Toward a Theory of Separatism in Post-Industrial Democracies 1 Contents The Quebec separatist movement 2 Data and Analysis 3 Generational Effect................................... 3 Period Effect...................................... 5 Latent Variable..................................... 6 Discussion 7 Toward a Theory of Separatism in Post-Industrial Democracies The Quebec Case Charles Tessier, Université Laval Yannick Dufresne, Université Laval Clifton van der Linden, University of Toronto Introduction September 5th 2015 The autumn of 2014 saw both Catalonia and Scotland hold referenda on independence. Although the Scottish independence movement was defeated having secured but 45% of the vote more than 80% of the votes cast in the Catalonian participation process favored an independent state. Turnout in the Catalonian case was notably low, most likely as the result of the referendum being non-binding. Nothwithstanding the failure of recent referenda to produce meaningful advancements towards independence, in both cases support for independence movements remains high. Junts pel Sí (JxSí), the Catalan independence coalition, maintains a sizable lead in opinion polls in advance of the Catalan parliamentary election. This mirrors the sweep by the Scottish National Party (SNP) in the 2015 United Kingdom general election. Whether this momentum can be harnessed so as to achieve the raison d être of these movements remains to be seen, but the Quebec case can be instructive as to the ebb and flow of support for independence. By many measures the heyday of the Quebec independence movement has passed. It too saw its political wing achieve stunning victories in previous decades, most notably during the 1990s with the return of the sovereigntist Parti Québécois to power and the emergence of the movement s federal arm, the Bloc Québécois, assuming status in the House of Commons
Toward a Theory of Separatism in Post-Industrial Democracies 2 as the Official Opposition, culminating in a referendum on independence in 1995 which proponents of a sovereign Quebec lost by a razor-thin margin. The ostensible waning of ambitions for independence in Quebec came into clear focus in the 2011 Canadian federal election, which saw the Bloc Québécois suffer a crippling defeat at the hands of the federalist New Democratic Party. Reduced to four seats in the House of Commons, the Bloc Québécois lost official party status and prompted questions both within Quebec and the rest of Canada about whether Quebec independence was merely a generational project. In this paper we interrogate the idea of Quebec independence as a generational project by examining the evolution of the Quebec independence movement. We draw together multiple data sources to explore how public sentiment towards independence in Quebec has changed over time, speculating as to causal mechanisms which may be responsible for said change. The findings from the paper form the basis of a general theory of support for separatism in post-industrial democracies, one which can be tested in cases such as Scotland and Catalonia. The paper proceeds in three parts. First, we provide an overview of the history of the independence movement in Quebec and review the relevant literature in the field. Second, we draw on multiple data sources to explore attitudes towards independence among Quebeckers, both past and present. Third, we employ a public opinion survey to develop a multi-item scale that offers a more nuanced measure of attitudes for independence in Quebec, making the case for said scale as a superior measure to the classic binary. The Quebec separatist movement The roots of the separatist movement in the province of Quebec date back to 1759, when the British conquered New France (Fenwick, 1981), setting a scene for tensions between native Anglophones and Francophones that continues to resonate today. But the idea of Quebec as a nation, a prerequisite for any separatist movement, began to take its contemporary form around the time of the Quiet Revolution in the 1960s. During this period, Quebeckers began drawing on the provincial construct of Quebec as a means to express their national identity (Balthazar, 2013; McRoberts, 1999). These events prompted the formation of the Parti Québécois the primary political vehicle of the separatist option and catapulted it to power in the 1976 Quebec provincial election. Two threads emerge in the academic literature on Quebec independence. The first encompasses efforts to explain the persistence of the movement and the variation in its support. Notwithstanding two failed attempts by the Parti Québécois to see Quebec secede from the Canadian federation by way of referendum, Millard (2008) notes that support for independence with an offer of partnership continues to percolate at between 40-50 percent, suggesting that the present state of the movement is best characterized as a ceasefire in rather than a decisive resolution of the wars over national unity. Attempts to explain the persistence of the movement have had limited success (Drouilly, 1997; Pinard, Bernier and Lemieux, 1997). Richez and Bodet (2012) argue that it is due to the entanglement of federal and provincial party effects. Several factors were considered, however, to explain variations in the support for independence. These include, among others, political crisis (Mendelsohn, 2002), the strengh of the separatist leadership (Pinard, Bernier and Lemieux,
