Freedom of speech is a fundamental value in democratic politics. Citizens are expected to respect the rights of others to express themselves publicly, regardless of whether they endorse the ideas espoused by others. The capacity to do this is captured by the concept of political tolerance, which is usually defined as “a willingness to ‘put up with’ those things one rejects or opposes. Politically, it implies a willingness to permit the expression of ideas or interests one opposes” (Sullivan et al., Reference Sullivan, Piereson and Marcus1982: 2). Traditional liberal definitions of tolerance require that all groups—even groups that fundamentally challenge a minority group's right to exist—be tolerated. Although there has been recognition that political tolerance is often at odds with other democratic values (Marcuse, Reference Marcuse, Wolff, Moore and Marcuse1969; Nelson et al., Reference Nelson, Clawson and Oxley1997; Peffley et al., Reference Peffley, Knigge and Hurwitz2001; Sniderman et al., Reference Sniderman, Fletcher, Russell and Tetlock1996), its current conceptualization does not fully account for the ways in which individuals distinguish various types of speech.
This is clearly the case when free expression collides with concerns about preventing discrimination in contemporary, multicultural democracies. There have been successive legislative and legal attempts, especially since the 1980s, to criminalize prejudicial behaviour in industrialized countries. This can be seen in employment equity legislation, anti-discrimination provisions and legal cases that challenge discriminatory behaviour on the part of governments and organizations. It can also be seen in the trend toward regulating racist speech, with hate speech legislation being present in most European countries and in Canada (Coliver et al., Reference Coliver, Boyle and D'Souza1992).
Such legislative restrictions are completely at odds with current conceptualizations of political tolerance as a citizen value. While attitudes toward free speech have been examined empirically since the 1950s (Stouffer, Reference Stouffer1963), there have been few attempts to examine specific target groups or the relationship between them. In this article, I argue that exclusionary speech—such as incitement of racial hatred, Holocaust denial and other forms of hate speech—poses a fundamental challenge to how we think about, and in turn measure, political tolerance. Target group distinctions are an under-specified source of variation in political tolerance judgments.
To examine these contentions, I rely on a unique comparative dataset of young people in Canada and Belgium. Young people in these two countries have grown up during a period of unprecedented racial and ethnic diversity and under legislative regimes that, at least officially, place severe sanctions on the expression of hatred. Given this, youth in these two countries are considered critical cases (Eckstein, Reference Eckstein, Greenstein and Polsby1975) for testing the presence of inter-target group distinctions in political tolerance judgments.
The findings, in general, support the idea that the youngest generation is balancing the need for social inclusion with individual rights to speech, rather than siding consistently with individual rights (absolute tolerance) or consistently censoring speech across the board (intolerance). A substantial proportion of youth see hate speech as outside the realm of legitimate democratic debate, while still permitting the expression of other objectionable ideas. This distinction is at the heart of the concept of multicultural political tolerance developed in this article.
Defining Political Tolerance
Tolerance is traditionally understood to imply restraint when confronted with a group or practice found objectionable (Cohen, Reference Cohen2004; Heyd, Reference Heyd1996; Horton and Nicholson, Reference Horton and Nicholson1992; Mendus, Reference Mendus1988, Reference Mendus1989; Sullivan et al., Reference Sullivan, Piereson and Marcus1979). Political tolerance typically refers to individual-level attitudes that permit groups to express opinions or maintain practices that a majority finds objectionable. It thus refers to the willingness to refrain from preventing people (or groups of people) from expressing their disliked opinions, lifestyles, preferences or world views (McKinnon, Reference McKinnon, McKinnon and Castiglione2003: 55–61; see also Walzer, Reference Walzer1997; Weissberg, Reference Weissberg1998). In practice, a citizen's degree of tolerance is assessed based on whether people agree that controversial groups should be allowed to participate in expressive activities, such as giving public speeches, holding rallies, or having books in local libraries. When respondents agree to extend civil liberties, they are providing tolerant answers. When they disagree, it is considered intolerant.
Two key features of this definition of political tolerance are the presence of prior disagreement and content neutrality. First, political tolerance is essentially about overcoming objection, and thus is only relevant for situations of disapproval (Sullivan et al., Reference Sullivan, Piereson and Marcus1979). If a person agrees with, or is indifferent toward, a viewpoint, tolerance is not applicable because there is nothing to which the person objects. The other feature of conventional political tolerance measures is their unified nature. While the targets of tolerance must provide an opportunity for objection on the part of the respondent, there has been little attempt to distinguish between types of objectionable speech. Although there has been recognition that the targets of intolerance vary by country (Sullivan et al., Reference Sullivan, Shamir, Walsh and Roberts1985), only a few studies have focused on specific types of target groups, such as extreme religious sects, racists or pornographers (Chong, Reference Chong2006; Davis, Reference Davis1995; Gross and Kinder, Reference Gross and Kinder1998; Lambe, Reference Lambe2004; O'Donnell, Reference O'Donnell1993). Much like American First Amendment jurisprudence, the concept of political tolerance has been constructed as essentially content neutral.Footnote 1
Prior objection and content neutrality are important concepts in the current measurement of political tolerance. These standards isolate the concept of tolerance from acceptance and ensure that both right-leaning and left-leaning citizens have an opportunity to express intolerance. Yet, from a comparative perspective, the emphases on prior disagreement and content neutrality create a situation in which citizens are considered intolerant even when the laws of their country permit certain types of censorship.
