Are today’s youth more tolerant? Trends in tolerance among young people in Britain

Attitudes towards social groups that have traditionally been marginalised or discriminated against have changed markedly in Britain over the past three decades. This change is particularly marked in attitudes towards homosexuality and racial diversity which, as public opinion surveys have regularly shown, have become more accepting over time. This change is often attributed to older, less tolerant generations being replaced by young cohorts who are more inclusive and open minded in their attitudes to cultural others. The paper explores this argument by examining trends in people’s attitudes towards a variety of minorities, including the said groups, but also immigrants and foreign workers. It starts with a discussion of several perspectives predicting different trends with regards to these attitudes. A distinction is made between optimistic ones (i.e. those anticipating rising levels of tolerance) and pessimistic ones (i.e. those expecting stable or declining levels of tolerance). Subsequently, the paper presents trend analyses and an analysis of age, cohort and period effects to broadly assess the explanatory power of these perspectives. Using these approaches, we find that tolerance towards racial minorities and homosexuality has indeed risen across the board, and that young people are also more accepting of these groups than their parents or grandparents and previous generations of young people. These trends broadly support the optimistic perspectives. However, we also find that prejudice has not disappeared from youth attitudes altogether; for a sizeable minority of youth, it has merely shifted its focus to immigration. Not only have unwelcoming attitudes towards immigrants generally become stronger, young people are not always the most tolerant age group regarding this social group. These findings are thus more in line with the expectations of the pessimistic perspectives.


Introduction
Over the past two decades there has been a marked shift in attitudes towards many social groups that have historically been marginalised and/or vilified in British society. People have become more accepting of homosexuality and interracial marriage, for instance. These changes have coincided with globalisation and openborders policies that have made West European societies more diverse, ethnically and religiously. Britain, in particular, has experienced a sharp increase in the number of immigrants, coupled with a gradual rise in the size of the Britishborn, ethnic minority population (Storm et al., 2017). There has been much debate, discussion and consternation in Britain about these social changes -particularly in relation to cultural diversity, immigration -and their implications. Contemporary youth are often considered to be more tolerant, open minded and inclusive because they grew up under these conditions. Yet, they have also suffered more from the last economic crisis and are likely to face more competition from immigrants than are older people. These last-named conditions are said to make people less tolerant, based on the argument (discussed below) that scarcity of resources fuels hostility towards out-groups.
In this article, therefore, we seek to address three questions: What are young people's attitudes towards cultural others? Is the current generation of young people more tolerant than their elders and/or previous generations of young people? And how can the patterns and trends on these attitudes be explained? To address these questions, we will draw on the data of two surveys (World Values Survey and British Social Attitudes) and present trend and regression analysis of attitudinal change in Britain over three decades. To the best of our knowledge, academic studies looking at trends in attitudes towards various social groups and comparing young people to both older people and to previous generations of young people are non-existent. That trends can differ quite dramatically across various indicators of (in)tolerance, and that there is thus every reason to investigate this in depth, becomes evident when we compare different studies. While Ford (2008) found racial intolerance to have declined in Britain in the 1980s and 1990s, Coenders and Scheepers (1998) found support for discrimination of immigrants to at first decline and then to rise during the same period in the Netherlands.
What our findings will show is that, on the whole, young people are indeed more accepting of homosexuality and of racial diversity than their parents or grandparents and previous generations of young people. However, we shall also see that intolerance has not disappeared from youth attitudes altogether; for a sizeable minority of youth, it has merely shifted its focus to immigration. Contemporary youth are markedly less welcoming of immigrants than young people 10 or 20 years ago. Given that immigration has been the most important political issue for the British electorate since May 2015 (Ipsos Mori, 2016a) and trumped the economy as the key concern motivating the vote in the EU Referendum in June 2016 (Ipsos Mori, 2016b), it is hard to exaggerate the relevance of these findings.
In the section following the next, we will discuss various theoretical perspectives offering different explanations and expectations regarding trends and generational differences in tolerance. We then present two empirical sections with analyses of survey data on the trends and drivers of tolerance. These sections serve to broadly test the predictions of the theoretical perspectives. The conclusion highlights our main finding that the target of intolerance has shifted and explains what the main contribution of this paper is to the existing literature.

