Are national identities inherently exclusionary or is it possible that some national identities act as encompassing forces that include newcomers? On the one hand, research in Europe and in the US seems to point to the former: people who care about their national identity and people who identify more with their nationality are more likely to hold negative attitudes toward immigrants (see Sniderman et al., Reference Sniderman, Hagendoorn and Prior2004; Theiss-Morse, Reference Theiss-Morse2009; Wong, Reference Wong2010). On the other hand, the long-standing explanation for Canada's success with immigration has been the central place played by immigration and multiculturalism in its national identity, making Canadian national identity an inclusive force. Although never tested empirically, this explanation seems to be vindicated by recent research showing that Canadians exhibit a different relationship between pride in their country and attitudes toward immigration (Citrin et al., Reference Citrin, Johnston and Wright2012).
The starting point of this research is that the effect of priming a given national identity should be conditional on the norms associated with it. Echoing Citrin and colleagues (Reference Citrin, Johnston and Wright2012) on the outcome of patriotism and chauvinism, national identity in itself should not be conceived as having an exclusionary effect on immigration attitudes (see also Pehrson et al., 2009). Canada represents an ideal case to test these propositions because of the link that has been made between its national identity, multiculturalism and immigration (see Bloemraad, Reference Bloemraad2006: 243; Joppke, Reference Joppke2004: 244; Johnston et al., Reference Johnston, Banting, Kymlicka and Soroka2010: 369; Kymlicka, Reference Kymlicka2003: 375). If Canadian identity cannot have this inclusive effect, the prospects for other national identities playing such a role are dire. From a Canadian standpoint, if raising the salience of national identity has exclusionary effects on attitudes toward newcomers, it would challenge a long-standing view of Canada as a success story of host society-immigrant relations.
Here, I test the possibility that some national identities might represent an inclusive force by building on a previous survey experiment done in the Netherlands (Sniderman et al., Reference Sniderman, Hagendoorn and Prior2004; Sniderman and Hagendoorn, Reference Sniderman and Hagendoorn2007) and applying a modified version in Canada. This new survey experiment uses a randomly selected and nationally representative sample (N = 1500), combining the internal validity advantage of the experimental method with the external validity of this type of sample. It also represents a hard test of the inclusiveness potential of Canadian national identity by using a priming strategy that does not assign a given normative content to the identity in question.
Comparing two different primes, this research demonstrates that making Canadian identity salient does not increase anti-immigration attitudes even when the prime used as treatment strongly reinforces boundaries between groups. A subtler prime, one that makes national identity salient without attaching any meaning to it, made respondents more inclusive in some instances. In addition to directly testing the effect of a salient national identity on attitudes toward immigrant and multiculturalism, this research demonstrates the need to pay attention to the content of national identity across contexts and opens up new directions for future research by offering a framework that can be replicated in other settings.
National Identity, Social Identity Theory and Exclusionary Attitudes
National identity can be a slippery concept, and a useful starting point in making sense of it is the important distinction proposed by Citrin and colleagues between three dimensions of the concept. The distinction they propose is that national identity consists of a cognitive dimension (identification as), an affective dimension (identification with) and a normative dimension. The last refers to “the particular set of ideas about what makes the nation distinctive, ideas about its members, its core values and goals, the territory it ought to occupy and its relation to other nations” (2001: 75-76). When looking at citizens' attitudes toward immigration, a measure of citizens' identification with their country (affective) or as a member of their country (cognitive) is not enough. After all, strength of attachment or pride, in itself, has nothing to do with attitudes toward immigration. The effect of attachment and pride will be a function of what this national identity means for respondents. We thus need to pay attention to the normative dimension of national identity. It is this dimension that will draw a boundary determining where the immigrant finds herself vis-à-vis her new society. Attention to the normative dimension—or to what Abdelal and colleagues call “constitutive norms” (Reference Abdelal, Herrera, Johnston and McDermott2006: 696) and Schildkraut calls “identity content” (2011: 6) —is central to viewing national identities as social identities.
