Too many immigrants: How
does local diversity contribute to
attitudes toward immigration?
Juta Kawalerowicz
The Institute for Analytical Sociology, Linko¨ping University, Sweden Department of Human Geography, Stockholm University, Sweden
Abstract
This paper investigates the link between residential context, perceptions and attitudes toward
immigrants by linking data from the British Election Study with Census statistics on
compo-sition of electoral constituencies in 2001 and 2011. I consider which type of local diversity is
most salient for natives’ attitudes by combining information on ethnicity, religion and skills.
Second, I look at whether base levels of and changes in local diversity affect anti-immigration
attitudes in the same way. I find that immigration is more salient when defined by ethnic
criteria, rather than criteria that combine ethnicity and religion or skills. Anti-immigrant
atti-tudes are more likely to be expressed by natives who live in constituencies where there has
been a large change in diversity between 2001 and 2011, but these responses depend on initial
diversity levels. For non-whites and skilled ethnic minorities higher residential segregation is
associated with more negative attitudes toward immigrants among natives.
Keywords
Immigration, public attitudes, intergroup conflict, realistic group threat, defended
neighbourhood hypothesis, United Kingdom
Introduction
Over the last couple of decades, immigration has reshaped Western European societies. With increasing diversity, more people are now living in areas with larger numbers of ethnic and religious minorities. British society is a good example of the magnitudes of such changes. Fifty years ago, the British population consisted of just over 1% non-white population (Rose, 1969); by 2011 this figure had increased to 13%. With the increase in diversity, the reaction of natives to the new minority groups became an increasingly urgent matter. Since the early 2000s, the British public has consistently identified immigration as one of the top issues facing the country ahead of education, the National Health Service and the economy (Blinder and Allen, 2016) while anti-immigration sentiments have been linked to the rise of the UK Independence Party and the outcome of the
Corresponding Author:
Juta Kawalerowicz, The Institute for Analytical Sociology, Linko¨ping University, Norrko¨ping, Sweden. Email: juta.kawalerowicz@liu.se
Acta Sociologica 2021, Vol. 64(2) 144–165 ªThe Author(s) 2021 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/0001699320977592 journals.sagepub.com/home/asj
2016 referendum on European Union membership (Evans and Mellon, 2019; Hobolt, 2016). How does diversity at a local level contribute to attitudes toward immigration? Two predominant sociological theories – contact and ethnic threat – give diverging answers. On one hand, exposure to diversity is a primary condition for inter-group contact, which reduces prejudice. On the other hand, increasing diversity can also lead to increased competition and a feeling of threat in either economic or symbolic domains. Recent studies examining contextual effects of diversity on attitudes toward immigrants show that increasing diversity is associated with more prejudice, but levels are not (Kaufmann, 2017; Newman, 2013; Van Heerden and Ruedin, 2019).
In this study, I examine data from the British Election Study in England and Wales and look at the role local diversity plays in shaping natives’ views on immigration. Individual-level information on attitudes, demographics and self-reported economic prospects is linked to contextual information from 573 constitu-encies in England and Wales.1I contribute to the existing literature in three ways. First, I show how different types of local diversity relate to anti-immigration attitudes in the British context. My data allow for directly looking at attitudinal questions rather than a preference for immigration reduction as used by Kaufmann (2017). Second to study how attitudes toward immigrants are related to natives’ perceptions of the local population, I test whether levels and changes apply in equal measures to local composition with regard to ethnicity, religion and skills. I test this by measuring the composition of ethnic minorities (i.e. those who did not identify as white British on the Census) in terms of the proportion of those who are non-white, Muslim and unskilled. My analysis suggests that ethnicity remains the most salient dimension. Next, I provide evidence for different effects of levels and changes while controlling for residential segregation. Researchers have looked at levels, changes or segregation, but to my knowledge, no other study has simultaneously considered all three indicators. Third, I observe an interaction effect between levels and changes in the proportion of non-white residents in the opposite direction from what has been previously reported by Newman (2013). While residents in areas with a high initial proportion of non-whites are generally more positive toward immigrants, further increases are associated with steeper declines in pro-immigration attitudes.