This is evident for what is commonly referred to as hate speech, usually defined as speech intended to incite hatred or promote genocide of minority groups. Legislation and legal interpretations that place restrictions on the expression of hate have been widely adopted in contemporary democracies, with the United States being the most notable exception (Cohen-Almagor, Reference Cohen-Almagor2000; Coliver et al., Reference Coliver, Boyle and D'Souza1992; Douglas-Scott, Reference Douglas-Scott1999). Defenders of such restrictions usually focus on the impact of hate speech on minorities. For example, Matsuda has argued, “The negative effects of hate messages are real and immediate to victims. Victims of vicious hate propaganda experience physiological symptoms and emotional distress … Victims are restricted in their personal freedom … As much as one may try to resist a piece of hate propaganda, the effect on one's self-esteem and sense of personal security is devastating” (Reference Matsuda, Matsuda, Lawrence, Delgado and Crenshaw1993: 24–25). Such a perspective is supported by recent work dealing with the effects of hate crimes and hate speech on victims (Boeckman and Turpin-Petrosino, Reference Boeckman and Turpin-Petrosino2002).
Hate speech also arguably plays a role in perpetuating hate organizations and hate crimes. Sumner (Reference Sumner2004: 162–63), for example, sees hate speech as serving primarily as a means of recruitment for and identification with hate organizations that have been linked to racial violence.Footnote 2 In studies of genocides and large-scale discrimination, hate speech serves to stigmatize a group and normalize discriminatory treatment of them (Bosmajian, Reference Bosmajian1974; Cortese, Reference Cortese2006; Mullen, Reference Mullen2001; Tsesis, Reference Tsesis2002). At the individual level, there is evidence that witnessing ethnic slurs or other derogatory comments directed at a minority can make majority group members feel more negatively toward them (Greenberg and Pyszczynski, Reference Greenberg and Pyszczynski1985; Kirkland et al., Reference Kirkland, Greenberg and Pyszczynski1987; Simon and Greenberg, Reference Simon and Greenberg1996).
In other words, hate speech is viewed as a distinct type of expression. Unlike other potentially objectionable speech, hate speech serves to negatively impact the psychological and physical well-being of racialized minorities, effectively impacting the equal ability of people to enjoy the rights accorded to them as citizens. This distinction is fundamental, and is well captured by Harel's distinction (Reference Harel and Heyd1996: 122) between inclusionary intolerance and exclusionary intolerance. Inclusionary intolerance arises in circumstances where minority groups are trying to fully participate in society by restricting the expression of prejudice directed at them. In other words, inclusionary intolerance involves restricting the rights of the intolerant. Preferential hiring regulations would be an example of inclusionary intolerance; such regulations limit the rights of employers to hire (or rather not hire) whom they want. Indeed, most anti-discrimination legislation could be considered as inclusionary intolerance as it limits the right of people to act on their biases in areas like employment and housing. Restrictions on hate speech, similar to other anti-discrimination measures, are a form of inclusionary intolerance. They limit individuals' right to express themselves in a manner that arguably hinders the full participation of others. Harel argues that inclusionary intolerance (restricting the rights of the intolerant) is more easily justified based on liberal democratic norms than other restrictions on rights.
The success of hate speech laws in effectively countering the societal problems they are said to address is, of course, a contentious issue (Braun, Reference Braun2004; Coliver et al., Reference Coliver, Boyle and D'Souza1992). Suppressing the expression of ideas, as Locke informed us long ago, does not necessarily change the ideas people hold, yet the extent of support for hate speech legislation and its relationship to other forms of intolerance is an empirical question. Current conceptualizations of political tolerance fail to recognize the possibility that restrictions on certain forms of speech are democratically defensible and are in fact restricted in most advanced industrialized democracies.
This shortcoming is particularly evident in attempts at assessing citizens' attitudes toward free speech, where there has been little effort to incorporate distinctions between inclusionary and exclusionary censorship practices. Typically, political tolerance is assessed either through the summing of responses to civil liberties questions about various target groups, similar to the early scales developed by Stouffer (Reference Stouffer1963) or by asking citizens which groups they dislike and then measuring their willingness to extend civil liberties to this group, referred to as the least-liked method of Sullivan and colleagues (Reference Sullivan, Piereson and Marcus1979). Both measures reflect the objective of content-neutrality, and indeed the least-liked method is designed specifically to be “content controlled” (Sullivan et al., Reference Sullivan, Marcus, Feldman and Piereson1981). Neither method for assessing political tolerance allows for the consideration of distinctions between various types of target groups, yet these two measurement techniques have been widely used to study citizens' political tolerance levels (see, for example, Chong, Reference Chong2006; Gibson, Reference Gibson1998, Reference Gibson2006a; Marcus et al., Reference Marcus, Sullivan, Theiss-Morse and Wood1995; Marquart-Pryatt and Paxton, Reference Marquart-Pryatt and Paxton2006; Mondak and Hurwitz, Reference Mondak and Hurwitz1998; Mutz, Reference Mutz2002).