National identity as a social identity
Most of the literature that identifies the exclusionary effect of national identity is rooted in social identity theory (see, for example, Citrin et al., Reference Citrin, Wong, Duff, Ashmore, Jussim and Wilder2001; Crepaz, Reference Crepaz2008; de Figueiredo and Elkins, Reference de Figueiredo and Elkins2003; Sniderman et al., Reference Sniderman, Hagendoorn and Prior2004; Theiss-Morse, Reference Theiss-Morse2009; Wong, Reference Wong2010). Social identity theory (SIT) (Tajfel, Reference Tajfel1982; Tajfel and Turner, Reference Tajfel, Turner, Worchel and Austin1986) contends that individuals' sense of who they are is based on the groups they think they belong to and that identification with these groups will have consequences for individual behaviours. For example, members will make theirs the concerns and goals of the group and will behave in order to increase the well-being of that group (Brewer, Reference Brewer, Ashmore, Jussim and Wilder2001). In addition, because their sense of self is linked to the group, members will tend to evaluate the group favourably to increase their self-esteem. Attachment to the group is also central because the effect of group membership and the internalization of group norms will depend on the strength of identification (for a review, see Huddy, Reference Huddy2001). Viewing national identities as social identities is now common in political science (for example, Huddy and Khatib, Reference Huddy and Khatib2007; Schildkraut, Reference Schildkraut2011; Theiss-Morse, Reference Theiss-Morse2009).
It is not surprising that most of the research that has emerged from this literature points to the exclusionary consequences of national identity since SIT offers clear hypotheses as to the effects of social identities on intergroup relations. For example, it suggests that the ingroup favouritism associated with a social identity will lead to outgroup hostility when one's group's identity is considered to be under threat (Brewer, Reference Brewer, Ashmore, Jussim and Wilder2001; Coenders et al., Reference Coenders, Lubbers, Scheepers and Verkuyten2008; Sniderman et al., Reference Sniderman, Hagendoorn and Prior2004). Viewing national identity as a social identity also means that the norms associated with this national identity and attachment to it will be important factors in determining attitudes toward outsiders (Schildkraut, Reference Schildkraut2011). National identities will always exclude some and not others; not everyone can be American or French, and being American is seen as different from being French. These identities set boundaries between the ingroup and the outgroup and these boundaries will have clear implications at the individual level. For example, Wong finds that respondents with a more restricted vision of who is part of the American community are more likely to be harsher on who qualifies for citizenship benefits (2010: 137). There is also evidence that hosts will evaluate newcomers based on their adherence to norms associated with this particular national identity (Hainmueller and Hopkins, Reference Hainmueller and Hopkins2012; Schildkraut, Reference Schildkraut2011). Making national identity salient should reinforce these boundaries.
One of the best demonstrations of the effect of this normative dimension on immigration attitudes comes from a survey experiment conducted by Sniderman and colleagues in the Netherlands. In this study, one group of respondents had their Dutch identity primed resulting in what these authors have called a mobilization phenomenon (2004: 44): people who usually did not oppose immigration were more likely to do so after having received the treatment. In this case however, it is Dutch identity that is primed. The treatment effect should be context dependent and vary depending on the norms associated with a given national identity. In other words, the prime brings an image of “Dutchness” in the respondents' mind; it is directly connected to, as Anderson (Reference Anderson1983) would put it, the imagined boundaries of the Dutch people, one where immigrants seem to be left out of the inner circle.
Although most of the research rooted in SIT concludes that national identity acts as an exclusionary force, viewing national identity as a social identity does not automatically imply such a relationship. In fact, SIT does offer potential mechanisms through which (or instances where) national identity may act as an inclusive force.
National identity as an inclusive force
Following Theiss-Morse (Reference Theiss-Morse2009), there are at least two ways in which national identity may play an inclusive role and both are related to its normative dimension. The first one is derived from the Common Ingroup Identity Model (Gaertner and Dovidio, Reference Gaertner and Dovidio2000) and is applied to the question of national identity by Transue. The idea behind this model is that people can shift from an “us” versus “them” dynamic by focusing on a superordinate identity that encompasses a more general “we.” Transue observes in a US metropolitan sample, that priming national identity among white respondents increases support for a tax raise directed at educational opportunities for minorities. He concludes that his results support “research from social psychology that shows that attachment to broader identities reduces intergroup bias rather than with theories that predict that raising salience of national identity would lead to hostility toward outgroups” (Reference Transue2007: 89). In that particular case, however, the identity of the outgroup is likely to be an important factor. The design used by Transue makes African-Americans the target of these policies rather than immigrants, and the former might be more likely to be seen as part of the American society than the latter. Contrary to Transue's argument, we cannot assume that every national identity has the potential to act as a superordinate identity that brings every ethnic group inside a more general “we.” For example, Citrin and Wright (Reference Citrin and Wright2011) tested to see if a similar mechanism could be applied to white Americans' attitudes toward immigrants' protest. A protest where immigrants were waving American flags (as opposed to Mexican flags) lessened the degree to which people said they were bothered by the protest but this small effect did not translate to any changes in opinion for specific immigration policies. Some national identities may act as superordinate for some groups but this will depend on the normative content associated with it; the normative content of American national identity seems to include African-Americans whereas the normative content of Dutch national identity clearly does not include immigrants.