Theories and hypotheses
The two dominant theoretical frameworks linking attitudes with local context are the power threat hypothesis and contact theory. The power threat hypothesis conceptualizes group hostility between dominant and subordinate groups. The main idea is that with sufficient numbers of newcomers, the hegemonic position of the dominant group is undermined, thus generating competition over resources such as employment, access to social housing or political influence (Blalock, 1967). Yet, under certain conditions, diversity can also be associated with reduction in prejudice. Research published in the wake of the Second World War showed a major decrease in racial prejudice among white military personnel if they served together with African American soldiers. Allport (1954) is credited for drawing attention to the role of interaction in attitude formation. Importantly, the presence of minority groups is not a sufficient condition to facilitate a reduction in negative attitudes. The type of contact that is meaningful in terms of prejudice reduction is characterized by small distance in terms of social status between groups, an environment that encourages cooperation and a broader framework of support from authorities, law and customs. In contrast, superficial contact refers to conditions when groups are heavily segregated or when one group profits from the hardship of another; both are examples of environments less conducive to prejudice reduction (Pettigrew, 1998).
What is the evidence in support of these hypotheses? Below I discuss the findings with respect to geographical scales at which threat and contact operate, conceptualization of immigrant populations, residential segregation and the role of diversity levels and changes.
The role of scale
First, it seems that threat operates at a larger scale and contact at smaller scales that are conducive to the formation of social ties. People are more likely to meaningfully interact with others living in the same
area but have little contact with those living on the other side of the country. Consequently, studies based on larger scales capture effects akin to threat and those based on smaller geographical scales often find support for contact theory (Van Gent et al., 2014; Rink et al., 2009).2
Earlier studies often looked at national scales and provided support for the notion that the size of the immigrant population was positively associated with perceived threat, exclusionary preferences and anti-foreigner attitudes (Quillian, 1995; Scheepers et al., 2002; Schneider, 2008; Semyonov et al., 2006, 2008).3 In later years, researchers examined whether this finding holds for smaller geographical scales. Studies that look at the subnational context prove to be less conclusive; a possible mechanism is that at smaller scales, the effect of contact overshadows threat and much depends on the size of the subnational units. Research-ers sought to distinguish between smaller and larger scales to show that threat and contact operate together but at different scales (Ha, 2010; Oliver and Wong, 2003). For example, Ha (2010: 35) investigated how attitudes are linked to context in both small neighbourhoods and larger metropolitan areas and reported that ‘attitudes toward immigrants vary across geographic areas [ . . . ] anti-immigrant sentiments tend to be greater in the metropolitan areas, and pro-immigrant sentiments are relatively larger in the neighbour-hoods’. Yet, studies of subnational contexts showed mixed results and reported that the concentration of immigrants may have a negative effect on attitudes toward immigrants (Quillian, 1996; Schlueter and Scheepers, 2010), have no or positive effect (Dustmann and Preston, 2001; Evans and Need, 2002; Fox, 2004; Hjerm, 2009; Semyonov, et al., 2004; Wagner et al., 2003), different effects depending on which immigrant groups are considered (Dixon and Rosenbaum, 2004; Ha, 2010; Taylor, 1998), or have effects which depend on regional labour market conditions (Markaki and Longhi, 2013).
Conceptualization of immigrant populations
In a meta-analysis of studies on contextual effects on attitudes toward immigrants, Pottie-Sherman and Wilkes (2017) argue that contextual effects of diversity on attitudes toward immigrants depend on the criteria taken to define the immigrant population as well as establishing which groups respondents have in mind when answering surveys (Pottie-Sherman and Wilkes, 2017). When studying how context affects attitudes toward immigrants, researchers have used a number of ways to define immigrant populations. These measures might have been detached from respondents’ own perceptions of who counts as an immigrant. Blinder emphasizes this issue, observing that the term ‘immigrant’ is inherently vague and one cannot be sure of respondents’ views on immigration because it is not known whom respondents regard as immigrants (Blinder 2015). Even if respondents are provided with clarification, they might hold different views about different groups – for instance, they might be more welcoming toward highly skilled foreign workers but hostile toward refugees (Card et al., 2005; Gorodzeisky, 2011).