Using citizens' support for the protection of racist speech as a test of political tolerance, then, may underestimate a citizen's commitment to the ideals of free speech. A willingness to curb racist speech may reflect the adoption of a more general approach to political expression that views some ideas as outside the realm of democratic discourse, as indeed many democratic governments and courts have maintained. One of the consequences of content neutrality in the measurement and operationalization of political tolerance is that we know little about the extent to which citizens view hate speech as distinct from other types of objectionable speech. The basic puzzle motivating this research is whether a willingness to restrict hate speech is simply a manifestation of political intolerance like any other or if individuals can be politically tolerant while still placing limits on specific types of speech. Unlike past research on political tolerance, this study rests fundamentally on the idea that speech associated with the promotion of hatred is viewed as particularly detrimental to democratic politics.
Redefining Tolerance
Rather than a view of political tolerance that prioritizes content neutrality, Harel's distinction (Reference Harel and Heyd1996), as well as the comparative juridical prominence of hate speech restrictions, suggests that political tolerance should not be conceptualized as a binary concept where one is either intolerant or tolerant. Rather, conceptual room must be made for those who distinguish between the speech of the intolerant and other objectionable ideas. Censoring speech is usually problematic because it serves to (or at least has the potential to) restrict free and open debate that is considered fundamental to the democratic process. Yet some speech arguably serves exclusionary purposes and aims to delegitimize the voices and experiences of marginalized communities in the democratic process. A redefinition of political tolerance judgments that takes this into consideration results in three possible types of responsesFootnote 3:
1. Intolerance. These individuals do not support speech rights for any objectionable group. Most research on political tolerance is actually concerned with this group of individuals.
2. Multicultural tolerance. Individuals who support speech rights for objectionable groups, but do not extend them to groups that promote hatred.
3. Absolute tolerance. Individuals who extend speech rights, irrespective of the target group.
While intolerance and absolute tolerance are well captured by current conceptualizations of political tolerance, the idea of multicultural political tolerance relies on the ability of citizens to distinguish hate speech from other speech that they find objectionable. Those who make such distinctions cannot possibly be captured using current techniques. They would be either categorized as intolerant if they chose a racist group as their least-liked using Sullivan and colleagues' content-neutral strategy (Reference Sullivan, Piereson and Marcus1979), or as somewhat intolerant based on a Stouffer-like scale. Yet, given the presence of such distinctions in the legal regimes of most industrialized countries, it seems essential to ask whether citizens are capable of (and in fact do) make such distinctions. I refer to this as multicultural political tolerance because it reflects contemporary discourses about the inclusion of various ethnic, racial and religious communities in the democratic process (Kymlicka, Reference Kymlicka1995, Reference Kymlicka2001; Young, Reference Young1990).Footnote 4
This typology, in part, addresses a recent concern expressed by Mondak and Sanders (Reference Mondak and Sanders2003: 496–97), who argue that absolute tolerance (that is, tolerance for all groups) is fundamentally different from variations in intolerance. They argue that when individual tolerance judgments are simply summed together across target groups, those who are tolerant of some are fundamentally different than those who respond tolerantly to all groups. They recommend only considering the extreme of the scale as tolerant, in line with the concept of absolute tolerance noted above.
The problem with their approach is that there is no consideration of how variations across target groups are conceptualized, leaving everything but the extreme of the scale as representing simple variation in intolerance. The definition developed here, in contrast, includes a theoretically driven explanation of this variation by focusing on the types of groups in the scale. Furthermore, this framework specifies the types of target groups driving the variation. The measure controls for disagreement and is not limited by the number of groups in total that the respondent finds objectionable. While Mondak and Sanders may consider this distinction between groups simply a specification of levels of intolerance, the distinction—if found to reflect significant patterns in line with the legislative norms of democratic politics today—is more fundamental than that.
Case Selection, Data and Methods
In order to test whether there is evidence that citizens do in fact make distinctions between hate speech and other types of objectionable speech, this study relies on a critical case studies approach (Eckstein, Reference Eckstein, Greenstein and Polsby1975). As the concept of multicultural political tolerance represents a significant departure from conventional conceptions of political tolerance, this study provides a first test relying on samples in which I most expect to find inter-target group distinctions. The case countries for consideration are Canada and Belgium. A key criterion for their inclusion in this study is that both have civil and criminal laws prohibiting hate speech. The presence of such legislation should make it more likely that citizens adopt similar limits on speech rights. Along with the presence of such legislation, both these countries also share structural characteristics, including two major linguistic communities that are regionally concentrated and a federal system of government. These similarities reduce the possibility that inter-country differences in political tolerance can be attributed to variation in structural characteristics.