This leads to the second way in which national identity might a play an inclusive role: if the norms associated with it are themselves inclusive (Guimond et al., Reference Guimond, De Oliveira, Kteily, Lalonde, Pratto, Sidanius, Crisp, Kamiejski, Kuepper, Levin, Tougas and Zick2013; Pehrson et al., Reference Pehrson, Vignoles and Brown2009; Theiss-Morse, Reference Theiss-Morse2009: 182). Identification with one's national identity means following and internalizing the norms associated with it and as Theiss-Morse writes, “inclusiveness and appreciation of diversity can have significant effects on people's willingness to accept marginalized group members as full members of the group” (2009: 183). These two mechanisms through which national identity can be inclusive are interrelated. For a national identity to act as a superordinate identity a degree of inclusiveness as a norm has to be associated with it. The Canadian “exceptionalism” idea supposes that the norm of inclusiveness is associated with the Canadian national identity and that, as such, it has the potential to act as a superordinate identity that includes immigrants. Making Canadian national identity salient should therefore also raise the salience of the norm associated with it and influence attitudes toward newcomers. The goal here is to investigate this possibility experimentally, where salience of national identity is manipulated.
The Canadian Case
One of the main hypotheses to explain why Canada has been impervious to the backlash against multiculturalism is that the Canadian national identity is linked with multiculturalism and immigration.Footnote 1 For instance, Will Kymlicka writes, “While the actual practices of accommodation in Canada are not unique, Canada is unusual in the extent to which it has built these practices into its symbols and narratives of nationhood” (2003: 375). Others have made a similar point (Bloemraad, Reference Bloemraad2006: 243; Joppke, Reference Joppke2004: 244). So far, the evidence presented to support this hypothesis has been based on observational data and often limited in the outcome under study. According to many studies, support for multiculturalism in Canada is always high compared to other countries (Adams, Reference Adams2007; Berry and Kalin, Reference Berry and Kalin1995; Kymlicka, Reference Kymlicka, White, Simeon, Vipond and Wallner2008: 109; Parkin and Mendelsohn, Reference Parkin and Mendelsohn2003). These results are presented as a proof that multiculturalism and immigration hold a central position in the Canadian national identity. However, the wording of these types of questions often makes them problematic. For example, questions that directly ask about support for multiculturalism might be subject to social desirability bias and respondents might not know what the term multiculturalism means or interpret it differently.
In fact, when we look at questions from the International Social Survey Programme (ISSP) that do not mention multiculturalism, we find that in 2003 only 29 per cent of Canadian respondents thought that ethnic minorities should maintain their distinct customs and traditions. On government giving assistance to ethnic minorities, 66 per cent of respondents disagreed or disagreed strongly. In other words, if we only use support for multiculturalism policies or for its principle, we do not see a distinctive Canadian position. Figure 1 places Canada in a comparative perspective based on these two questions. It clearly shows that if there is a Canadian exceptionalism, it is not on support for multiculturalism policies or for its principle.
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160921111549-55462-mediumThumb-S0008423915000268_fig1g.jpg?pub-status=live)
Figure 1 Canada in comparative perspective
Two recent studies make an important step in testing the proposition that there is something different about Canadian national identity. First, Johnston and colleagues find that Canadian national identity is associated with pro-immigrants attitudes and that this identity also mitigates the negative effect that anti-immigration attitudes might otherwise have on support for the welfare state.Footnote 2 These results lead the authors to conclude that “the very self-conception of the country has come to embrace the idea of a multicultural society which successive waves of immigrants have helped build” (2010: 369). Citrin and colleagues followed with a study using observational data from the ISSP to compare the effect of national pride in the US and Canada. They find that in Canada, pride is positively related to support for multiculturalism but that this relationship is the opposite in the US. However, when the authors look directly at attitudes toward immigration and multiculturalism, Canadians are not more inclusive than Americans but rather less inclusive. This seemingly contradictory finding leads them to conclude that: “Canadian identity may foster an inclusive orientation but in doing so it starts from a less inclusive base” (2012: 547). The authors also look at the relationship between the same pride index and anti-immigration sentiment. Again, Canada differs from the US and exhibits a negative relationship. Canada, however, is not the only country exhibiting this relationship between pride in the country and immigration attitudes. The same data show that it is also true for France, Portugal and New Zealand.Footnote 3
In addition to this link between pride and more inclusive attitudes exposed by Citrin and colleagues (Reference Citrin, Johnston and Wright2012), Guimond and colleagues (Reference Guimond, De Oliveira, Kteily, Lalonde, Pratto, Sidanius, Crisp, Kamiejski, Kuepper, Levin, Tougas and Zick2013) show that Canada also differs in the perceived importance of the multicultural norm. Multiculturalism is perceived by Canadian respondents as a norm endorsed by the population. The proportion seeing multiculturalism as a norm is larger in Canada than in the US, the UK, and Germany. Again, however, Canadians' attitudes themselves are not that different from the attitudes of other Western countries. What seem to be different is the normative dimension of Canadian national identity and the inclusiveness that it promotes. It is this possibility that is tested experimentally in the following section.