Toward a Theory of Separatism in Post-Industrial Democracies 3 1997), and anticipation of economic gain (Blais, Martin and Nadeau, 1995). In addition, Drouilly (1997) describe the support for independence as strata. Within this conceptualization there are two outer boundary strata comprised of strong sovereigntists and strong federalists, respectively, and multiple inner strata comprised of voters with a less resolute opinion on Quebec independence. The second thread in the literature concerns the demographics of supporters themselves. Many variables were studied, but despite recent emphasis on the generational aspect of the Quebec independence movement, no consensus exists in the literature as to the effect of age on support for independence. According to some studies, the effect of age is more of a period effect. In other words, younger people tend to support independence more than older people (Cloutier, Guay and Latouche, 1992; Nadeau, 1992). Alternatively, some argue that the effect of age is rather a cohort effect a reflection of common, generational experience (Crête, 1984; Martin, 1994). According to Pinard, Bernier and Lemieux (1997), citizens that have been socialized after 1960 demonstrate more favorable views toward independence than citizens that have been socialized before 1960. Finally, some scholars opt for a middle ground position, and claim that both a generational effect and a period effect are present (Monière and Guay, 1996). However, the literature in this area is especially sparse, often leaving the public to make inferences from commercial polls commissioned by media organizations, which are interpreted to suggest that the independence movement is the product of two generations: Baby Boomers and Generation X (Gagnon, 2014a). Younger Quebeckers, especially those between the ages of 18 and 25, are generally considered to be much less supportive of the idea of independence(asselin, 2014; Gagnon, 2014b). Whether a generational or a period effect is reponsible for this dynamic requires further empirical investigation. Data and Analysis For the purposes of this analysis, we take age as our variable of interest to ascertain whether the Quebec independence movement is subject to generational or period effects. We use data from the Canadian Election Study and from the /em Boussole électorale, or Vote Compass, an online civic engagement launched during the 2011 Canadian federal election, as well during the 2012 and 2014 Quebec provincial elections. The survey-based application generated nearly 100,000 usable observations for this study. It offers a unique opportunity for fine-grained analysis of attitudes towards independence in Quebec. Generational Effect Some studies on the demographics of support for Quebec independence argue that the effect of the age variable is solely or partly a generational effect, or a cohort effect. To test for this effect we apply a standard OLS estimator to respondent data from the Vote Compass dataset, specifically the proposition that reads Quebec should become an independent state. Responses are registered on a Likert scale using the categories of strongly disagree, somewhat disagree, neutral, somewhat agree, and strongly agree. We set
Toward a Theory of Separatism in Post-Industrial Democracies 4 this as the independent variable in our regression and include age, gender, education and language as our dependent variables. The age variable is treated as a factor variable. In other words, the model will include a dummy variable for each date of birth available. This way we avoid the imposition of constraining assumptions on the age variable, but its effect is controlled for with other sociodemographic variables. Because of the extraordinary sample size, we can operationalize age as a continuous variable and thus a factor in a parametric estimator as opposed to recoding into ordinal categories, which can distort the results on the basis of arbitrary categorization. In the absence of such a large sample, researchers are often forced to poll data from several surveys fielded decades apart, which can distort the results and stymie efforts to identify the causal mechanisms at play. An addition advantage of the Vote Compass dataset is that responses are measured on a Likert five-point scale as opposed to the classic binary option on support for independence included in most public opinion research in the area. The results of the regression analysis are presented in Figure 1 [Figure 1 about here.] It is worthwhile to note that the graph includes date of birth up until 2000. While citizens born after April 1996 did not have to right to vote in the last federal or provincial elections, they were included in the graph since their responses can be of interest as citizens who will eventually vote. The graph also indicates the delineation between the different generations. The first conclusion we can draw from Figure 1, even before looking specifically at the generations, is the cyclic nature of the support for independence. According to these data, we can already put aside the idea that independence is the project of one or two subsequent generations. Then when we look at the different generation on the graph, we can see that some part of the cycle do fit somewhat