The critical difference between the countries, and comparative advantage of this study, is that Canada has a longer history of multicultural policies and much higher levels of racial and ethnic diversity than Belgium. While the primary goal of this article is to assess if citizens are capable of distinguishing between hate speech and other types of objectionable speech, the study also provides an important opportunity to explore the circumstances in which it is most likely to emerge. Canada was the first country to adopt an official policy of multiculturalism. In part due to open immigration policies, its actual levels of racial, ethnic and religious diversity are higher than most advanced industrialized democracies, including Belgium. Belgian politics have been marked by more open hostility to racial and ethnic diversity, notably in the popularity of the Vlaams Blok, a right-wing anti-immigrant party which has garnered substantial portions of the vote there (Billiet and De Witte, Reference Billiet and De Witte1995). While both countries provide crucial case studies for the presence of inter-target group distinctions, their differences also provide some leverage in explicating the institutional and social bases of multicultural tolerance.
This initial test also focuses on a specific age cohort. Given the importance of early experiences to the socialization of political values and behaviours (Gerber et al., Reference Gerber, Green and Shachar2003; Hooghe, Reference Hooghe2004; Plutzer, Reference Plutzer2002), the youngest generation is considered the most likely to espouse new norms around the limits of free speech in multicultural democracies. Evidence suggests that attitudes toward diversity in these countries have become increasingly open over time, and this appears to be particularly true among the younger generations (Inglehart, Reference Inglehart1997; Nevitte, Reference Nevitte1996; Wilkes et al., Reference Wilkes, Guppy and Farris2008). Furthermore, there is some evidence that more supportive attitudes toward diversity and multicultural policies have been accompanied by increased hostility toward hate speech among younger people (Chong, Reference Chong2006; Harell, Reference Harell2008). If some citizens do distinguish between hate speech and other types of objectionable speech, they would most likely be found among youth in countries with strong legal frameworks restricting hate speech.
The Comparative Youth Survey (CYS) provided the data used in this analysis (Stolle and Hooghe, Reference Stolle and Hooghe2006). Surveys were conducted with students in grades 10 and 11 in Canada and Belgium during the 2005–2006 school year. In Canada, students were sampled in schools from seven cities in Ontario and Quebec.Footnote 5 Six cities were selected to vary in terms of size and were “matched” across provinces. The largest city was selected in each province (Toronto and Montreal), along with two medium-sized cities of approximately 150,000 inhabitants and two small towns with approximately 15,000 inhabitants. A second small town was also surveyed in Ontario that included a substantial French-speaking minority to facilitate inter-language and inter-province comparisons. Schools were intentionally selected to vary in terms of the socio-economic status of students and the homogeneity of the student population.Footnote 6 In the medium and small towns, all school boards were contacted and an effort was made to survey as many schools as possible in each setting. In total, 3334 respondents completed the self-administered questionnaire. Within each city, the socio-economic and linguistic backgrounds of the students are similar to the city in which they were sampled, and the distribution of schools approximates the language and public/private distribution of schools in the cities.Footnote 7
The Belgian sample was a stratified sample of secondary schools in ten provinces in the French and Flemish communities, with an over-sampling of five additional Dutch-speaking schools in Brussels. The schools were randomly selected and match the distribution of school types present. In total, 6265 students completed the survey. The average age of respondents in both surveys was 16 years old.
The main variables of interest derive from a tolerance battery. Modified from commonly used tolerance batteries, the goal was to include a number of potentially objectionable groups that differ in the exclusionary nature of their speech, their ideological association and their salience in the two contexts. The final battery includes five different potentially objectionable groups: racists, skinheads, radical Muslims, gay rights activists and Quebec/Flemish separatists. Racists and skinheads were included to represent groups associated with hate speech.Footnote 8 For each group, the respondent is asked to indicate whether they should be allowed to 1) hold a peaceful march in the respondent's neighbourhood and 2) talk on public television about their views. The answer categories are dichotomous (yes or no).
Importantly, the respondent was also asked to indicate their level of agreement or disagreement with each group on an 11-point Likert scale. The inclusion of this last item allows replication of a modified version of the least-liked methodology created by Sullivan and colleagues (Reference Sullivan, Piereson and Marcus1979) where respondents pre-select their most objectionable group. Unlike the method developed by Sullivan and colleagues, where respondents are required to rank groups by level of dislike, this question format allows the respondent to find multiple groups equally objectionable and therefore allows for a controlled comparison across different objectionable groups. Its advantage over traditional Stouffer-like batteries where responses are summed across groups is that disagreement can be controlled.Footnote 9
Based on these questions, a tolerance score is calculated for each target group and activity pair by limiting the analysis to individuals who expressed prior disagreement with the group.Footnote 10 Unlike studies that simply sum the number of tolerant responses, this methodological approach allows for a comparison of the nature of tolerance decisions and, importantly, whether young citizens make distinctions between the two hate groups in the survey and the other three groups, while controlling for the varying levels of disagreement across target groups. A composite measure based on these items is then developed to represent the three typologies presented earlier and a brief exploration of the significant demographic and political correlates of multicultural tolerance is provided. These include the level of political activism (0 to 5 or more), political knowledge (scale from 0 to 1 based on 3 questions), organization involvement (0 to 4 or more organizations), gender (1=female), racialized minority status (1=non-white respondent), urban/rural (1=urban), parental education level (1=one or both parents university educated), religious affiliation (non-religious is reference category) and religious attendance (0=never to 4=more than once a week).