Hypotheses
Based on what we know of the Canadian case and on the “national identities as social identities” literature, we can derive expectations about the effect of making Canadian respondents' national identity salient before answering questions on immigration and multiculturalism.
Social identity theory is clear on the effect of threat on outgroup hostility. Accordingly, even if the normative content of Canadian national identity acts as an inclusive force, we should expect respondents who feel like their national identity is threatened to be more opposed to immigration and multiculturalism (H1). The meaning of Canadian national identity should come into play, however, in the priming effect. Making respondents' national identity salient before asking them questions on immigration should remind them of the norms associated with such an identity. This means that Canadian respondents who receive the primes should not be more opposed to immigration and multiculturalism. In fact, one would expect them to become more supportive of both (H2). There is also the possibility that the priming effect will depend on some individual characteristics. For example, respondents for whom national identity is important will have internalized its norms more thoroughly and the strength of attachment to the Canadian national identity should interact with the prime. Thus, strong identifiers, those for whom Canadian identity is more important, should be more supportive of immigration and multiculturalism and we should see a more important treatment effect among them (H3).
Finally, the prime should not produce the “mobilization effect” found by Sniderman and colleagues in the Netherlands. Here the expectation is that the prime will not have a negative effect and this should be true at all levels of “cultural threat.” However, this cultural threat may diminish the effect of the norms associated with Canadian national identity. This means that priming Canadian national identity should not have a negative impact on attitudes toward immigration and this should be true at all levels of cultural threat. The primes should have a stronger inclusive effect at lower levels of threat (H4). The logic here is that the priming effect might not be strong enough to counterbalance the effect of feeling of threat. Thus, the effect will be more important at low level of cultural threat. In order to test these hypotheses, the following section presents a survey experiment that builds on prior work by Sniderman and colleagues in the Netherlands (2004; Sniderman and Hagendoorn, Reference Sniderman and Hagendoorn2007).
Experimental Design and Data
The objective of the design is to prime respondents' national identity while letting them decide what this national identity means. The survey experiment unfolds in three steps: a pre-treatment question, a priming treatment and a series of questions on immigration and multiculturalism. Figure 2 displays the complete design.
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160921111549-72237-mediumThumb-S0008423915000268_fig2g.jpg?pub-status=live)
Figure 2 Experimental design and question wording
The first element of the design is a question that asks respondents if they feel like Canadian values are threatened. The assumption here is that this measure of threat can serve as a proxy for a pre-existing bias against immigration without actually mentioning identity or immigration (Sniderman et al., 2004; Sniderman and Hagendoorn, Reference Sniderman and Hagendoorn2007). The treatment is then block-randomized on answers to this question. The logic behind this manipulation is that Sniderman and colleagues found that people who do not usually see immigration as a problem are most influenced by the prime. This manipulation controls for a “feeling of threat,” an obvious confounder of the effect of priming national identity, and makes it possible to look at heterogeneity in treatment effects.Footnote 4 One could argue that this question in itself represents a prime of the respondent's national identity but this pre-treatment question and the prime were separated by 30 unrelated questions. This pre-treatment question also voluntarily leaves open the source of the threat in order to not create or prime the association between threat and immigrants in the respondents' mind. For some respondents this feeling of threat might arguably be caused by factors other than immigration and diversity, such as current federal policies or the growing acceptance of more liberal lifestyles. However, if this pre-treatment question is associated with more exclusionary attitudes, it increases the confidence that it can be used as a proxy for pre-existing anti-immigration attitudes.Footnote 5
The treatment then takes two forms (plus a control group): one group receives a vignette replicating Sniderman and colleagues' study and another receives a question-as-treatment designed to prime the respondent's national identity and to measure the strength of attachment to this identity.Footnote 6 Using two different primes also allows for methodological advancement by testing the effect of different “strengths” of priming strategy. The vignette used by Sniderman and colleagues to prime national identity represents a strong prime that may be responsible for their results. Although their formulation mimics more closely what an anti-immigration politician might say, I argue that this prime is not a weak intervention or a “mere mention of consideration of collective identity” as they suggest (Sniderman and Hagendoorn, Reference Sniderman and Hagendoorn2007: 119). Telling the respondent that each nation is different and reinforcing the importance of this difference sharpens the boundaries between groups. It is not isolating the effect of national identity but lumping it with the effect of strong and exclusionary wording. Respondents in the question-as-treatment condition are asked how important their national identity is to them. This represents a mere mention of national identity, one that is more neutral. A design that compares primes of different strengths makes it possible to test if indeed, a mere mention of national identity induces exclusionary attitudes, or if these attitudes are only caused by a prime that raises the salience of group differences.