Ethnic criteria. When answering questions about immigration, respondents often associate immigration with ethnicity. The condition for threat reaction is that the minority group is visible. Hjerm (2009) finds that in Swedish municipalities, a higher portion of immigrants from outside Western Europe, USA, Canada, Aus-tralia and New Zealand is associated with stronger anti-immigrant attitudes and concluded that with higher visibility comes higher feeling of threat. This is similar to the concept of ethnic hierarchies used by Ford (2011) who showed that British respondents are more hostile toward immigrants from Southeast Asia and the West Indies (i.e. non-whites) and more positive toward European and Australian immigrants. The author argues that ethnic hierarchies are a key dimension distinguishing ‘acceptable’ and ‘unacceptable’ immigrants.
Skills as criteria. Alternatively, researchers have argued that anti-immigrant attitudes are driven by concerns over economic well-being or cultural homogeneity. Two economic theories have been particularly prominent: labour market competition, where influx of immigrants threatens natives’ wages and employment prospects, and fiscal burden, where more negative attitudes toward low-skilled immigrants are explained by the expected burden they put on public services (Hainmueller and Hopkins, 2014). Researchers have observed a preference against low-skilled immigrants; for instance, Hainmueller and Hiscox (2010) used an experimental survey
design to show that American opposition to the arrival of more immigrants depends on whether the description of immigrants portrayed them as highly skilled or skilled. Interestingly, both highly skilled and low-skilled respondents had a strong preference for highly low-skilled immigrants. These results suggest that attitudes toward immigrants are not necessarily rooted in economic self-interest but beliefs about the impact of immigration on a wider society. Markaki (2014) notes a similar pattern in her study of regional determinants of public support for restriction to immigration in the United Kingdom. Her findings suggest that support for limiting immigration is associated more with who the immigrants are rather than characteristics and condi-tions of natives in the region. Additionally, it is reasonable to expect that this distinction between high- and low-skilled immigrants may be of particular importance in Britain in recent years. While Britain has policies aimed at limiting low-skilled migration from outside the EU there were no similar restrictions to the inflow of low-skilled EU migrants, which led to a large increase in the low-skilled immigrant population after 2004. The vast majority of this increase can be attributed to Eastern European migration from new EU countries.
Cultural and symbolic criteria. Another line of research highlights the primacy of cultural or symbolic concerns. In a meta-analysis of 55 studies of contextual effects on attitudes toward immigrants carried out by Pottie-Sherman and Wilkes (2017), the authors found that studies that use symbolic threats tend to find larger effect sizes than those reported for economic or political factors. McLaren (2003) argues that in Western Europe, the preference to expel immigrants is driven to a lesser extent by self-interest or economic concerns and more by the perception of a cultural and religious threat. Religious minorities, particularly Muslims, are mentioned as a potential source of symbolic threat (McLaren and Johnson, 2007). Survey research has confirmed that respon-dents are generally hostile toward religious diversity. For instance, back in 2003, 47% thought that British Muslims could never be truly committed to Britain, and a further 62% declared that British Muslims were more loyal to other Muslims around the world than they were to other people in Britain (McLaren, 2003). Similarly, Card and colleagues (2012) examined attitudes toward immigrants in 21 European countries and showed that cultural concerns (including a preference for religious homogeneity) were two to five times more important in shaping attitudes toward immigrants than economic concerns about wages and taxes.
Following from this discussion on different criteria used to define salient immigrant populations, I expect that:
Hypothesis 1a: Higher base levels of non-white ethnic minority residents are associated with more negative attitudes toward immigrants.
Hypothesis 1b: Higher base levels of low-skilled ethnic minority residents are associated with more negative attitudes toward immigrants.
Hypothesis 1c: Higher base levels of Muslim ethnic minority residents are associated with more negative attitudes toward immigrants.
