Introduction
Public opinion surveys are extremely useful for capturing the distribution of a variable among a given population. They make it possible to determine how widespread an attitude or a behaviour is among a group. Surveys are also frequently used to examine relationships between variables. They help to identify the determinants of an attitude or a behaviour. However, the presence of a statistical association between the responses to two survey questions does not allow us to know whether one variable is really the cause of the other. To increase our confidence in the existence of a causal relationship between two variables, various investigative approaches are possible. On the one hand, we can turn to an experimental design—for instance, a survey experiment—where one variable is manipulated to discern whether the other variable is affected. On the other hand, panel surveys offer the opportunity to examine the passage of time in order to ascertain whether changes in one variable really lead to subsequent movements in the second variable.
This research note uses panel surveys to shed light on a question that remains unanswered about the factors that influence support for Quebec sovereignty. The desire to see the Canadian province become an independent country has long been correlated to citizens’ expectations about the consequences of independence on economic conditions and on the linguistic situation. Those in favor of Quebec sovereignty tend to think that it will improve the health of the economy and the French language, while those opposed to sovereignty generally anticipate negative consequences for the economy and the French language. The question that we tackle here is whether a person’s preference regarding Quebec sovereignty stems from their economic and linguistic expectations, or whether those expectations merely represent a projection of their opinion about the desirability of Quebec independence. We rely on the panel surveys of the Canadian Election Study conducted in 2004, 2006, 2008 and 2011. Methodologically, we employ cross-lagged models. We find that economic and linguistic expectations have some impact on support for sovereignty, however, the reverse relationships are much more important. For many citizens, it is the preference concerning Quebec sovereignty that shapes their expectations about the consequences of independence.
Determinants of Support for Quebec Sovereignty
The debate about Quebec’s political future has been ongoing for more than half a century. Some people want the Canadian province to separate and become an independent, sovereign country. Others prefer it to remain within Canada. Even though the proportion of people in each camp rises and falls over time, this issue structures political competition and affects the electoral behaviour of citizens decade after decade (Cloutier et al., Reference Cloutier, Guay and Latouche1992; Bélanger and Nadeau, Reference Bélanger and Nadeau2009; Yale and Durand, Reference Yale and Durand2011; Bélanger et al., Reference Bélanger, Daoust, Mahéo and Nadeau2022).
What factors explain support for and opposition to Quebec sovereignty? Two components are considered to be among the most important determinants in previous research: an identity and affective dimension, and a cost-benefit calculation dimension (Blais and Nadeau, Reference Blais and Nadeau1992; Martin, Reference Martin1994; Blais et al., Reference Blais, Martin and Nadeau1995; Nadeau and Fleury, Reference Nadeau and Fleury1995; Nadeau et al., Reference Nadeau, Martin and Blais1999; Mendelsohn, Reference Mendelsohn2003; Bélanger and Perrella, Reference Bélanger and Perrella2008). The first dimension refers to the relative attachment that individuals feel towards Quebec and Canada. People more attached to Quebec tend to be sovereignist, while those more attached to Canada tend to be federalist.Footnote 1 The second concerns people’s expectations about the impact of sovereignty on the economy and the French language, compared with the status quo. This is a comparative and forward-looking perspective. People who believe that sovereignty would have positive economic and linguistic consequences are more inclined to be in favour of it, while those who expect negative consequences are more likely to be against it.
How should we interpret these correlations between support for sovereignty and economic and linguistic expectations? On the one hand, researchers from the Montreal school consider that “the evaluation of anticipated costs and benefits influences the choices of a sufficient number of individuals to have a decisive effect on the final outcome” (Blais and Nadeau, Reference Blais and Nadeau1992; Martin, Reference Martin1994; Blais et al., Reference Blais, Martin and Nadeau1995: 638; Nadeau and Fleury, Reference Nadeau and Fleury1995; Nadeau et al., Reference Nadeau, Martin and Blais1999). The notions of utility maximisation (Blais et al., Reference Blais, Martin and Nadeau1995) and rationality (Martin, Reference Martin1994) are invoked. It is thus assumed that expectations are the independent variables and that support for sovereignty is the dependent variable. On the other hand, some studies question this conception and propose that causality flows in the opposite direction (Howe, Reference Howe1998; Mendelsohn, Reference Mendelsohn2003).Footnote 2 Rather than a rational choice, these correlations could represent rationalization.Footnote 3 More specifically, “rather than sunny assessments of the economic effects of sovereignty generating sovereignty support, it is enthusiasm for sovereignty that causes people to be optimistic about its economic consequences” (Howe, Reference Howe1998: 32). Here, it is support for sovereignty that influences expectations about the consequences of sovereignty.