Once the respondents have received the treatment, they are asked to answer four questions on immigration and multiculturalism. The first three of these questions come from Sniderman and colleagues and the fourth comes from the International Social Survey Programme (National identity module). The first question is a direct measure of support for immigration while the other three measure different aspects of support for ethnic minorities and multiculturalism as a policy.Footnote 7 These questions take two forms depending on the respondent being in either of the treatment groups or in the control group: a treated respondent is asked for her answer “as a Canadian.” The treatment can be conceived as consisting in a prime (a preamble or a question-as-treatment) plus a repetition of “as a Canadian” in following questions.Footnote 8
The survey experiment was conducted over the phone and administered to 1500 Canadian respondents by the polling firm Research House between June 12 and June 23, 2012, as part of an omnibus survey. Respondents were selected through a random digit dialing sampling technique and they all reside in English Canada.Footnote 9 The following analysis is based on the full sample, which means that it is not solely composed of white English Canadians. An analysis was also done with a restricted sample (N = 1283) that excluded respondents whose first language learned and still spoken was not English and gave similar results.Footnote 10
Results
Looking at the distribution of answers on the four questions on immigration and multiculturalism, a consensus emerges on two of them whereas, on the other two, Canadians appear to be far more polarized. Figure 3 shows the distribution of preferences on these four questions in the full sample.
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160921111549-21200-mediumThumb-S0008423915000268_fig3g.jpg?pub-status=live)
Figure 3 Attitudes towards immigration and multiculturalism
Surprisingly, the two questions on which there is a consensus point in different directions. 83 per cent of the respondents agree (strongly or somewhat) that ethnic minorities living in Canada should have the same political rights than other Canadians. In contrast, 74 per cent disagree (strongly or somewhat) that ethnic minorities should receive government assistance to preserve their customs and traditions. It could be argued that the way this last question is framed raises more opposition because it mentions government assistance. Some respondents may oppose it not because they do not want ethnic minorities to preserve their customs and traditions but because they do not want to see an increase in public spending. Nonetheless, a basic idea of multiculturalism is that the government should at least fund ethnic groups' organizations to support cultural activities, bilingual education or mother-tongue instruction (Banting et al., Reference Banting and Kymlicka2006: 56–57). The opposition to government assistance is more important than in the 2003 ISSP sample where 66 per cent of respondents disagreed or disagreed strongly with the same statement (see Figure 1). Respondents are more divided on the other two questions where 41 per cent believe that the different ethnic cultures present in Canada are a threat to Canadian values and 54 per cent believe that immigration to Canada should be made more difficult than it is now. Based on these results, one could argue that Canadians are not overwhelmingly welcoming of immigrants and that they definitely do not support a basic tenet of multiculturalism: government financially supporting ethnic minorities. These results are in line with what Citrin and colleagues (Reference Citrin, Johnston and Wright2012) have found using data from the ISSP, which led them to conclude that if there was a Canadian “exceptionalism,” it was clearly not on basic attitudes toward immigration and multiculturalism.