Residential segregation
Next to scale, another important geographical factor is residential segregation. Areas that are similar in terms of diversity may present different environments for intergroup interaction depending on how segregated these groups are from one another (Pettigrew, 1998). Having immigrant friends, acquaintances or colleagues diminishes threat responses (Dixon and Rosenbaum, 2004; Ellison et al., 2011; Schneider, 2008) while segregation inhibits opportunities for creation of social ties and affects inter-group attitudes. Rocha and Espino (2009) found that among Anglo-Americans, a preference to make English the official language depends not only on changes in the fraction of Spanish-speaking Latinos in the area but also on the level of segregation. Recently, Laurence and colleagues (2019) showed that community segregation moderates the effect of diversity on prejudice, affecting inter-group contact as well as perceived threat. The authors found that in the UK, communities that are becoming more diverse while also becoming more segregated at the
same time experience an increase in prejudice. In contrast, communities that become more integrated see prejudice at stable or even declining levels.
Hypothesis 2: Segregated areas have higher levels of negative attitudes towards immigrants, even when controlling for immigrant population shares.
Levels or changes?
When examining how the presence of immigrants relates to attitudes, researchers often use a static framework to understand a dynamic process. Although immigrants’ arrival might attract media attention and mobilize people to express opposition, over time residents with ethnic minority backgrounds are less likely to be seen as belonging to the immigrant group. While it is plausible that natives react adversely to changes, once they have time to get used to the new environment, they might re-evaluate their initial assessment. Advancements in studying the effect of contextual factors have come from studies that, rather than focusing exclusively on levels, looked at the impact of changes (Hopkins, 2010) and the interplay between base levels and changes (Green et al. 1998; Newman, 2013). Importantly, while some studies conceptualize change as merely a replacement of levels, others have shown that considering levels and change jointly can provide a fuller understanding of the relation between diversity and attitude formation. One such perspective, rooted in a classical interpretation of the power threat hypothesis, holds that members of the majority group exhibit more hostile views toward minority groups when immigration challenges the interests of the majority group. This focus on the balance of power is a prominent theme in work on ethnic competition. For instance, Olzak (1992) argues that conflict is more likely to erupt when the ethnic and racial hegemony of the dominant group is challenged and the racial order begins to break down. Consequently, there is a higher rate of collective action against ethnic minorities, not only because of the higher levels of the minority groups but also as a reaction to the rapid changes that further increase hostility. Although some researchers predict that hostility toward immigrants would be higher in areas with substantial immigrant populations, other studies suggest that anti-immigration attitudes are more common when immigrants constitute a smaller share of the population. The latter view is in line with prejudice reduction mechanisms; when there are no opportunities for contact, natives’ ideas about minority groups are based on stereotypes, which are often negative.
How can population levels interact with change in shaping attitudes? An alternative to ethnic competition as presented by Olzak is found in the ‘defended neighbourhood hypothesis’ introduced in a paper by Green et al. (1998). The authors predict that hostility toward members of the out-group would be strongest in homogeneous areas experiencing rapid growth of the out-group population. The emphasis on demographic changes suggests that the feeling of threat is not simply a function of the minority population but also depends on prior expectations about the local environment. Green and colleagues found support for the defended neighbourhood hypothesis when studying incidences of hate crimes against ethnic minorities in New York City. Such crimes were more common in neighbourhoods previously described as white strongholds, which experienced an influx of non-white residents in recent years, and less common if these minority groups had resided in the area in significant numbers for a long time. Shifting from action to attitudes, Newman (2013) provided evidence that white Americans’ responses to local demographic changes also depend on the prior demographic balance in the area. In counties where there were few Hispanics, an influx of Hispanics increased threat response and fostered anti-immigration attitudes. Similar to predictions given by the defended neighbourhood hypothesis, Newman says that rapid growth of the immigrant population seems to be threatening to those living in areas with small pre-existing immigrant populations and less threatening to those residing in areas where immigrants are more established. The author draws attention to the process of acculturation – natives react adversely to growth in the immigrant population at first; however, the same rate of growth is associated with less negative reactions in areas where residents have had a chance to interact with immigrants over a longer period.
The ethnic competition and defended neighbourhood perspectives agree on the effect of growth but disagree on the impact of population levels. From the ethnic competition perspective, higher levels of the minority group population are a source of competition and threat, whereas the defended neighbourhood hypothesis focuses on the acculturation process that is a consequence of contact and higher levels of diversity. Moreover, these theories make a similar prediction regarding the impact of increasing diver-sity (i.e. change is associated with higher levels of hostility) but differ with regard to the mechanism; the ethnic competition perspective treats growth of the immigrant population as a challenge to the hegemony of the dominant majority group, whereas the defended neighbourhood perspective treats it as a catalyst for action among those who seek to preserve homogeneity.