Determining which of these two views is more empirically sound has not proved possible to date. For example, Howe admits that in the absence of more appropriate data, he offers only speculative and suggestive evidence to support his argument (1998: 32–33, 51–53). With cross-sectional surveys, as Mendelsohn points out, “Regression analysis in the end cannot perfectly sort out the dependent from the independent variables” (Reference Mendelsohn2003: 518).
Cross-Lagged Models
To examine the relationships between support for sovereignty and economic and linguistic expectations, we propose to use panel data and cross-lagged models (Campbell and Stanley, Reference Campbell and Stanley1963; Kenny, Reference Kenny1975; Finkel, Reference Finkel1995). This type of analysis consists of two stages. First, the correlation between expectations at time T-1 and support at time T is examined, while controlling for support at time T-1. We therefore test whether previous expectations predict change in support for independence. Then, we turn things around. We examine the correlation between support at time T-1 and expectations at time T, while controlling for expectations at time T-1. The aim is to see whether past support predicts changes in expectations. Finally, the results of the two stages are compared to decide whether the influence follows one direction more than the other.
This approach has notably been used by Lenz (Reference Lenz2009; Reference Lenz2013) to study the impact of issue positions on vote choice and, conversely, the impact of vote choice on issue positions. Lenz concludes that vote choice affects issue positions more than the opposite. While it is generally believed that citizens modify their electoral preference to support the political party that defends their positions, these results indicate instead that citizens modify their positions so that they are in line with their electoral preference. In other words, voters tend to adopt the positions of their favourite party. However, this dynamic is not dominant in all contexts. Analysing the 2017 French presidential election with cross-lagged models, Guntermann and Lachat (Reference Guntermann and Lachat2021) show that citizens’ issue positions explain their electoral choices when a new political party appears. Those who held centrist positions before the emergence of Emmanuel Macron were more likely to vote for this candidate on election day. To a lesser extent, there was also rationalization; some Macron supporters changed their positions and adopted those of the En Marche! movement. However, during this election, the influence was mainly from issue positions to vote choice rather than the other way around.
Some studies have used the same method to assess whether partisan identification leads to changes in issue positions, and whether issue positions lead to changes in partisan identification (Layman and Carsey, Reference Layman and Carsey2002; Carsey and Layman, Reference Carsey and Layman2006; Highton and Kam, Reference Highton and Kam2011; Engelhardt, Reference Engelhardt2021). Other research has done a similar exercise to assess the links between partisan identification and political values (Goren, Reference Goren2005; Evans and Neundorf, Reference Evans and Neundorf2020). A case can be made that opinions about Quebec sovereignty are very strong and can be conceptualized as the equivalent of partisanship in the United States. To our knowledge, no study has applied the technique of cross-lagged models to the case of support for Quebec sovereignty.
Data
The Canadian Election Study is a series of public opinion surveys conducted at each federal election since 1965. A large sample representative of the voting-age population is contacted each time (3,000–4,000 respondents). Approximately a quarter of the respondents are from the province of Quebec. The analyses here are limited to residents of Quebec. For three elections between 2006 and 2011, part of the sample was made up of respondents who had taken part in the study for the previous election. We can therefore use panels for the following pairs of elections: 2004/2006, 2006/2008 and 2008/2011.
In each of these surveys, support for Quebec sovereignty and expectations are consistently measured the same way. The support question asks: “Are you very favourable, somewhat favourable, somewhat opposed, or very opposed to Quebec sovereignty, that is Quebec is no longer part of Canada?” The question on economic expectations is as follows: “If Quebec separates from Canada, do you think your standard of living will get better, get worse, or stay about the same as now?” For linguistic consequences, the question is: “If Quebec separates from Canada, do you think the situation of the French language in Quebec will get better, get worse, or stay about the same?” All variables are recoded to range from 0 to 1 (where a high score indicates a positive opinion on sovereignty and the economic and linguistic consequences). Table 1 displays the means of these variables for each year. The central tendency of each attitude is relatively stable over time. The majority of respondents is opposed to Quebec sovereignty and anticipates that sovereignty would have negative economic consequences and positive linguistic consequences.
Table 1. Variable Averages
All variables range from 0 to 1.
Results
To illustrate the findings of previous research, Table 2 reports the results of regression analyses based on cross-sectional data (that is, measured in the same survey). In each of the four elections, expectations about the economic and linguistic consequences of sovereignty are strongly correlated with support for sovereignty. Individuals who are more optimistic about the impact of independence are more likely to be in favour of secession. The two expectation variables alone explain 40 per cent or more of the variance in support for sovereignty. It is important to note that the coefficients of expectations remain strong and significant when control variables, such as relative attachment to Quebec, are added.Footnote 4
Table 2. Determinants of Support for Quebec Sovereignty: Cross-Sectional Analyses
OLS non-standardised coefficients and standard errors. All variables range from 0 to 1.
