golder. 2006. “Understanding Interaction Models: Improving Empirical Analyses.” Political Analysis 14: 63-82.
Bricker, D., and J. Ibbitson. 2013. The Big Shift: The Seismic Change in Canadian Politics, Business, and Culture and What It Means for Our Future. Toronto: HarperCollins Canada.
Cochrane, C., and A. Perrella. 2012. “Regions, Regionalism and Regional Differences in Canada.” Canadian Journal of Political Science 45 (4): 829-53. doi:10.1017/S0008423912001011
Druckman, J.N., and T. Bolsen. 2011. “Framing, Motivated Reasoning, and Opinions about Emergent Technologies.” Journal of Communication 61 (4): 659-88. doi:10.1111/j.1460-2466.2011.01562.x
Elazar, D.J. 1985. “Federalism and Consociational Regimes.” Publius 15 (2): 17-34. doi:10.2307/3329962
Gidengil, E., A. Blais, R. Nadeau, and N. Nevitte. 1999. “Making Sense of Regional Voting in the 1997 Canadian Federal Election: Liberal and Reform Support Outside Quebec.” Canadian Journal of Political Science 32 (2): 247-72. doi:10.1017/S0008423900010489
Goren, P., H. Schoen, J. Reifler, T. Scotto, and W. Chittick. 2016. “A Unified Theory of Value-Based Reasoning and U.S. Public Opinion.” Political Behavior 38 (4): 977-97. doi:10.1007/s11109-016-9344-x
Henderson, A. 2004. “Regional Political Cultures in Canada.” Canadian Journal of Political Science 37 (3): 595-615. 10.1017/S0008423904030707
Héroux-Legault, M. 2016. “Substate Variations in Political Values in Canada.” Regional & Federal Studies 26 (2): 171-97.
Jacoby, W.G. 2006. “Value Choices and American Public Opinion.” American Journal of Political Science 50 (3): 706-23. doi:10.1111/j.1540-5907.2006.00211.x
This publication was published as part of Canada’s Changing Federal Community research program under the direction of F. Leslie Seidle. The manuscript was copy-edited by Barry Norris, proofreading was by Zofia Laubitz, editorial coordination was by Francesca Worrall, production was by Chantal Létourneau and art direction was by Schumacher Design.
Éric Montpetit is a professor of political science and vice-dean of faculty life in the Faculty of Arts and Sciences at the Université de Montréal. He has published extensively on policy-making in the Canadian federal context, as well as on the politics of scientific expertise. His latest book is In Defense of Pluralism: Policy Disagreement and Its Media Coverage (2016). He received a PhD in comparative public policy from McMaster University in 1999.
Erick Lachapelle is an associate professor in the Department of Political Science at the Université de Montréal, where he teaches courses on environmental politics, public policy and research methods. He joined the faculty at the Université de Montréal in August 2011 after receiving his doctorate from the University of Toronto. He is the lead researcher behind the Canadian Surveys on Energy and the Environment, and he has published numerous articles on the comparative politics of climate change, environmental public opinion and the politics of scientific expertise.
Simon Kiss is a political scientist in the digital media and journalism and leadership programs at Wilfrid Laurier University. He teaches courses on journalism, public opinion, research methods and political communication. His research is mostly in the politics of managing environmental risks in Canada, and he is currently a member of a multidisciplinary research initiative examining ways of improving the use of scientific evidence in the Canadian policy process. He received his PhD, which examined the evolution of government communications in the province of Alberta, from Queen’s University.
To cite this document:
Montpetit, Éric, Erick Lachapelle and Simon Kiss. 2017. Does Canadian Federalism Amplify Policy Disagreements? Values, Regions and Policy Preferences. IRPP Study 65. Montreal: Institute for Research on Public Policy.
Montreal – Canadians’ differences over values have a considerably greater effect on their public policy preferences than does the region where they live, finds a new study from the Institute for Research on Public Policy.
“Policy disagreements have often been interpreted through the prism of regionalism. Our findings show that although regions matter, policy disagreements are primarily the product of value conflicts that are similar from one region to the next,” say Éric Montpetit, Erick Lachapelle and Simon Kiss.
Their analysis is based on an original survey developed to measure the degree to which Canadians are divided along three value inclinations: egalitarianism, traditionalism and legal rigorism (a belief that laws should apply strictly and uniformly to everyone). They find that individuals who share the same values, regardless of where they live, have similar policy preferences. For example, egalitarian values have a strong impact on Canadians’ welfare spending preferences, and the effect is similar across all regions.
More broadly, the authors observe that there may have been a tendency to exaggerate fears that disagreements among regions will be harmful to policy acceptance and institutional legitimacy.
They conclude that public policy should be designed, framed and promoted to appeal primarily to values, not to regions. Although the composition of values varies among regions, this does not prevent the formation of cross-regional alliances that can help build national acceptance.
The survey was conducted in March and April of 2015, with a sample of around 1,000 Canadians in each of five regions: British Columbia, the Prairie provinces, Ontario, Quebec and the Atlantic provinces.
Does Canadian Federalism Amplify Policy Disagreements? Values, Regions and Policy Preferences, by Éric Montpetit, Erick Lachapelle and Simon Kiss, can be downloaded from the Institute’s website (irpp.org).
The Institute for Research on Public Policy is an independent, national, bilingual, not-for-profit organization based in Montreal. To receive updates from the IRPP, please subscribe to our e-mail list.
Media contact: Shirley Cardenas tel. 514-594-6877 firstname.lastname@example.org