Threat
For a country that has often been labeled as one of the most successful in dealing with ethnic diversity, it is quite striking to see the proportion of Canadians who feel as if their values are threatened. Almost 70 per cent of respondents agreed with the statement that these days, Canadian values were threatened. Although fifteen years separate this research and that of Sniderman and colleagues, it is interesting to note that in the Netherlands it was 51 per cent who somewhat agreed or agreed strongly with a similar statement. Even more striking is the 33 per cent in the Netherlands who strongly disagreed with the statement compared to 9 per cent in Canada. Arguably, these respondents may feel that their values are threatened by something other than immigration but the relationship between feeling of threat and anti-immigrant sentiment is strong. Figure 4 plots the mean response for the four questions on immigration at each level of threat.Footnote 11
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160921111549-44040-mediumThumb-S0008423915000268_fig4g.jpg?pub-status=live)
Figure 4 Threat and attitudes towards immigration and multiculturalism
As predicted by hypothesis 1, the more respondents feel as if their values are threatened the more they are opposed to immigration and to multiculturalism. This is especially true for questions 1 and 3, where respondents at low level of threat and those at high level of threat find themselves on opposite sides of the mid-point in the 0 to 1 scale (the difference between agreeing and disagreeing with the statement). In the case of question 1 for instance, it means that people who feel as if Canadian values are threatened are more likely to think that immigration to Canada should be made more difficult than it is now. It is worth noting that the two questions where threat is having less of an impact are the ones for which respondents are either overwhelmingly inclusive toward immigrants (question 2 on political rights) or exclusive (question 4 on government assistance). Even after controlling for known predictors of opposition to immigration (such as education), the “threat” measure remains a statistically significant predictor of exclusionary attitudes. Consequently, we can be confident that this pre-treatment question represents a useful proxy for measuring pre-existing anti-immigration attitudes and that it can be use to look at heterogeneity in treatment effects at different levels of threat.
Importance of national identity
The question-as-treatment (QAT) prime was designed to test for heterogeneity in treatment effect at different levels of importance of Canadian national identity. Unfortunately, the distribution of answers to this question-as-treatment turned out to have little variance and made it impossible to test hypothesis 3. Only 18 respondents out of the 538 in this treatment condition answered that their Canadian identity was not very important or not important at all.
Priming
The main expectation is that because the norm of inclusiveness is an important feature of Canadian identity, making this identity salient should increase inclusive attitudes. Looking at Figure 5, one can see that it is not the case for questions 1, 3 and 4.Footnote 12
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160921111549-69392-mediumThumb-S0008423915000268_fig5g.jpg?pub-status=live)
Figure 5 The effect of Priming on anti-immigrant attitudes
Importantly, we do not observe an exclusionary effect, either; the control group and the two groups who have received a prime do not differ. When the means differ, they do so slightly and the differences are far from being statistically significant. Priming Canadian identity had an inclusive effect in one instance. When asking whether minorities should have the same political rights than the Canadians people, respondents in the QAT condition were more likely to agree than people in the Sniderman treatment and in the control group. Both differences are statistically significant (two-tailed p = .046 and .016) but the difference between the Sniderman condition and the control group is not (two-tailed p = .64). This means that compared to the control group, asking respondents how important their Canadian identity was to them, made them more inclusive but that reading them a preamble emphasizing the importance of belonging to the Canadian nation did not have an effect important enough to be detected with a sample of this size.
It is interesting to notice that Sniderman and colleagues' design replicated in part here had one peculiarity: it asked respondents if they felt as if the Dutch culture were threatened before the treatment and asked them after the treatment if they as if like all the cultures present in the Netherlands were threatening the Dutch culture. In Canada, as reported earlier, many respondents felt as if Canadian values were threatened. If Canadian national identity played a truly inclusive role, we should see the prime having an effect on the second question. In other words, priming a Canadian respondent's national identity should remind him of how this identity is built around immigration and multiculturalism and make him see all those cultures present in Canada as less of a threat. But, here again, the hypothesis of no effect cannot be rejected; the effects are small and not statistically significant. One element that should be emphasized, however, is that adding “different cultures” as the threat to Canadian values shifts respondents to a more inclusive position: 31 per cent disagreed with the original threat statement compared to 58 per cent when “different cultures” are added as the threat. This result is likely due, at least partially, to a social desirability bias.
More importantly, Sniderman and colleagues (Reference Sniderman, Hagendoorn and Prior2004; Sniderman and Hagendoorn, Reference Sniderman and Hagendoorn2007) found that priming national identity had an effect for people who did not feel as if the Dutch culture was threatened a priori, creating what they called the mobilization phenomenon. The prime made them more likely to agree that immigration should be made more difficult. Contrary to the Netherlands and, as expected, the prime increases the proportion of Canadians who strongly disagree with the statement. However, the hypothesis of no effect cannot be rejected. Here, the means for the two treatment groups are lower than the mean for the control group (.34 and .35 compared to .38) but these differences are far from being significant (two-tailed p > .6 for all differences). The effect size is also small compared to the results in the Netherlands.Footnote 13 The Canadian sample only has 76 respondents at a low level of threat. With the effect size found in Canada, a sample of 7698 respondents would be necessary to reach p < .05. Thus, both the sample size at low level of threat and the small effect size in Canada compared to the Netherlands are responsible for this statistically non-significant result.