Hypothesis 3a: Higher base levels of immigrant populations are associated with more positive attitudes toward immigrants.
Hypothesis 3b: Growth of immigrant populations is associated with less positive attitudes toward immigrants.
Hypothesis 3c: Growth of immigrant populations is associated with less positive views about immigrants in areas with low base levels of immigrant populations (interaction effect).
Data and methods
This analysis distinguishes two levels of analysis: the individual and the contextual measured in the Westminster constituency. Individual data come from the British Election Study (BES). The focus of the BES is political participation, but respondents also answer questions about their views on immigration. Here, I use Wave 4 from the latest study, which was conducted in March 2015 on a sample of over 30,000 respondents in England, Scotland and Wales.
In their review of existing research on attitudes toward immigrants, Hainmueller and Hopkins (2014) remarked that attitudes on immigration are about ‘groups of people and about challenges to group bound-aries’ with prior research emphasizing two types of groups – those based on nationality and ethnicity. Because I want to examine the perceptions and attitudes of those whose non-immigration status is rarely questioned, I focus on a subsample of white British respondents. What makes BES data suitable for my purpose is that I can match residents to their constituencies. Although these units are larger than neighbour-hoods, in a way they are the natural unit of analysis for competition over political influence. Ideally, I would prefer to use a finer scale; however, this is the smallest level of aggregation available in a survey that asks questions about respondents’ views on immigration. Still, it should be emphasized that this level is smaller than the regional distinctions found in other subnational studies.
My dependent variable measures attitudes toward immigrants. To construct this variable, I use the following three questions: (1) ‘Do you think immigration is good or bad for Britain’s economy?’ (2) ‘Do you think that immigration undermines or enriches Britain’s cultural life?’ and (3) a rating of the statement, ‘Immigrants are a burden on the welfare state’. The first two questions are measured on a scale from 1 to 7. For economic impact (1), the scale ranges from ‘Bad for economy’ to ‘Good for economy’. For cultural impact (2), the scale ranges from ‘Undermines cultural life’ to ‘Enriches cultural life’. The third variable is measured on a scale from 0 – ‘Strongly agree’ to 5 – ‘Strongly disagree’. The variables are highly correlated (Cronbach’s alpha¼ .89), and I follow a common practice of using principal factor analysis to construct an index of attitudes (Hjerm, 2007; Semyonov et al., 2006; Sides and Citrin, 2007). Higher values indicate more positive attitudes toward immigrants. The peak at the minimum value is for people who have the most negative assessment of immigrants on all three measures (13% of those who replied to all three questions). Other variables at the individual level are sex, age and lack of educational qualifications. I also used two variables that measure subjective economic prospects: self-assessed risk of unemployment and risk of poverty. Both are measures on a scale from 1 (Very unlikely) to 5 (Very likely), but I recoded it into a
binary variable equal to 1 if a respondent said that they are likely or very likely to find themselves in unemployment or poverty. While sex, age and lack of qualifications are classic control variables (because temporally they are located before the association I investigate – the presence of immigrants in the area cannot influence the respondents’ sex, for example), the remaining variables might not be antecedents to the variables I investigated. In particular, this might apply to subjective views about one’s employment pros-pects; it is plausible that the presence of immigrants could influence individual evaluation of the chances for employment, which in turn affects attitudes toward immigrants (here subjective perception of employment prospects is an intermediate variable that mediates the impact of context on attitudes). On the other hand, not including individual-level variables would weaken my analysis because while I know that sex, age and educational qualification are strong predictors of anti-immigration sentiments, so are variables capturing economic deprivation (Lancee and Pardos-Prado, 2013).