A brief note on terminology
In this article, we understand the concept of tolerance in a broad sense; that is, as denoting acceptance of, and favourable and inclusive attitudes towards, various minority groups that are often marginalised and/or discriminated against by the majority. This means we consider its antonyms to be hostility, prejudice and exclusionism (see also Dejaeghere et al., 2012). We are aware that this goes beyond its original meaning, which revolves around the idea of enduring and respecting something one dislikes. As Vogt (1997: 3) put it: 'Tolerance is intentional self-restraint in the face of something one dislikes, objects to, finds threatening, or otherwise has a negative attitude towards'. We understand the term in a much wider sense, i.e. as also referring to attitudes towards cultural others that do not necessarily include this element of dislike, such as positive opinions about cultural others and a willingness to accept them as equals. We further acknowledge that the attitudes we hold it to embrace are not equally demanding in relation to implications for the treatment of cultural others. Believing that others are fundamentally equal to one's own group and therefore deserving of the same rights (i.e. inclusive attitudes), or supporting an unrestrained endorsement of these others obviously require more accommodating policies towards these groups than a 'resigned acceptance for the sake of peace' (Walzer, 1997: 10; see also Dobbernack and Modood, 2013). These finer distinctions are glossed over in the more colloquial use of the term, however. In the media and in everyday conversations, the term is often used as a catch-all phrase referring to all kinds of positive attitudes towards various minorities. We adopt the term in this idiomatic sense because we need an accessible term to denote a wide set of attitudes often seen to be interrelated.
In view of the sensitivity of the topic under investigation, we take extra care in the terms we use to refer to the object of tolerance; that is, the group which people are asked to express their opinions about. Although no terms are completely devoid of negative or positive connotations, we will use concepts that are as neutral as possible. Hence we will use 'social groups' or 'cultural minorities' to refer to immigrants, homosexuals and distinct racial minorities, and not describe them as 'outgroups', since the latter could be read as implying that we as researchers consider these groups as somehow deviant from the norm, which is not our intention. Nonetheless, when certain terms are used in the wording of survey questions (such as 'immigrants' and 'homosexuals'), we will copy this terminology to avoid confusion. In situations where the majority population needs to be referred to, we will use 'majority' or 'White British' rather than the 'native' or 'indigenous' population as use of the latter implies that we would consider the objects of tolerance as somehow less native. In the case of immigrants, this is undoubtedly true, but not in the case of other minorities.

The drivers of tolerance
There is extensive literature on tolerance and prejudice, particularly regarding immigrants (see Hainmueller and Hopkins, 2014), but considerably less debate about where these attitudes come from and why individuals (and societies more generally) become more or less tolerant of cultural others at particular points in time. This varied literature contains theories that are optimistic about the future of tolerance, and some that are decidedly more pessimistic. The optimistic perspectives predict rising levels of tolerance while the pessimistic ones anticipate stable or declining levels of tolerance. These perspectives can further be subdivided into those discerning clear generational differences in tolerance related to different circumstances in people's formative years and those envisaging rising or falling levels of tolerance due to more general processes affecting everyone equally. The former can be said to anticipate cohort effects while the latter would expect to find period effects.

Optimistic perspectives
The optimistic theories tend to focus on the fact that societies as a whole are becoming more tolerant over time and can point to a number of social changes, technological developments and cultural shifts to explain these changes. Developments in biology and genetics, for instance, have exposed racist ideology as fundamentally flawed (Ford, 2008). Politically, racist views have become completely unacceptable due to their association with the Holocaust and Apartheid (Ford, 2008). Racial intolerance should thus have declined as the assumptions on which such attitudes are based have been shown to be false and morally objectionable. At the same time, an international normative discourse on human rights stressing the dignity and equality of all human beings has become ever more influential as evidenced by the treaties and agreements that many states have signed (Osler and Starkey, 2006). This process can be also be expected to have increased tolerance by increasing the normative pressure on people to respect other people's cultures and views. Tolerance may also have risen because of British society having become ever more ethnically diverse. This assumption is based on the idea that the growing visibility of ethnic minorities in all domains of society and the increasing cross-cultural contacts in everyday settings have made the majority (White British) population more familiar with, and more inclusive in its attitudes towards, ethnic minorities (Ford, 2008). This echoes contact theory, the key tenet of which is that informal intercultural contact reduces prejudice and enhances sympathy and identification with the cultural other (Allport, 1954;Pettigrew and Tropp, 2006).