Target Group Distinctions and Tolerance Norms
If individuals do make distinctions between exclusionary speech and other types of objectionable speech, one would expect that the aggregate level of tolerance for exclusionary groups would be lower than for other objectionable groups. Figure 1 presents the levels of tolerance for each of the two civil liberties activities in each country. The results are presented only for respondents who disagreed (and highly disagreed) with each group in order to more accurately capture the concept of tolerance. Note that the levels of tolerance for talking on television are expected to be higher than for peaceful marches because the proximity of the latter activity makes it more threatening—an “in your face” activity that has more potential to affect the respondent and would be harder to ignore if it occurred (Gibson and Bingham, Reference Gibson and Bingham1985; Marcus et al., Reference Marcus, Sullivan, Theiss-Morse and Wood1995). Figure 1 bears this expectation out. Indeed, in almost every single case, the tolerance level in each country is higher in the television scenario than it is for the peaceful march scenario.
Figure 1 Percent Tolerant by Activity, Country, and Disagreement Level
Source: Comparative Youth Study, 2005–2006.
What is more noteworthy is the variation across target groups within each scenario. First, the percentage of respondents in Canada and Belgium who were willing to tolerate each group talking on television about their views follows the hypothesized pattern: racists and skinheads, who represent exclusionary groups, are less likely to be tolerated than radical Muslims, gay rights activists or separatists.Footnote 11 In the Canadian sample, racists and skinheads receive the lowest levels of tolerance, with only about one in five young people permitting them to talk on public television about their views. This is in contrast to the other three target groups that receive between 35 and 63 per cent tolerance. In the Belgian data, racists and skinhead again receive the lowest levels of tolerance (17 and 15 per cent respectively) whereas over twice as many respondents were willing to allow gay rights activists and Flemish separatists to talk on television about their views. Radical Muslims, on the other hand, only receive slightly higher levels (19 per cent).Footnote 12 The gap between the most and least tolerated groups is more than 40 points in the Canadian sample, and over 20 points among Belgian youth.
A similar pattern emerges when respondents were asked if each group should be allowed to hold a peaceful march in their neighbourhood. As expected, levels of tolerance for each group are lower for this more threatening activity, but the distribution across groups is consistent with the talking on television scenario. Almost 30 points separate the tolerance levels of racists and skinheads from gay rights activists and separatists in the Canadian sample, and over 16 points separate these groups in the Belgian sample.
Some might argue that this variation is driven solely by the fact that young people in both countries find hate groups more objectionable. Respondents did disagree most heavily with racists and skinheads in the full sample.Footnote 13 This is hardly a surprising finding, and in fact reflects the primary contention of this research: exclusionary speech is more objectionable to many young people precisely because it conflicts with other values, like social inclusion. As one might expect when limiting the sub-sample to only those who highly disagree with a target group (that is, where the respondent reported a 0 on the 0–10 disagree/agree scale), tolerance levels for those groups that were less objectionable on average decrease more substantially than for groups that were already found, on average, to be more objectionable.Footnote 14 For example, willingness to allow gay rights activists to talk on television drops from 50 per cent to 32 per cent among Canadian youth. The parallel drop for racists and skinheads is only a couple of percentage points. Thus, one of the reasons for the variation across target groups is the difference in the levels of aggregate disagreement.
That being said, this reduction in variation does not radically change the observed pattern, suggesting that level of disagreement is an important control but does not fully explain the aggregate level variations. Individuals continue to be less willing to extend civil liberties to exclusionary groups than to other groups, even under the most stringent conditions (high disagreement X high threat). Clearly, more young people find it harder to tolerate exclusionary speech than other forms of speech. As expected, these gaps are largest in Canada. While the samples across countries are not identical, and thus prevent a rigorous test of differences between the countries, the greater divergence between exclusionary and other objectionable groups in Canada is suggestive. Canada's longer history of multicultural policies and substantially higher levels of racial and ethnic diversity, especially in Central Canada from which the sample is drawn, should promote the free speech balancing act consistent with multicultural tolerance. Yet, in both Canada and Belgium, target group variation is clearly present, albeit at different levels.
The evidence provided so far only demonstrates this distinction at the aggregate level. One of the strengths of the question format developed here is it allows an examination of how individuals who disagree with multiple groups make distinctions between them. On average, respondents indicated some level of disagreement with just over three groups, of which two on average were rated highly objectionable. Racists and skinheads were among the group many respondents' objected to, but the vast majority of those finding at least one exclusionary group objectionable also disagreed with one of the remaining three groups. In fact, 80 per cent of respondents disagreed with at least one of each type of group, allowing for a comparison of tolerance judgments at the individual level.
Table 1 (Canadian data) and Table 2 (Belgian data) capture these distinctions by presenting the differences in the percentage tolerant of one target group compared to another target group limiting each dyad to only those individuals who found both groups objectionable.Footnote 15 Percentages close to zero imply that when respondents tolerated (or not) the group listed in the column, they responded similarly to the group listed in the row. A higher percentage indicates the difference in tolerance levels between the column and row groups for each activity. If exclusionary speech is less tolerated than other types of speech, one would expect a higher positive percentage among dyads that include both an objectionable group in the column and an exclusionary group in the row. This implies that individuals were more likely to tolerate objectionable speech than they were to tolerate exclusionary speech. Likewise, tolerance values between pairs of objectionable groups or between pairs of exclusionary groups are expected to be closer to zero.