***p < .001; **p < .01; *p < .05.
We now turn to the analysis of the panel data. The first stage of the cross-lagged models tests the existence of a genuine relationship between expectations and support for sovereignty. We examine the impact of past expectations on current support, while taking past support into account.Footnote 5 Table 3a presents the results for economic expectations, while Table 3b presents those for linguistic expectations.Footnote 6 The magnitude of the coefficients of the lagged dependent variable shows that support for sovereignty is very stable over time. In each of the three models in Table 3a, economic expectations have a significant effect on opinions towards sovereignty. Economic expectations are responsible for subsequent changes in support for independence. However, these effects are much smaller than those found in the cross-sectional data (Table 2). In the case of linguistic expectations (Table 3b), only one of the three coefficients is significant (Model 6). Linguistic expectations are therefore less relevant than economic expectations in explaining support for sovereignty.
Table 3a. Cross-Lagged Analyses: Economic Expectations and Support for Sovereignty
OLS non-standardised coefficients and standard errors. All variables range from 0 to 1.
***p < .001; **p < .01; *p < .05.
Table 3b. Cross-Lagged Analyses: Linguistic Expectations and Support for Sovereignty
OLS non-standardised coefficients and standard errors. All variables range from 0 to 1.
***p < .001; **p < .01; *p < .05.
The second stage of the cross-lagged models tests for the presence of rationalization. It asks whether prior support for sovereignty affects current expectations when controlling for prior expectations. The coefficients of the lagged dependent variables indicate that economic expectations (Table 3c) and linguistic expectations (Table 3d) are less stable over time than support for sovereignty (Tables 3a and 3b). Models 7, 8 and 9 reveal that views about independence systematically have a significant impact on Quebeckers’ economic expectations. The same dynamic is observed for linguistic expectations (Models 10, 11 and 12). Those who are more favourable towards sovereignty tend to develop more optimistic expectations about the consequences of separation on the state of the economy and the French language.
Table 3c. Cross-lagged analyses: Support for Sovereignty and Economic Expectations
OLS non-standardised coefficients and standard errors. All variables range from 0 to 1.
***p < .001; **p < .01; *p < .05.
Table 3d. Cross-Lagged Analyses: Support for Sovereignty and Linguistic Expectations
OLS non-standardised coefficients and standard errors. All variables range from 0 to 1.
***p < .001; **p < .01; *p < .05.
Finally, we need to compare the results of the two stages to determine whether influence flows more in one direction than the other. To facilitate the comparison, we have calculated the averages of the standardised coefficients for each trio of models (these coefficients hence differ from the non-standardised coefficients reported in Tables 3a–3d). The average standardised impact of economic expectations on support for sovereignty is 0.12 (Models 1–3), while that of support for independence on economic expectations is 0.30 (Models 7–9). Thus, the influence of support for sovereignty on economic expectations is much greater than the influence of expectations on support.Footnote 7 Despite the fact that economic expectations do affect some respondents’ views about secession, there is a great deal of rationalization at work.
The story is the same for linguistic expectations. The average standardised effect of linguistic expectations on support for sovereignty is 0.09 (Models 4-6), while that of support on expectations is 0.26 (Models 10–12). The influence from expectations to support is much smaller than the influence in the opposite direction.Footnote 8
Conclusion
Two divergent views of the determinants of support for Quebec sovereignty coexist. One considers that people’s expectations regarding the economic and linguistic consequences of independence lead them to develop a favourable or unfavourable opinion about sovereignty. The other view considers that the attitude towards the desirability of sovereignty generates positive or negative expectations about the consequences of secession. Analyses based on cross-sectional survey data have so far been unable to ascertain what drives what.
In this research note, we rely on panel surveys and cross-lagged models to assess the accuracy of the two views. It turns out that both interpretations are partially valid. Some people use their predictions about the economic effects of sovereignty to decide whether or not they are in favour of independence. To a lesser extent (in one election out of three), this also applies to linguistic expectations. However, other individuals exhibit rationalization. Their economic and linguistic expectations simply stem from their constitutional preferences. This second dynamic is more important than the first. There is more rationalization than cost-benefit calculation in citizens’ decisions about Quebec sovereignty.
Our findings question the relevance of the rational choice perspective that puts much emphasis on cost-benefit evaluations as the primary determinant of support for or opposition to Quebec sovereignty. Yet, such evaluations do come into play, but their impact is limited. The question thus becomes: What does drive Quebeckers’ views about independence? We are not able to tackle that question in this research note. Our hunch is that attitudes about sovereignty, like many other attitudes, are shaped by feelings, in this case how strongly or weakly attached one feels about Quebec and Canada.
Open access