Other 'optimistic' perspectives would argue that rising levels of tolerance are mainly due to a process of generational substitution whereby new, young and more tolerant cohorts replace more intolerant, older ones. The process of growing diversity might partly influence tolerance through this mechanism, as not all age groups experience diversity equally. Young people have more opportunities for cross-cultural contacts than do older people because the immigrant population tends to be young (Ford, 2011(Ford, : 1024. As they also experience higher levels of diversity than previous generations of young people have, they are likely to be more tolerant than both older age groups and earlier generations of young people. The idea of rising levels of tolerance due to generational replacement is mostly associated with the work of Ronald Inglehart on post-materialism. Inglehart and Welzel (2005) argue that the steady rise in living standards in the Western world after the Second World War meant that new generations grew up under ever more prosperous and secure conditions. With their basic needs satisfied, these generations developed so-called post-materialist values in their formative years, including self-fulfilment, freedom of choice and expression and tolerance of cultural others, and have retained these values in the remainder of their lives. Inglehart and Welzel (2005) discern a steady shift from the 1960s towards post-materialist values generated by a process of young, post-materialist generations replacing older, materialist ones. Thus, with tolerance being one of the core dimensions of this values syndrome, we would expect it to have risen over the last 50 years and we would anticipate the younger generations to show the highest levels of tolerance.
A final optimistic perspective predicting rising levels of tolerance due to generational replacement highlights the steady process of educational expansion. OECD data show that the United Kingdom has indeed experienced a significant educational expansion in the last 15 years: The percentage of 25-to 65-year-olds completing tertiary level education has increased from 26% in 2000 to 42% in 2014 (OECD, 2015: 44). This process, which led to new generations being ever better educated than previous ones, is likely to have produced a growth in tolerance because of the strong association at the individual level between educational attainment and tolerance, which is one of the most consistent findings in social science research (e.g. Scheepers et al., 2002;Stouffer, 1955;Sullivan and Transue, 1999). Several reasons have been proposed as to why more educated people are more tolerant. First, education enhances the knowledge and reasoning skills of people, allowing them to refute prejudiced claims and dismiss irrational fears about cultural others (Nunn et al., 1978). Second, education cultivates tolerance directly through socialisation: the longer individuals stay in the education system, the more they are exposed to tolerance as the core value that it promotes and thus the more likely they are to internalise it (Hyman and Wright, 1979;Stubager, 2008). Third, education enhances the competitive position of individuals and diminishes feelings of economic insecurity. Educated people therefore see people from other cultures as less threatening, which makes them more welcoming and inclusive in their attitudes towards outsiders (Lipset, 1981;Stubager, 2008).

Pessimistic theories
Although education tends to have a positive effect at the individual level, the relationship between education and tolerance can be complex and it also features in one of the pessimistic theories of tolerance. To understand this perspective, we must first distinguish between the absolute and the positional effect of education (see Campbell, 2006;Nie et al., 1996). While the first two reasons noted above relate to the absolute effect of education, the last reason, that is the argument that education improves the competitive position of individuals, refers to the positional effect. If the effect is absolute, any increase in education levels should, over time, yield higher aggregate levels of tolerance among the society as a whole. In that case, the effect of educational expansion on tolerance is simply compositional (i.e. the sum of individual attainment levels). Educational expansion can then be expected to lead to a rise in tolerance through the substitution of older, less educated generations by younger, more educated ones (see, e.g. Stubager, 2008). By contrast, if the effect of education on tolerance is mainly positional, educational expansion cannot be expected to have increased overall levels of tolerance much since the competitive position of more educated individuals will not improve when other people also become more educated. Put simply, if levels of education are raised for everyone, then having even more education than previous generations of young people does not increase one's competitive advantage today, and nor does it enhance a sense of economic security, which is a contributory factor to explaining why people with more education are more likely to be tolerant (see, e.g. Borgonovi, 2012).
The notion of competition is also key to the second pessimistic perspective -the perceived threat theory. This theory contends that the scarcer the resources in society, and the greater the competition for them, the more people will tend to view outsiders as threatening and the more negative and hostile their opinions will tend to be towards such groups (Blalock, 1967;Bobo, 1983;Olzak, 1992;Quillian, 1995;Scheepers et al., 2002;Semyonov et al., 2006). If material conditions are indeed a decisive factor, one would expect young people, in particular, to show declining levels of tolerance as this age group has disproportionately been affected by the rising costs of education and housing and the growing precariousness of jobs in Britain. Moreover, at the same time when these conditions were worsening, the country experienced unprecedented levels of immigration, with net inward migration steadily increasing from the mid-1990s onwards to reach an all-time high of 300,000 people in 2014 (ONS, 2016). As the competition over resources, real or perceived, is not just a function of the amount of resources themselves, but also of the number of people seeking access to them, high levels of immigration can equally spark feelings of threat and intolerance among the 'indigenous' majority (Coenders and Scheepers, 1998;Gorodzeisky and Semyonov, 2009).