Table 1 Difference in Tolerance Levels within Group Dyads in Canada
Note: The percentage is the percentage point difference in the aggregate level of tolerance for the row group compared to the column group. Positive numbers indicate greater tolerance for the column group, and negative numbers indicate higher levels of tolerance for the row group. Within each pair, only respondents who disagreed with both groups were included. The number is represented by the n for each pair of groups.
Source: Comparative Youth Study, 2005–2006.
Table 2 Difference in Tolerance Levels within Group Dyads in Belgium
Note: The percentage is the percentage point difference in the aggregate level of tolerance for the row group compared to the column group. Positive numbers indicate greater tolerance for the column group, and negative numbers indicate higher levels of tolerance for the row group. Within each pair, only respondents who disagreed with both groups were included. The number is represented by the n for each pair of groups.
Source: Comparative Youth Study, 2005–2006.
Both Table 1 and Table 2 present patterns consistent with these expectations. Turning first to tolerance of talking on television in the Canadian sample, consider the dyads that contrast tolerance levels of gay rights activists and separatists with skinheads and racists in the television scenario. These are found in the lower left-hand corner of Table 1. For individuals who disagree with both groups in the dyads, tolerance levels are still between 27 and 45 percentage points higher for gay rights activists and separatists than for the two exclusionary groups. In contrast, the difference that emerges between tolerance levels of gay rights activists and separatists is only 8 percentage points in favour of Quebec separatists. Similarly, the difference between tolerance of racists and skinheads is only 4 percentage points. Dyads that include radical Muslims are less consistent. While they do receive higher levels of tolerance than skinheads and racists (14 and 17 percentage points respectively), they receive substantially lower tolerance levels than gay rights activists and separatists.
The results for holding a peaceful march indicate an almost identical pattern, although the differences are less dramatic as might be expected in the higher threat scenario. The largest differences (between 14 and 29 points) emerge between exclusionary groups and gay rights activists and separatists, even after limiting the comparisons to only those who find both groups objectionable.
A similar pattern emerges in the Belgian data. Considering young people's tolerance of a target group talking on television, a difference of over 20 percentage points exists between those tolerating either gay rights activists or Flemish separatists and each of the exclusionary groups. It is noteworthy that similar gaps in tolerance also exist in this sample between gay rights activists or Flemish separatists and radical Muslims. This suggests that there is an added distinction being made with respect to radical Muslims. The results for holding a peaceful march mimic this pattern. Importantly, the results in Tables 1 and 2 control for prior disagreement with groups (hence the different sample size in each dyad) and provide strong support for the contention that young people distinguish exclusionary speech from other objectionable speech.
What does this mean for our understanding of political tolerance judgments? Clearly, some young people do make distinctions across target groups when deciding whether or not to extend certain civil liberties. This is particularly evident for civil liberties activities that are more distant and less “in your face.” These differences are not simply an artifact of varying levels of objection. Rather, it seems that some groups are viewed as more legitimate participants in democratic debate, in spite of any objection to the point of view being expressed.
Multicultural Political Tolerance among Youth
Clearly, there is evidence that at least some young people are distinguishing between hate speech and other objectionable speech. This is consistent with the concept of multicultural political tolerance, yet to fully highlight the extent of target group distinctions, it is necessary to identify the extent to which youth are making these distinctions. For this, a categorical variable has been developed that captures three types of individuals in the CYS study that correspond to the three types of tolerance dispositions developed here: intolerance, multicultural tolerance and absolute tolerance. A respondent who is unwilling to allow at least one exclusionary group and at least one other objectionable group to hold a march and talk on television is coded as intolerant. Conversely, a respondent who is willing to allow both an exclusionary group and another objectionable group to hold a march and talk on television is coded as an absolute tolerator. Finally, those who allow at least one objectionable group to do both civil liberty activities but deny them to at least one exclusionary group are coded as multicultural tolerators.
A choice was clearly made to force respondents to allow at least one objectionable group to participate in both of the civil liberties activities asked about in the survey, rather than at least one of the two activities. This was intentional in order to ensure respondents applied rights judgments in a principled manner, rather than simply responding to the threatening stimulus of the march scenario. This is also an attempt to make this a stricter test of the hypothesis that tolerance distinctions emerge based on target group distinctions (and not distinctions between situations where tolerance is asked to be applied).Footnote 16
Table 3 provides the breakdown by each category. It demonstrates that multicultural tolerance is, in fact, characteristic of a substantial portion of young people's thinking. In the Canadian sample, 55 per cent of respondents are categorized as multicultural tolerators when any level of prior disagreement is used as the basis for determining tolerance judgments. Another 33 per cent responded in an intolerant manner, and only 11 per cent were willing to extend civil liberties judgments across groups.Footnote 17 Similarly, about 11 per cent of the Belgian sample qualifies as absolute tolerators. The difference is that in Belgium, the intolerant account for 49 percent of the sample compared with 40 percent who were coded as multicultural tolerators.Footnote 18
Table 3 Breakdown by Type of Tolerance
Note: The disagreement breakdown excludes 1598 respondents who did not find at least one of each type of target group objectionable, as well as 1482 respondents who failed to complete the tolerance battery. Significance calculated using a two-group mean comparison t-test (one-sided) where *** means p<.01.