Figure 6 plots the means on the four dependent variables by levels of threat for each condition.
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160921111549-10320-mediumThumb-S0008423915000268_fig6g.jpg?pub-status=live)
Figure 6 Priming, threat and anti-immigration attitudes
Almost all of the differences in means for individual questions are not statistically significant. The only exceptions are the differences between the QAT condition and the Sniderman condition on question 2 (political rights) for high and low levels of threat. Both at low level and at high level of threat, respondents in the Sniderman condition were more opposed to ethnic minorities getting the same political rights than other Canadians compared to those in the QAT condition (high threat: QAT = .23, Sniderman = .32, two-tailed p = .04; low threat: QAT = .07, Sniderman = .19, p = .05). In both cases, the control group finds itself in the middle but without any statistically significant difference to the treatment groups. This makes it difficult to evaluate if this result is due to the QAT treatment having an inclusive effect, the Sniderman treatment having an exclusionary effect or both.
So far, results have been presented in a straightforward manner, in the form of “difference-in-means.” When analyzing experimental data, simple procedures are often preferable to more complex models that require assumptions (Dunning, Reference Dunning, Collier and Brady2010), but it can be useful to include covariates as control in a multivariate regression framework when these pre-treatment covariates strongly predict the outcome (Freedman, Reference Freedman2008; Green, Reference Green2009: 19).
Table 1 shows the results for these multivariate regressions. After controlling for important covariates, the treatment effects remain small and far from being statistically significant with one exception.
Table 1 Treatment Effects in a Multivariate Context
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160921111549-01285-mediumThumb-S0008423915000268_tab1.jpg?pub-status=live)
p < .1; **p < .05; ***p < .01
This table shows multivariate regressions for all four dependent variables. Given that all four questions are four-category Likert scales, coefficient estimates (and standard errors) are from an ordered probit regression. Reference categories for explanatory variables are only elementary schooling for Education and rural community for size of community. Treatment conditions are compared to the control group.
For the multiculturalism policy question (question 4), being in the QAT condition decreases the likelihood of strongly disagreeing with the statement that ethnic minorities should receive government assistance by 8 per cent and increases the likelihood of being in the somewhat agree and strongly agree categories by 4 per cent and 2.5 per cent respectively. Even though this is the only result that reaches the usual threshold of statistical significance, it is important to note that specifying multivariate models for the different dependent variables results in negative effects for the QAT treatment (more inclusive attitudes) and positive effects for the Sniderman treatment (more exclusive attitudes) for three out of the four questions. When compared to the Sniderman prime, the QAT condition also has a statistically significant effect on both question 2 on political rights and question 4. The main point remains, however, that even after controlling for covariates the hypothesis of no effect cannot be rejected in most cases.
To summarize, in most cases, priming national identity did not have a statistically significant effect on attitudes toward immigration and multiculturalism. The effect sizes were small and were far from the conventional threshold for statistical significance making it impossible to reject the hypothesis of no effect. However, when the QAT prime had an effect, it made respondents more inclusive and inversely, when the Sniderman prime had an effect, it was an exclusionary one. The difference in the direction of the effect for the two primes is interesting and represents an important contribution. It shows that, contrary to Sniderman and Hagendoorn's argument, their prime does not represent a “mere mention of consideration of collective identity” (2007: 119) but rather reinforces boundaries. This is potentially due to the fact that the preamble reminds the respondent that nations are different and that this difference is a crucial component of national identity. However, it is important to note that in many instances even this highly restrictive prime was not enough to move Canadians in an exclusionary direction. On the other hand, a true “mere mention” of national identity, asking the respondent how important his national identity is to him, when it had an impact, made the respondent more inclusive. The analysis also shows that priming the respondents' national identity as Canadian either through the Sniderman prime or the QAT prime did not completely offset the exclusionary effect of feeling of threat on attitudes. The more people feel like their values are threatened the more they show exclusionary attitudes.
Conclusion
This paper investigates the relationship between national identity and exclusionary attitudes toward immigration in the Canadian context. The “real-world” importance of this question lies in the fact that the exclusionary effect of national identity is what makes possible the “flash politics” surrounding immigration issues (Sniderman and Hagendoorn, Reference Sniderman and Hagendoorn2007). Anti-immigration politicians use national identity as a mobilization tool to gather supporters and gain votes. One mechanism through which this chauvinistic discourse has an impact on attitudes is that it primes national identity among the population, makes salient and moves the imagined boundary of the nation. Results obtained by Sniderman and his colleagues in the Netherlands showed that this was particularly the case for people who were usually not opposed to immigration: priming their identity as Dutch made them more opposed to immigration. Replicating this experiment in Canada did not produce the same results. Although Canadians are not as welcoming of immigrants as would be expected from a country that prides itself on its successes with immigration, priming Canadians' identity did not make them more opposed to immigration. However, in most cases, the prime did not make them more welcoming either.