Some have argued that the majority need not only feel threatened in an economic sense. A sudden influx of immigrants could also spark a sense of cultural threat: a feeling that one's identity, established ways of life, and one's norms and values are being challenged by outsiders (Chandler and Tsai, 2001;Sears and Henry, 2003). Either way, the result is that people retreat into their own community, develop hostile attitudes towards the ethnic minorities living in the same neighbourhood and reject a discourse of racial equality and multiculturalism, as has been well demonstrated by case studies of White British youth in London and the northern industrial towns of Oldham and Rochdale (Hewitt, 2005;Thomas and Sanderson, 2013). These case studies also show how localised and contingent on particular circumstances these feelings can be. Such feelings of threat can, moreover, be activated by political entrepreneurs claiming that immigrants bring nothing but misery to the country (Bohman, 2011;Ford, 2011). According to Hopkins (2010), negative political rhetoric influences how people evaluate a recent increase of immigrants in their home region. As Britain experienced a severe economic crisis after 2008, attracted high numbers of immigrants and saw the rise of a political party critical of immigration and the European Union (the United Kingdom Independence Party), one would, based on these theories, expect tolerance of cultural others and minorities to have decreased among all age groups over the last 15 years. By this logic, this downward trend should be particularly evident among young people; this generation fared worse in the aftermath of the global recession and this, coupled with the increasingly insecure place in the labour market, could fuel anti-immigrant sentiment (Mieriņ a and Korol¸eva, 2015: 187). The young, moreover, face more competition from immigrants in areas such as the job and housing markets than older age groups due to the youthful age profile and residential preferences of immigrants (Camarota, 2013).
The idea of a resurgent nationalism can be seen as another pessimistic perspective. This perspective questions whether the post-national discourse on human rights and cosmopolitan citizenship, as discussed earlier, has truly become more salient. Scholars embracing this view observe that nationalism has made a comeback since the turn of the century, as manifested by the electoral successes of populist anti-immigrant parties in Europe (see Mieriņ a and Korolȩva, 2015;Mudde, 2013;Pilkington and Pollock, 2015), growing discord within the European Union and a growing realism in international relations (Kaplan, 2012;Rachman, 2014). Mainstream parties have adopted some of the rhetoric and proposed policies of these populist parties, leading to more restrictive immigration and naturalisation regimes everywhere and a discourse more generally of protecting and privileging the majority population (Van Spanje, 2010). These recent trends suggest that the taboo on expressing negative sentiments towards cultural others, especially regarding immigrants, has weakened, leading possibly to growing intolerance in all age groups. That said, since the perspective is not making claims about generational differences, young people are not expected to show any higher or lower levels of tolerance than other age groups.

Data and methods
We draw on data of the British Social Attitudes survey (BSA) and the World Values Survey (WVS) to investigate trends in tolerant attitudes and examine some of the drivers of tolerance as suggested by the aforementioned perspectives. While the BSA has relevant data going up to 2013, the WVS allows for an assessment of trends only until 2006 since the United Kingdom did not participate in the latest round of this survey. The WVS is useful nonetheless as it allows us to trace trends further back in time, particularly regarding attitudes on immigrants. Both surveys include nationally representative samples of the British population, with sample sizes varying between 1000 and 3400 respondents across the different waves of the two studies.
First, we present a series of trend graphs showing the responses of different age groups on various tolerance indicators. In each graph a particular age group at a certain point in time is compared to that same age group at an earlier or later point in time. One and the same age group thus represents different cohorts across different points in time. These graphs allow us to compare young people to older age groups at the same point in time and to compare them back in time to their parents when they were young. They further permit a provisional evaluation of the aforementioned perspectives.
We complement the review or trends with a multivariate (OLS regression) analysis of age, period and cohort (APC) effects, based on the pooled samples of Waves 1996Waves , 1999Waves , 2003 and 2013 of the BSA. 1 APC analysis is appropriate as some perspectives suggest that changing attitudes reflect a cohort effect (e.g. postmaterialism and growing diversity) while others posit they represent period effects (e.g. resurgent nationalism). APC analysis has some important limitations, most notably the classic identification problem, which means that, as the three effects are completely linearly dependent, they cannot be distinguished; for instance, knowing how old someone is (age) and the year of the survey (period), means knowing when this person was born (cohort) (Tilley and Evans, 2014). This problem has led some scholars to abandon statistical analysis of APC effects altogether and exclusively use descriptive and graphical investigation of data (e.g. Voas and Chaves, 2016). We believe a statistical approach does have its merits, however. On the one hand, the three effects can still be estimated by using proxies for one or more of the three effects (see, for instance, Coenders and Scheepers, 1998;Ford, 2008). We follow Ford's approach in using conditions relating to life stages, such as being married, being a widower and owning a home, as proxies for age. These conditions have been shown to be related to authoritarianism (Altemeyer, 1995) and political orientations (Binstock and Quadagno, 2001), and are therefore also likely to influence tolerance. Including these proxies in the models helps to prevent multicollinearity between the APC variables (Ford, 2008: 619). On the other hand, by assigning different significance levels to the various effects, the statistical approach allows researchers to arrive at a better assessment of the relative importance of the different factors than a trend analysis. Indeed, a multivariate APC analysis allows us to draw more precise inferences regarding the explanatory power of some of the aforementioned perspectives (see further later).