Source: Comparative Youth Study, 2005–2006.
The difference in levels of multicultural tolerance between Canada and Belgium are significant (p<.01) and in the hypothesized direction. While sampling differences cannot be fully ruled out as a source of these differences, the evidence is at least suggestive that countries with greater openness toward racial and ethnic diversity should be more likely to view hate speech as intolerable. Given the importance that ethnocultural diversity plays in Canadian discourses around identity, it should not be surprising that young people in that context are more likely to fall into the multicultural tolerance category.
The prominence of multicultural tolerance, especially in the Canadian context, raises the question of who is most likely to make such distinctions. Clearly, the extensive literature on political tolerance points to a number of key correlates of political intolerance (for reviews, see Gibson, Reference Gibson2006b; Sullivan and Transue, Reference Sullivan and Transue1999). Here, I focus primarily on demographic and political correlates of intolerance. In terms of demographic characteristics, gender, urban/rural status, education and religiosity are examined. Previous research has shown that men, those living in urban areas, those with more education and less religiously involved express greater political tolerance (Marcus et al., Reference Marcus, Sullivan, Theiss-Morse and Wood1995; Nie et al., Reference Nie, Junn and Stehlik-Barry1996; Stouffer, Reference Stouffer1963). There is also an expectation that involvement in the political system is supposed to foster knowledge of the rules of democratic politics as well as facilitate the ability of individuals to apply general democratic principles to specific situations (Finkel and Ernst, Reference Finkel and Ernst2005; Fletcher, Reference Fletcher1990; Peffley and Rohrschneider, Reference Peffley and Rohrschneider2003).Footnote 19 These finding have largely been reproduced among youth samples, with the exception of gender (Sotelo, Reference Sotelo1999, Reference Sotelo2000).
If multicultural tolerance is, in fact, a unique tolerance disposition, then the expectation is that the traditional correlates of tolerance should help to distinguish multicultural tolerance from intolerance. Furthermore, I have argued that multicultural tolerance does not simply represent a midpoint on a scale from intolerance to tolerance, either. If multicultural tolerance reflects contemporary norms of political tolerance, then one might expect that traditional correlates of intolerance should do little to differentiate multicultural from absolute tolerance.
To test these contentions, Table 4 presents separate multinomial logistic regressions. The reference category for the analysis is multicultural tolerance, which means that the results should be read essentially as tests of the impact each independent variable has on the likelihood of intolerance or absolute tolerance compared to the reference category. The coefficient should be interpreted as providing the direction of effects.
Table 4 Political and Demographic Correlates of Tolerance
Note: Multinomial logistic regressions are presented, where multicultural tolerance is the reference category, and standard errors are adjusted for clustering. The variable for Jewish has been dropped in the Belgian regression due to small subsample size.
*** p<.01;
** p<.05;
* p<.10;
a p<.15.
Source: Comparative Youth Study, 2005–2006.
The first contention is that traditional correlates of intolerance should distinguish multicultural tolerance from intolerance, and this appears to largely be reflected in the models presented in Table 4. As expected, the intolerant in both countries appear to be less politically knowledgeable and engaged compared to those in the multicultural tolerance category (p<.01). They are also likely to attend religious services more often compared to multicultural tolerators. Like Sotelo (Reference Sotelo1999), I find that young women have a greater probability of being tolerant of the civil liberties of some groups compared to men. There are some inter-country differences that emerge, although in each case the significant effect goes in the expected direction. In short, traditional correlates of intolerance seem to distinguish the intolerant from multicultural tolerators, despite the fact that these individuals would likely be considered intolerant based on a least-liked approach.
Perhaps more enlightening are the effects of these variables on the likelihood of absolute tolerance compared to multicultural tolerance. As mentioned previously, the alternative to a least-liked methodology is to sum the number of tolerant responses across target groups. Such an approach would make multicultural tolerance a midpoint on the scale, as respondents have provided some tolerant responses and some intolerant responses. The implication would be that multicultural tolerators should be found “between” the intolerant and absolute tolerators on many of the important correlates of intolerance. However, the results in Table 4 shed doubt on such a linear interpretation. In both the Canadian and Belgian data, fewer coefficients are significant, but more importantly, the direction of these effects are opposite of what would be expected from a linear view of political tolerance. At least in the Canadian case, the likelihood of absolute tolerance decreases as youth are more politically active and knowledgeable about politics. Conversely, those who display greater religiosity are more likely to fall into the absolute tolerance category. At the same time, those from Catholic and Jewish background were less likely to be in this category, compared to multicultural tolerance. Racialized minorities, as might be expected given the role of hate speech in the categorization of multicultural tolerance, were also less likely to be absolutely tolerant.