All explanatory variables used to test hypotheses are contextual. To construct them I use small area (Output Area) statistics from the 2001 and 2011 censuses and aggregate them to 2011 Westminster con-stituencies. I assume that peoples’ expectations about the present are based on experiences from the past; hence, for base levels, I use the share of immigrants measured in 2001. Regarding changes in immigrant population, one can think of the change simply as the influx of people. However, this would overlook the fact that constituencies have different numbers of residents. Hence, I denominate the increase by the number of people in the constituency in 2001, creating what amounts to a measure of growth of immigrant populations. To test Hypotheses 1a, 1b and 1c I distinguish between different minority groups. In my models, I look at levels and changes with regard to ethnicity, religion and skills to see their role in the subsequent attitudes of white Britons toward immigrants. The construction of minority groups used in the study is described in Figure 1. For each constituency, I first subtracted the number of white British from all residents; the remaining population was then divided according to ethnicity (white and non-white), religion (Muslim and non-Muslim) and skills4(unskilled and skilled). For testing hypotheses 2 and 3 I calculated levels and changes and segregation indexes at constituency level. For segregation, I used an index of dissimilarity, calculated over 8546 wards and electoral divisions and 573 English and Welsh constituencies. The common interpretation of the dissimilarity index that it measures the share of the group’s population that would have to move for each smaller unit to have the same share of that group as the larger unit. Values range from 0, which indicate no segregation to 1, which means complete segregation. For each immigrant population the segregation index measures the extent to which it is segregated from the majority white British group.
Contextual variables for population density at the constituency level and the percentage of the unemployed are also included as well as a regional dummy variable for 10 government office regions in England and Wales (NUTS1). Descriptive statistics for dependent and independent variables are shown in Table 1 and distribution of respondents (weighted) in Table 2.
My data included information on individuals and the composition of the constituencies in which they are located. One method for analysing data in which individuals are nested within constituencies is multilevel modelling; my level 1 is the individual, and level 2 is constituency. Random coefficient models are preferred when dealing with nested sources of variability and when the association between dependent variables and level 1 might vary between level 2 units (Rabe-Hesketh and Skrondal, 2008). Regressions are weighted using the BES importance weights for Wave 4. In the appendix, I present random intercept models. There, I also present results from an unweighted random intercept model, following the advice of Winship and Radbill (1994) on the use of weighting in regression. These models give broadly similar results in terms of the main variables of interest; in particular, they all show that models based on ethnicity offer the greatest reduction in akaike information criterion (AIC) score and show the different effect of levels and changes.
Results
To assess whether local context is important for attitudes, I looked at the percentage of variance in attitudes explained by cross-constituency differences compared to individual differences. I ran the null
model with a constant and random variation between and within constituencies. This gives the Interclass Correlation (ICC), calculated by dividing the variance at the constituency level by the total variance. The
Figure 1. Measuring diversity according to different criteria: ethnic, combined ethnic and religion,
ICC shows that only about 6% of total variance in attitudes is attributable to a difference across constituencies, with the remaining 94% attributed to individual differences. This in itself is an important observation; natives’ views about immigration are not influenced by their local context.
The scale of my dependent variable ranges from1.57 (the most negative attitudes) to 1.86 (the most positive attitudes) with a standard deviation equal to 1. To test Hypotheses 1a–1c, I ran three separate
Table 1. Descriptive statistics.
obs. mean sd min max
Dependent variables Attitude index 17,373 –0.16 0.94 1.5 1.84 Independent variables Individual Male 17,373 0.52 0.5 0 1 Age 17,373 50.54 16.18 16 95 Risk of unemployment 17,373 0.14 0.35 0 1 Risk of poverty 17,373 0.23 0.42 0 1 Constituency No qualifications 573 0.23 0.06 0.1 0.39 Density 573 7.73 4.3 1.08 39.5
Level: white immigrants 573 0.04 0.04 0.01 0.3 Level: non-white immigrants 573 0.08 0.12 0 0.66 Change: white immigrants 573 0.02 0.02 0 0.12 Change: non-white immigrants 573 0.06 0.07 0 0.38 Segregation: white immigrants 573 0.15 0.08 0.02 0.47 Segregation: non-white immigrants 573 0.24 0.11 0.05 0.66 Level: unskilled immigrants 573 0.02 0.03 0 0.16 Level: skills immigrants 573 0.05 0.06 0.01 0.33 Change: unskilled immigrants 573 0.02 0.02 0 0.11 Change: skilled immigrants 573 0.04 0.04 0.01 0.23 Segregation: unskilled immigrants 573 0.25 0.11 0.04 0.63 Segregation: skilled immigrants 573 0.02 0.03 0 0.16 Level: Muslim immigrants 573 0.18 0.09 0.03 0.52 Level: non-Muslim immigrants 573 0.03 0.05 0 0.39 Change: Muslim immigrants 573 0.09 0.11 0.01 0.61 Change: non-Muslim immigrants 573 0.02 0.04 0.01 0.3 Segregation: Muslim immigrants 573 0.06 0.06 0.01 0.3 Segregation: non-Muslim immigrants 573 0.34 0.13 0.08 0.77
Table 2. Distribution of respondent in regions, with weighting.