Cohort is measured with year of birth rather than five or 10-year periods (such as birth decades) because of the arbitrariness in establishing the boundaries for the latter. Period is represented by BSA Round. Although we have data on just two or three rounds for the outcomes to be analysed, these rounds are quite far apart (e.g. 1996 and 2013), meaning that they capture very different circumstances and zeitgeists. In 1996, Western countries experienced robust economic growth, liberal democracy reigned supreme and globalisation was seen as a welcome and/or inevitable process; in 2013 Western economies lay in tatters, anti-immigrant parties with protectionist agendas were quickly gaining popularity and authoritarianism was on the rise globally. The BSA rounds thus truly reflect very different periods. It is important to note, though, that period represents a multitude of macro-level influences, including educational expansion, net immigration, the economic crisis, the alleged growing salience of a human rights discourse and the alleged comeback of nationalism. We cannot include variables representing these influences in the model as there are not enough data points (points in time) to assess the effects of these variables individually. Therefore, if we find a period effect we cannot be sure what this effect precisely stands for. It could represent the influence of any one of these variables or a combination of them.
We further added highest qualification attained (HQA) to the models (see Table 1 for the categories of this variable). The outcomes are racial tolerance, support for equal treatment of immigrants and perceptions of ethnic competition (see Figures 3, 6 and 7 for the wording of the items on which these outcomes are based). The analyses on the latter two outcomes are based on a sample of those identifying as British; the analysis of the first-named outcome only includes those identifying as White British. We could not run the analyses on the same group of respondents as ethno-racial identity was not available in all the BSA waves for the latter two outcomes, while the item on racial intolerance was asked of only White British respondents. 2 The APC analysis enables us to look more closely at the optimistic perspectives in particular. If HQA is the only variable with a significant (and positive) effect on tolerance, rising levels of tolerance are exclusively due to increasing levels of education, which would strongly support the education as absolute effect argument. However, if birth cohort is the only variable with a significant (and positive) effect, it is not their higher education levels but something else that makes young people more tolerant, indicating that the post-materialism and growing diversity perspectives are better able to account for the trends in tolerance.
It is more difficult to evaluate the different pessimistic perspectives because it is not quite clear which influence is captured by period. Nonetheless, some more precise inferences can be made. If the effects of period and cohort are both negative in a model with tolerance on immigrants as outcome, tolerance towards this group has declined and is lowest amongst the young, which would indicate that the perceived threat theory has excellent predictive power. If only the effect of period is negative in the said model and all the other effects are insignificant, then the results are more in accordance with the expectations of the resurgent nationalism argument or the education as positional effect thesis.
Trends over time: Are young people today more tolerant than previous generations of youth?
Have people become more tolerant of minorities in general in their immediate environment or does this only apply to some minorities? Figures 1 to 4 show that the latter is the case. Over time, Britons have become steadily more accepting of homosexuals and people of a different racial background, and this trend can be seen in all age groups. In other words, the current generation of young people are more tolerant than the young people of the 1980s and 1990, and the same applies for middle-aged people and pensioners. This development has, in fact, been so pronounced that, by the end of the time series of the WVS (2006), only small minorities in each age group still have issues with living next door to homosexuals or people of a different race (see Figures 1 and 2). Data from the BSA going up to 2013 confirm this trend for racial tolerance. Ever smaller numbers of people state that they would have problems with one of their close relatives marrying a person of Black or West Indian origin ( Figure 3; the question was asked only of White British respondents). The fact that the lowest levels of intolerance can be observed at the end of the time series when Britain was in the midst of a severe recession  suggests that the economic crisis has not affected racial tolerance in the slightest way. We further see that the youngest age group tends to be the most tolerant and the oldest group the least tolerant whatever the period or the social group we are examining (only regarding the BSA question on racial tolerance do we see that the 35-to 54-year-olds are marginally more tolerant than the youngest age group towards the end of time series). Together, these results provide some provisional support for the optimistic perspectives, i.e. the international human rights discourse, the growing diversity perspective, post-materialism theory and the absolute effect of education hypothesis.