In some ways, then, multicultural tolerators seem to share the democratic qualities of absolute tolerators—they are as, if not more, knowledgeable and engaged in politics—while their social backgrounds distinguish them equally well from the intolerant. Multicultural tolerators are distinct from the intolerant, despite the fact that the least-liked methodology would often categorize them as intolerant. A targeted intolerance of racist speech, then, appears fundamentally different than intolerance of other speech. This casts doubt on a linear conception of political tolerance and points instead to the benefits of examining the role that inter-target group distinctions play in understanding contemporary civil liberties judgments. Developing a measure that can capture such distinctions opens up new avenues of research into what type of people are most likely to set limits on exclusionary speech. The preliminary examination of the correlates of multicultural tolerance provides support for the view of multicultural tolerance as a unique disposition more akin to tolerance than intolerance. It also casts further doubt on the ability of conventional measures to fully capture how the next generation is making civil liberties decisions.
Balancing Rights in Multicultural Democracies
Democratic politics is a balancing act. In multicultural democracies, this balancing act sometimes brings the rights of individuals into conflict with the rights of groups. The public expression of exclusionary ideas is such an instance: individual rights to free expression must be balanced against the rights of minorities to live free from harassment and prejudice. In many advanced, industrialized democracies, there is room for the courts to decide in favour of the rights of minorities in such instances, despite the overwhelmingly absolute nature of free speech in much of the political science literature. As Horton notes, “What we need to recognize is that any inculcation of the virtue of toleration (and any coherent form of multiculturalism) must attend to questions about what it is reasonable to object to, as well as about which of those things that are objectionable should be tolerated and which should not” (Reference Horton and Heyd1996: 37) .
In contemporary democracies, one thing that most would agree is unreasonable to object to is skin colour or ethnic origin. On the other hand, it seems perfectly reasonable, and indeed desirable, to object to racial and ethnic prejudice. The normative question, for academics as well as citizens, then becomes whether to tolerate the latter, given the unreasonableness of the former. The answer to this question is contested because both responses at their core have a desire to ensure freedom and facilitate the healthy functioning of democratic politics.
While the normative implications of a more multicultural form of tolerance are beyond the scope of this article, the empirical evidence presented suggests that the study of political tolerance needs to move beyond questions of the degree of tolerance and intolerance to the ways in which people distinguish across target groups. It is clear that a substantial portion of young people do indeed favour some limits on speech in line with the legislative norms in their countries. Hate speech has a legal status as a prohibited form of speech in many countries, and the evidence presented here suggests that many young people in Canada and Belgium recognize it as such. They are in turn far less willing to permit its public expression, despite the fact that they are generally tolerant of other objectionable ideas. This distinction is at the heart of the concept of multicultural political tolerance.
While a rich research tradition exists into the correlates of political intolerance, an understanding of politically tolerant attitudes in multicultural democracies requires researchers to examine the nature of the limits people place on speech rights. While this article has developed a typology of tolerance dispositions based on theoretically driven distinctions between types of speech and has presented an initial test of the demographic and political correlates, future research will need to unpack the causes of such distinctions. Threat is a key variable in understanding tolerance decisions (Duch and Gibson, Reference Duch and Gibson1992; Gibson and Gouws, Reference Gibson and Gouws2001; Huddy et al., Reference Huddy, Feldman, Taber and Lahav2005; Marcus et al., Reference Marcus, Sullivan, Theiss-Morse and Wood1995; Stouffer, Reference Stouffer1963; Sullivan et al., Reference Sullivan, Marcus, Feldman and Piereson1981). It may well be that exclusionary groups are seen as more threatening, and this is the reason for the distinctions documented across groups here. Unfortunately, there is no direct measure of threat available to test this contention in the CYS. This is an empirical question for future research and a potentially fruitful way of understanding what it is about exclusionary groups that make them particularly likely to be censored by the next generation.
In conclusion, young people can and do make distinctions across different types of speech they find offensive. Documenting such distinctions provides a better understanding of the way people balance the sometimes competing demands of individual rights and the inclusion of ethnic, racial and religious minorities in public life. As immigration continues to change the demographic realities in liberal democracies, this article has pointed to one way in which public opinion reflects support for such competing rights. In doing so, it has challenged current conceptualizations of political tolerance as an absolute democratic value and provided substantial evidence of inter-target group distinctions. The concept of multicultural tolerance developed here reflects the ways in which many young citizens set predictable limits on speech which reflect both the legal realities in Canada and Belgium and the increasing acceptance of proactive measures to combat discrimination.
Acknowledgments
The author is grateful to the principal investigators of the Comparative Youth Study, Dietlind Stolle and Marc Hooghe, for their collaboration in collecting the data used in this article. The Social Science and Humanities Research Council (SSHRC), the Fonds Québécois de la Recherche sur la Société et la Culture (FQRSC), the Fonds Wetenschappelijk Onderzoek Vlaanderen and the Bijzonder Onderzoeksfonds, K.U. Leuven provided financial support for the data collection. I would also like to thank to Elisabeth Gidengil, Stuart Soroka and two anonymous CJPS reviewers for useful comments provided on earlier drafts and the American Association of University Women whose financial support made this research possible.