with the generations. The cuts are far from perfect, but trends are still observable. It is worth to recall that these generations were not included in any way in the model and that not assumptions of this kind were impose on the age variable. The tendency in the the pre-boomers and X s years seem to be toward federalism, while the tendency for the Boomers and the beginning of the Y s years seem to be toward independence. These are tendencies only because it is clear by looking at the graph that the generations are not homogeneous blocks. Any attempts to label the generations as separatist or federalist is an oversimplification. The other interesting thing on this graph is the Y generation. According to our data, Y s born before 1994 are in the high section of the cycle, while other respondents of the same generation born later clearly move toward the bottom of the cycle. Most of these respondent did not have the right to vote when they answered this question, but it will be interesting to follow these voters when they enter the electorate. If this trend is observe in the future surveys like this one, it might be an indication of a new generation. To sum up, support for Quebec independence might be cyclical, roughly fitting with the generations. We could therefore conclude there is a generational effect in the relation between age and independence. In addition, the younger voters cannot be tagged as a group opposed to the idea of an independent Quebec. In fact, some of the younger voters are in the peak of the cycle. In the next section, we will move forward with the second part of this three-step analysis, where we will try to tackle the question of the period effect.
Toward a Theory of Separatism in Post-Industrial Democracies 5 Period Effect The objective in this second part of the analysis is to verify if there is a period effect related to the support to independence. In other word, is the fact of getting older affect respondents from all generation the same way. In a perfect world, we would have Vote Compass-like data going back 40 years, and we could run similar analysis. Unfortunately, this application is relatively new and the first data come from the 2011 federal election. We can hardly measure any period effect over a so short period of time. We will then use the data from the Canadian Election Study (CES). The question on Quebec independence was asked for the first time in 1968 and in nine other surveys since then 1. This number of surveys and the time span they cover allows us to study the period effect. The nature of the data of the CES is quite different from the Vote Compass data we used in the previous section. First, the sample size is more limited. The number of respondents in each surveys is between 379 and 1251. These sample sizes make it impossible to treat the age variable as we did in the last section. Second, contrarily to the Vote Compass, respondents in the CES had to answer Yes, No or don t know to the question. Given these differences the analysis will be a little different in this section. The objective here is to compare respondents with the same age at different point in time. For example, we want to compare 20 years old respondent in 1968, 1979, 1988, etc. To do that, we will run a basic logit model for each of the survey. The independent variable set will include gender, education, language and age. Then several predictions of the support for independence will be compute for each model keeping every variable constant except the age variable which will be increase in each new iteration of the prediction. Then we will compare the predictions for each survey and will be able to determine if there is a common trend for all surveys. The results are presented in Figure 2 [Figure 2 about here.] The y-axis on Figure 2 indicates the chances a respondent have to support independence, and the x-axis is the values of the variable used to compute the predictions. Contrarily to the last section when we used the date of birth, we now use the actual age, so the graph is inverted (young voters on the left, older voters on the right). Each line represent a unique survey. The predictions show a common tendency for a large majority of the surveys included in our analysis. The chance of supporting independence decrease slightly as the age variable increase. Obviously, the trends of each surveys don t perfectly fit, the 2011 survey for example don t fit with the others, but in general the period effect appears pretty clearly. We can therefore conclude there is, in addition to a generational effect, a period effect. 1 The question was not include in the 1974, 1980 and 1984 surveys