Immigrants -the new 'other'?
By contrast, when we look at how attitudes towards immigrants have changed over time, a different pattern emerges. Between 1981 and 1998, opposition to having   immigrants or foreign workers as neighbours declined among the young and the middle aged. And at each of the three time points we looked at, young people (age 15-29) expressed the least opposition to this. These results are in keeping with the trends in tolerance towards the other groups discussed above. Yet this trend is reversed in the 2000s, and indeed in 2006, intolerance of immigrant neighbours had not only started to rise again among the under 50s, but opposition was higher than it was in the early 1980s (see Figure 4). The deviating pattern regarding trends on immigrants is even more pronounced when looking at recent trends in exclusionary attitudes and perceptions of competition. We see, for instance, a very pronounced rise among the young and an almost equally salient increase among middle-aged people in the numbers supporting the idea that employers should give priority to the majority population over immigrants in times of crisis (see Figure 5); by contrast, there has been virtually no change in attitudes among the over 50s. What is more, the absolute levels are also noteworthy: in each age group, clear majorities supported this form of unequal treatment by the mid-2000s.
Declining support for equal treatment can also be seen when turning to the more recent data from the BSA survey. Asked about whether 'legal immigrants to Britain who are not citizens should have the same rights as British citizens', all age groups expressed much lower levels of agreement with this idea in 2013 than 10 years earlier ( Figure 6). Falling support for equal treatment is accompanied by rising perceptions of competition as all age groups show increases in the percentages agreeing with the statement that 'immigrants take jobs away from people who were born in Britain' (Figure 7). Again, the overall levels are remarkably high: almost 50% of the 18-to 34-year-olds and a majority of the 55-plus group support the statement. These rising levels of unwelcoming attitudes are all the more remarkable in view of the growing share of ethnic minorities in the British population (Storm et al., 2017) and the tendency of ethnic minority respondents, both of the first and second generation, to be significantly more inclusive in their attitudes towards immigrants than the majority population (Janmaat, 2014: 815). Perhaps their own experience as immigrant or that of their parents makes these groups identify with immigrants and express greater sympathy towards them (Janmaat, 2014). Nonetheless, with the passing of generations, links with immigrant pasts may weaken. It has been noted, for instance, that Britain's ethnic minorities also favour a reduction in migration rates and start to think less positively about immigrants, although the White British are still more critical on immigration (Amrani, 2016;Blinder, 2011). Equally interesting is that the youngest age group becomes virtually indistinguishable from the middle-aged groups in their tolerance levels. On some indicators they even become slightly less tolerant than the latter. Towards the end of   the time series, middle-aged people (age 35-55) are, for instance, less supportive of employers discriminating against immigrants than young people, and they do not appear to perceive as much competition from immigrants on the labour market. Young people are still the most supportive of extending equal rights to immigrants.
In sum, the trend analysis shows that young people's attitudes towards minorities are more complex than it is sometimes portrayed. When the focus is on the acceptance of racial and sexual minorities, we see that young people are more tolerant that their elders or than previous generations of young people. Yet this trend towards tolerance does not necessarily extend to other minorities. When we examined attitudes towards immigrants, including support for equal treatment and perceptions of competition, we found declining or stabilising levels of tolerance and in some cases, slightly lower levels of tolerance among young people than middleaged respondents.
These contrasting trends are all the more remarkable as previous research has found attitudes towards different racial and cultural groups to be so strongly interrelated that they have been interpreted as manifestations of a latent dimension of tolerance (e.g. Brewer, 1986;Dejaeghere et al., 2012;Scheepers et al., 2002). It is perhaps the determination to find such a singular dimension that has prevented scholars from looking at the variation in the attitudes towards different groups or from assessing whether these attitudes cluster equally strongly among different social groups or generations. Indeed, using Wave 5 of WVS and correlating the items of Figures 1, 2, 4 and 5 to one another, we find a marked difference between the youngest (those born after 1975) and the oldest group (those born before 1954) in the clustering of these attitudes. 3. Among the young, attitudes towards other racial groups and homosexuals are not significantly related to attitudes on immigrants and support for discrimination of immigrants; however, among the old respondents, all four attitudes are closely interlinked. In other words, while a more general disposition of tolerance may well inform attitudes on immigrants among the old, the young seem to decouple immigration issues from the norm of accepting and embracing cultural minorities. These diverging attitudes among the young towards different minorities in a way make sense in view of the greater competition young people face from immigrants and their generally more vulnerable position on the labour market, as discussed earlier. They also shed further light on our finding that contrasting trends are particularly evident among the young (compare, for instance, Figures 1 and 2 with Figure 5).