Region %
North East 5.45
North West 12.02
Yorkshire and the Humber 9.84
East Midlands 8.39 West Midlands 9.46 East of England 10.58 London 10.89 South East 15.72 South West 10.36 Wales 7.29
models and examined which model led to the greatest reduction of AIC score compared to the base model (which did not include diversity measures). Akaike’s information criterion measures relative goodness of fit; it aids selection of a model that offers an acceptable fit to the data while minimizing the number of estimation parameters (Burnham and Anderson, 2004). The results are shown in Table 3 and Table 4. First, note that with a reduction in AIC score, models that include contextual diversity are much better than the base model. Model 1 is associated with the greatest reduction in AIC score compared to the base model (45,197.86 45,130.69 ¼ 67.17), suggesting that ethnic diversity with ethnic minority residents divided into white and non-white groups performs better at explaining variance in attitudes toward immigrants than criteria based on ethnicity and religion or ethnicity and skills. This suggests that ethnic criteria are the most salient. Yet, contrary to what was expected from Hypotheses 1a–c, higher 2001 levels of immigrant population are for all groups associated not with more negative but more positive attitudes towards immigrants, the only group for which the coefficient is not statistically significant at 0.05 level is the levels of unskilled ethnic minority in Model 3.
Hypothesis 2 about the impact of residential segregation is supported for the non-white ethnic minority. Rather surprisingly, the only other group for which segregation is statistically significant is the skilled ethnic minority group. The magnitude of the coefficient is similar in Model 1 and Model 3 (b¼ 0.438 for non-white ethnic minority and b ¼ 0.495 for skilled ethnic minority, but note that segregation has on average higher values for non-white ethnic minorities): both are negative suggesting that higher segregation translates to more negative attitudes towards immigrants.
In Model 1 and Model 2 the coefficient for levels is higher for non-visible minorities: whites (compare b ¼ 1.695 for white ethnic minority with b ¼ 1.023 for non-white ethnic minority) and non-Muslims (compare b¼ 1.443 for non-Muslim ethnic minority with b ¼ 0.95 for Muslim ethnic minority). What is associated with more negative views about immigrants is the change in visible minorities between 2001 and 2011: in constituencies experiencing larger growth of non-white ethnic minority and Muslim ethnic minority, the attitudes tend to be more negative (note that changes are not significant in Model 3 for skills). The different effect of levels and changes provides strong support to Hypotheses 3a–3c.
I also observed that the size of the coefficient for change was three times as large for change in Muslim ethnic minorities compared to that for non-white ethnic minorities (b¼ 2.433 compared with b¼ 0.849). This could suggest that change in the fraction of residents from this group is a source of greater threat, translating to more negative attitudes. To test Hypothesis 3c, I used the model with the lowest AIC score and added an interaction effect between the base levels and change for white and non-white ethnic minority groups. In Model 4, the interaction effect was statistically significant for non-non-whites (b¼ 3.148). Figure 2 shows how both variables affected the estimated attitude index. Attitudes toward immigrants were most positive in constituencies with larger base levels of non-white population and little change in the non-white population. Living in constituencies with low base levels of non-white population was associated with a more negative evaluation of immigrants; however, in this case, further changes did not seem to matter as much. For those living in constituencies with high base levels of non-white population, changes mattered more (note that the negative slope for the high base level is steeper). Model 4 also shows that higher levels of residential segregation were associated with more negative attitudes towards non-white ethnic minority groups; the effect was not significant for white ethnic minority groups, similar to the effect of levels and changes for this group. One possibility is that this is because of visibility; natives associated diversity with non-whites more because they are easier to notice.