Toward a Theory of Separatism in Post-Industrial Democracies 6 Latent Variable In the first to part of this three-step analysis, we concluded to the presence of a generational effect and a period effect on the support for independence. In this third and last part, we move from a conception of this question as being a dichotomous response as we would find on a referendum ballot. We will rather look at the support for independence as an attitude. The objective here is to assess how polarized voters are regarding this question, and more specifically, if the polarization is different among younger and older cohorts of voters To consider the support for independence as an attitude as important implications on the way we measure it. According to?, to take a person s response to any one question as evidence of a theoretical true attitude seems a grievous simplification. Therefore, using only one question, as we did in the last two sections may not be the best. It s even worst if the question offers a dichotomous choices set, since we can t measure the strength of the attitude with such a question. In this third part of the analysis, we will use a series a question we consider relevant to measure support for independence as a complex attitude and use these to build an index. An index is the ideal tool to measure the strength of an attitude and the distribution of a population. To do that we will use a Canada-wide survey conducted by the Vox Pop Labs firm in 2014. The Quebec subsample include 512 respondents. To create the index, we use a standard Factor analysis on twelve questions we selected in the survey. Before presenting any results, it is important to verify if the twelves question seem to measure the same latent variable. The eigenvalue of the first dimension of the factor analysis is 7.82 and the Cronbach alpha is 0.87. These diagnostic seems to indicate that the questions fit nicely together and measure the same attitude 2. Now that we have an index, we will compare how younger and older voters are distributed on it. Voters between 18 and 35 years old are included in the younger voters category. Figure 3 present the distribution for both groups. [Figure 3 about here.] The two samples used for the figure above are the same size, therefore any differences between the two groups are solely due to their respective distributions and not because of the sample sizes. Generally speaking, this figure confirm that strata description of the support for Independence described by Drouilly (1997). We also see that this polarization is true for both group, but less accentuated for younger voters. According to these data younger voters are less polarized on this issue than their older counterparts and that they are, on both side, more ambivalent on this issue. This finding have implication on the way we measured support for Quebec Independence among younger voters. In a more highly polarized setting, using a question with a dichotomous response set, as it is the case in most commissioned surveys, may produce accurate results. This may not be the case however in less polarized setting. This may results in an inaccurate assessment of the support for Independence of younger cohorts of voters. 2 The complete list of question and the diagnostic are available in the Appendix.
Toward a Theory of Separatism in Post-Industrial Democracies 7 Discussion It is difficult to say what exactly will happen in Scotland after the 2014 referendum. Even if London do not deliver what they pledged, the movement could either weakened until it is not a political force anymore or thrive and lead in the future to another referendum. The relation between support for independence and the age of the supporters might be the key to have a better idea of the movement future. How this variable evolve over time while newer cohort of voters replace older ones is of first importance to understand the evolution and the persistence of separatist movements. The objective of this paper was to use the case of Quebec, an older movement, to study this variable and somewhat maybe have a glimpse to the future of the movement in Scotland. We used a three-step analysis to study the relation between age and support for Independence on three different point of view. In the first part, we used the data from the Vote Compass to study the generational effect. The sample size allowed us not to put any constraint on the age variable and the five-point scale of the dependant variable gave more flexibility to the analysis. We found that the effect was mostly cyclic and the changes in support fits roughly with the generations. In the second part, we used the data from the Canadian Election Study to study the period effect. These data did not offer the same flexibility as the Vote Compass, but it had the advantage of returning back to 1968. We compared similar age cohort over time and found evidence of a period effect. In the third part of the analysis, we used a survey conducted by Vox Pop Labs in which several question related to support for independence were asked. Using these questions, we created an index to study how voters are polarized on the issue, and more importantly, is there a difference between younger and older voters. We found that older voters are more polarized than yougner voters, who seems to have less strong attitude, on both side of the index. Finally, this paper makes important contributions. First, it update the scientific literature on the support for independence in Quebec since not much was published on the topic in the last 15 years. Most of what we know these days on the topic rather come from newspapers analysis of opinion survey they commissioned. This paper offers more reliable analysis that meets scientific standards. In fact, the contribution of this paper is not merely to update the literature. By using the Vote Compass data, we perform analysis with levels of precision and flexibility that were not possible before. Lastly, these analysis might inform on the way some of the separatist movement may evolve over time.