The trends also raise questions about the optimistic theories discussed above. The declining levels of tolerance regarding immigrants in the youngest age group are particularly noteworthy in view of educational expansion, which means that young people today are more highly educated than young people at any point in the past. This might indicate a positional effect of education. In the next section, we will examine the explanatory power of the various theoretical perspectives more closely.
A closer look at the perspectives explaining tolerance: APC effects Table 1 presents the results of the APC analyses. A striking finding is that educational attainment exerts a strong positive influence on all three outcomes. That is *P < 0.05; **P < 0.01; ***P < 0.001. Note: None of the three analyses exceeded critical multicollinearity levels, as gauged by the variance inflation factor (VIF). Source: British Social Attitudes Survey (Waves 1996(Waves , 1999(Waves , 2003(Waves , 2013. the only commonality across the three analyses, however. The effects of year of birth (cohort) and BSA Round (period) vary radically across outcomes.
If we look at racial intolerance, we see a strong negative cohort effect indicating that young generations are significantly more racially tolerant than older ones. It is worth mentioning that this effect occurs in addition to a strong negative effect of educational attainment. This means that the generational effect is not only due to younger cohorts being more highly educated than older ones. Evidently other influences must be driving this effect as well, such as, for instance, more prosperous conditions during one's formative years, or more contact with cultural others. We further note that the effect of year of birth is significant while controlling for marital status and tenure, which indicates that it does represent a cohort effect and not, or not just, an age effect. (We have to add the words 'or not just' as we can never be sure whether marital status and tenure fully capture the effect of aging. Only a longitudinal study with several panels would permit a better disentanglement of age and cohort effects.) Of further interest is that the period effect is non-significant. It is difficult to say what this precisely means. It could be that macro-level processes, such as rising rates of immigration, the post-2008 economic crisis, growing nationalist rhetoric or an increasingly influential post-national discourse of human rights, are not influential. It could also mean that the first three processes enhance intolerance, but that these influences are cancelled out by an increasingly prevalent human rights discourse that promotes tolerance.
Together, these findings are quite consistent with the optimistic perspectives, such as those of post-materialism and growing diversity. As the period effect is non-existent, they also suggest that the effect of education on racial intolerance is absolute rather than positional (if the effect of education had been positional we should have seen a positive effect of period). The strong cohort effect further suggests that racial tolerance is likely to increase in the near future as more tolerant young generations replace less tolerant older ones.
In contrast, the effect of cohort is insignificant and that of period highly significant when we look at the two other outcomes of interest: namely support for equal treatment and perceptions of ethnic discrimination. Regarding this period effect, the more recent the BSA round, the less supportive people are of offering equal rights to immigrants and the more they feel that immigrants take jobs away from the majority population, controlling for all other conditions. This indicates growing exclusionism and intolerance towards immigrants, which is the opposite of what the human rights perspective would predict. Educational attainment has a positive effect on support for equal treatment and a negative one on perceptions of ethnic discrimination. In combination with the non-effect of cohort, this suggests that higher tolerance levels amongst the young are entirely due to younger generations being more highly educated than older ones; once you control for educational attainment, differences between the generations disappear. 4. This means the post-materialism and the growing diversity perspectives cannot account for the patterns on these outcomes. Instead, the results are much more consistent with the 'pessimistic' perspectives. Support for one of these perspectives, the perceived threat theory, is mixed, however: on the one hand, the theory correctly predicts growing exclusionary attitudes towards immigrants, but on the other hand, and contrary to expectation, young people are not more hostile to immigrants than older generations despite having taken the brunt of the economic crisis and facing more competition from immigrants. Lastly, as the period effect is as strong, or stronger, than the effect of educational attainment and as exclusionary and intolerant attitudes on immigrants have increased (as shown by the trend analyses), there is more evidence for a positional than an absolute effect of education. We could not determine, however, whether the resurgent nationalism idea or the positional effects model has most power in accounting for the attitudes on immigrants because the period effect can represent a number of macro-level influences simultaneously.
More generally, the analyses of APC effects confirm the observations regarding trends. Reported acceptance of racial diversity differs markedly across generations and shows a steady upward trend undisturbed by fluctuations in the economy or immigration. However, the story is completely different for tolerance outcomes relating to immigrants. Processes such as rising immigration or economic contraction are likely to be highly influential (although it could not be determined which one of these precisely had a decisive influence) and differences between the generations are small, reflecting only differences in aggregate educational attainment. In short, patterns on racial tolerance cannot be generalised to attitudes on immigrants. The British, and particularly young people, thus appear to distinguish between different minorities. It seems that they increasingly accept people of a different skin colour as one of 'us' (provided they are British), while immigrants are ever more feared and considered as 'them'.