This is surprising when considering recent studies showing that pride in Canada is positively related to support for multiculturalism and immigration (Johnston et al., Reference Johnston, Banting, Kymlicka and Soroka2010; Citrin et al., Reference Citrin, Johnston and Wright2012) and that multiculturalism is also perceived as the accepted norm in the country (Guimond et al., 2013). It is important to note, however, that the design presented here represents a hard test of the inclusive potential of Canadian national identity. The objective of this study was to see the picture that emerged when no content was attached to a national identity made salient other than the content attached by the respondents themselves. An easier test would have been to prime the inclusiveness itself or insist on immigration as a foundation of Canadian identity (see for example Esses et al., Reference Esses, Wagner, Wolf, Preiser and Wilbur2006). In addition, by letting respondents decide what it means to be Canadian, the design might be priming different conceptions of this identity. In most countries “multiple traditions” (Schildkraut, Reference Schildkraut2011) or different constitutive norms will coexist and priming national identity will have a different impact depending on which of these competing norms the respondent adheres to when thinking about immigration issues. It is highly possible that there is heterogeneity in what Canadians see as the normative content of national identity when it comes to the role of immigrants and that these different conceptions are cancelling each other. This, in itself, presents a more complex vision of Canadian national identity than one that places immigration and multiculturalism as defining features. Future research could prime respondents' national identity by asking them to list a series of keywords that they associate with this identity, making it possible to measure the content as well.
Another possible explanation for the absence of more important treatment effects is that perhaps priming national identity has an asymmetrical impact on attitudes toward immigration and multiculturalism; priming an exclusionary national identity has a negative impact on attitudes but priming an inclusive national identity does not make these attitudes more inclusive. This idea of asymmetry in the effect of national identity is in line with what Citrin and Wright (Reference Citrin and Wright2011) found in the US.
Small treatment effects aside, this research represents an important step in trying to make sense of how national identity and its normative dimension relate to immigration attitudes in a comparative perspective. Although it presents evidence for a single country, its aim is to highlight the need for more comparative work on the effect of national identity on attitudes (Pehrson, 2009; Wright, Reference Wright2011). The results presented here make it clear that national identities that have been able to include immigration as one of their defining aspects might not have the same effect on attitudes. The role played by the normative dimension of a given national identity is somewhat in the background when looking at determinants of anti-immigrant attitudes in a given country and comparative work brings it to the forefront. The next step would be to extend the framework and apply it to other countries of immigration. The province of Quebec represents another interesting case because its citizens can identify both as Canadians and as Quebeckers. Quebec has also seen far more heated debates around immigration and integration than the rest of Canada. These differences would make it possible to prime respondents' identity as Canadians (hypothesized as being more inclusive) and as Quebeckers (hypothesized as being more exclusionary) and compare their effect on immigration attitudes.
Finally, this research also adds to the literature linking cultural threat with exclusionary attitudes toward newcomers. Surprisingly for a country that has presented diversity as one of its defining features, the number of Canadians respondents who feel like Canadian values are threatened is considerable. Not only are people agreeing with the “threat” statement when the source of this threat is left open (69 per cent) but an important proportion of respondents (41 per cent) are willing to admit to the interviewer that they feel like Canada's values are threatened by the different cultures represented in the country. These results also demonstrate that priming Canadian national identity does not damper the effect of cultural threat. The strong relationship between this feeling and anti-immigration views is unequivocal and suggests the need for more research on the determinants of emotions such as feeling of threat and anxiety (Brader et al., Reference Brader, Valentino and Suhay2008). This should give pause to Canadian triumphalism regarding its success with immigration and multiculturalism. One of the established findings in the literature on attitudes toward immigration is that cultural threat is one of the most important driving forces of anti--immigration attitudes. Consequently, Canadians' attitudes toward immigration may not be as firmly and unequivocally welcoming as we might think. What the results presented here show, however, is that even with an important perception of cultural threat among the Canadian population, Canadian national identity does not have the exclusionary potential of many of its European counterparts.
Appendix A
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary-alt:20160921111549-25406-mediumThumb-S0008423915000268_fig7g.jpg?pub-status=live)
Figure A1 National pride and anti-immigration attitudes