Toward a Theory of Separatism in Post-Industrial Democracies 8 References Asselin, Pierre. 2014. Le parti aveugle. Le Soleil June 10th 2014:17. Balthazar, Louis. 2013. Nouveau bilan du nationalisme au Québec. Montréal: VLB Éditeur. Blais, André, Pierre Martin and Richard Nadeau. 1995. Attentes économiques et linguistiques et appui à la souveraineté du Québec: une analyse prospective et comparative. Canadian Journal of Political Science 28(4):637 657. Cloutier, Édouard, Jean-Herman Guay and Daniel Latouche. 1992. Le virage: comment le Québec est devenu souverainiste. Montréal: Québec/Amérique. Crête, Jean, ed. 1984. Comportement électoral au Québec. Chicoutimi: Gaëtan Morin éditeur. Drouilly, Pierre. 1997. Indépendance et Démocratie. Montréal: Harmattan. Fenwick, Rudy. 1981. Social Change and Ethnic Nationalism: An Historical Analysis of the Separatist Movement in Quebec. Comparative Studies in Society and History 23(2):196 216. Gagnon, Katie. 2014a. Le projet de deux générations. La Presse June 3rd 2014:4 5. Gagnon, Katie. 2014b. Radiographie politique d une génération. Le Droit June 2nd:15. Martin, Pierre. 1994. Générations politiques, rationalité économique et appui àla souveraineté au Québec. Canadian Journal of Political Science 27(2):345 359. McRoberts, Kenneth. 1999. Quebec: Social Change and Political Crisis. Toronto: Oxford University Press. Mendelsohn, Matthew. 2002. Measuring national identity and patterns of attachment: Quebec and nationalist mobilization. Nationalism and Ethnic Politics 8(3):72 94. Millard, Gregory. 2008. Secession and Self: Quebec in Canadian Thought. Montreal: McGill- Queen s University Press. Monière, Denis and Jean-Herman Guay, eds. 1996. La bataille du Québec - Troisième épisode: 30 jours qui ébranlèrent le Canada. Boucherville: Fides. Nadeau, Richard. 1992. Le virage souverainiste des Québécois, 1980-1990. Recherches sociographiques 33(1):9 28. Pinard, Maurice, Robert Bernier and Vincent Lemieux. 1997. Un combat inachevé. Sainte-Foy: Les Presses de l Université du Québec. Richez, Emmanuelle and Marc André Bodet. 2012. Fear and Disappointment: Explaining the Persistence of Support for Quebec Secession. Journal of Elections, Public Opinion and Parties 22(1):77 93.
Toward a Theory of Separatism in Post-Industrial Democracies 9 Figure 1: Generational Effect on Support for Independance Separatist Neutral Federalist 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Pre boomers Boomers X's Y's 1966 1967 1968 1969 Year of Birth
Toward a Theory of Separatism in Post-Industrial Democracies 10 Figure 2: Period Effect on Support for Independence 1.00 0.75 Pr(Support for Quebec Separation) 0.50 1968 1979 1988 1993 1997 2000 2004 2006 2008 2011 0.25 0.00 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 Age
Toward a Theory of Separatism in Post-Industrial Democracies 11 Figure 3: Distribution of voters on the separatist index by age cohort 160 120 Density 80 Older Younger 40 0-2 -1 0 1 2 Separatist Scale