Conclusions
In this paper we set out to examine the oft-cited claim that young people are more tolerant than their elders and/or previous generations. We found that, on the whole, contemporary youth in Britain are more tolerant of racial diversity and of homosexuality than older age groups and previous generations of young people. Trend analysis clearly shows that intolerance towards these groups has declined markedly since the 1980s, and indeed, some measures of racial diversity suggest that among young people opposition to these types of racial diversity has almost disappeared. Yet we found that quite a different pattern emerged once we examined youth attitudes towards immigrants and foreign workers. Over the past four decades, attitudes towards these groups have become more rather than less negative; this pattern is true across all age groups, and young people are not necessarily the most tolerant age group. Young people today are less accepting of immigrants than previous generations of young people. This is remarkable because they are more educated than their past peers and because well-educated people are usually more tolerant in their views on immigrants than poorly educated ones.
Through further analysis we also examined whether these attitudinal changes could be attributed to the replacement of generations (i.e. a cohort effect) or to conditions prevailing in a certain period (i.e. a period effect). We found that the increase in racial tolerance and acceptance of homosexuality primarily represents a cohort effect, induced not just by rising levels of education but also by a broader cultural shift in attitudes. By contrast, shifts in attitudes towards immigrants appear to be more a function of the prevalent social conditions, confirming pessimistic scenarios about intolerance rising in times of crisis and enhanced competition over scarce resources or becoming more salient as part of a virulent resurgence of nationalism. For many young people, the prevailing conditions have meant fewer opportunities in housing and employment, and a media environment that tends to link these challenges to a surge in immigration. These contemporary contextual issues may well explain why young people are less accommodating in their attitudes towards immigrants than towards other cultural minorities.
This paper contributes to the wider literature by showing that, even though attitudes towards a variety of cultural minorities may be strongly interlinked, they need not change in the same way over time, nor be influenced by the same conditions. Putting the spotlight on contrasting trends in such attitudes is a focus that few studies share. Most of the existing studies zoom in on one outcome among a series of outcomes capturing tolerance and tend to be quite unequivocal in their conclusions as to the theoretical perspectives that can best explain the variation in this outcome (e.g. Coenders and Scheepers, 1998;Dejaeghere et al., 2012;Ford, 2008;Janmaat, 2014;Semyonov et al., 2006). Even some review studies restrict themselves to a single outcome. Thus, Hainmueller and Hopkins (2014) focus on attitudes towards immigrants and observe that concerns about immigrants constituting a cultural threat to the community are a stronger factor in explaining such attitudes than personal economic circumstances. Our study shows that there is no one factor or theoretical perspective that can best account for the variation in attitudes towards various minorities; which factor has most explanatory power depends on the minority that people are asked to express their opinions about. This finding, in particular, casts doubt on those studies that collapse a range of tolerance outcomes into one outcome and proceed by investigating the determinants of this outcome (e.g. Dejaeghere et al., 2012).
More generally, our findings guard against the assumption that dissolving boundaries regarding one group can easily be generalised to other groups. As we have seen, growing acceptance of people of different racial backgrounds and different sexual preferences coincide with a hardening of the split between the majority population and immigrants. We are certainly not the first to note such a process of shifting boundary making. Back (1993), for instance, found a similar dynamic amongst a group of White and Black youth in South London; the engagement of this group with Afro-American and Afro-Caribbean popular culture engendered a de-racialised 'neighbourhood nationalism' (p. 220), which, however, became highly exclusionary and hostile towards Asian minorities and Vietnamese in particular. The findings of Back and our own study thus show that the development of seemingly more inclusive identities only leads to new forms of exclusion and othering. It is all the more sobering to realise that such new patterns of exclusion continue to happen despite new generations being better educated than previous ones. This tempers the optimistic view that rising levels of education should result in greater broadmindedness towards cultural others more generally.

Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the ESRC (grant ref: ES/J019135/1).

Notes
1. Unfortunately, due to shortcomings in the data regarding the measurement of key variables we could not perform APCE analysis on the WVS data. 2. The BSA also asked White British respondents to express their attitudes towards Asians, but these questions were not asked of other ethnic groups. It is somewhat surprising that only this group was questioned about their racial tolerance of other groups. The survey organisers may have assumed that issues of racial intolerance only, or mainly, apply to the White British population. 3. The results of these analyses can be obtained from the authors upon request. 4. Indeed, once educational attainment is omitted from the model, the effect of cohort becomes significant for both outcomes (results can be obtained from the authors upon request).