Discussion
As observed by Fieldhouse and colleagues (2019), public concern tracks actual levels of immigration remarkably well in the United Kingdom, yet the mechanisms that connected the two are not well estab-lished. To what extent attitudes toward immigration are shaped by local demographic realities as opposed to other sources has been an issue of some debate. This article aims to contribute to this debate by testing
Table 3. Attitudes toward immigrants -random coefficient models. Fixed effects Null Base Ethnicity (Model 1) Ethnicity þ religion (Model 2) Skills (Model coef. se coef. se coef. se coef. se coef. Male 0.113 0.015 0.113 0.015 0.112 0.015 0.112 Age 0.038 0.003 0.037 0.003 0.037 0.003 0.037 Age squared 0 0.000 0 0.000 0 0.000 0 No qualifications 0.388 0.028 0.391 0.028 0.391 0.028 0.39 Risk of unemployment 0.083 0.025 0.08 0.025 0.081 0.025 0.081 Risk of poverty 0.183 0.020 0.184 0.020 0.185 0.020 0.185 Share with no qualifications 2.03 0.218 1.543 0.246 1.429 0.219 1.453 Population density 0.006 0.003 0 0.000 0.002 0.002 0.001 Region: (ref. North East) North West 0.049 0.052 0.059 0.048 0.056 0.049 0.057 Yorkshire and the Humber 0.011 0.055 0.018 0.050 0.022 0.051 0.011 East Midlands 0.021 0.057 0.032 0.054 0.034 0.053 0.002 West Midlands 0.011 0.052 0.066 0.052 0.094 0.052 0.067 East of England 0.048 0.054 0.107 0.052 0.111 0.052 0.091 London 0.039 0.057 0.313 0.066 0.38 0.069 0.337 South East 0.069 0.052 0.121 0.050 0.115 0.050 0.099 South West 0.021 0.054 0.018 0.053 0.005 0.045 0.015 Wales 0.041 0.058 0.013 0.054 0.005 0.056 0.013 Segregation: white ethnic minority 0.241 0.171 Levels: white ethnic minority 1.695 0.497 Change: white ethnic minority 1.335 0.763 Segregation: non-white ethnic minority 0.438 0.112 Levels: non-white ethnic minority 1.023 0.193 Change: non-white ethnic minority 0.849 0.290 Segregation: non-Muslim ethnic minority 0.111 0.163 Levels: non-Muslim ethnic minority 1.443 0.212 Change: non-Muslim ethnic minority 0.464 0.344 Segregation: Muslim ethnic minority 0.097 0.104 Levels: Muslim ethnic minority 0.96 0.382 (continued) 154
Table 3. (continued) Fixed effects Null Base Ethnicity (Model 1) Ethnicity þ religion (Model 2) Skills (Model coef. se coef. se coef. se coef. se coef. Change: Muslim ethnic minority 2.433 0.596 Segregation: skilled ethnic minority 0.495 Levels: skilled ethnic minority 1.839 Change: skilled ethnic minority 0.43 Segregation: unskilled ethnic minority 0.075 Levels: unskilled ethnic minority 1.157 Change: unskilled ethnic minority 1.171 Intercept 0.155 0.012 1.43 0.095 1.273 0.104 1.222 0.098 1.285 Random effects Male 0.015 0.01 0.016 0.01 0.016 0.01 0.016 A ge 00 0 0 0 0 00 No qualifications 0.047 0.02 0.043 0.02 0.044 0.02 0.045 Risk of unemployment 0.1 0.07 0.1 0.07 0.1 0.07 0.101 Risk of poverty 0.054 0.04 0.056 0.04 0.055 0.04 0.055 Intercept 0.052 0.01 0.022 0.02 0.015 0.01 0.016 0.01 0.018 Residual 0.826 0.01 0.747 0.73 0.747 0.73 0.747 0.73 0.747 N 17,373 17,373 17,373 17,373 17,373 AIC 46,518.88 45,197.86 45,130.69 45,137.10 45,169.64 coef: coefficient; se standard error; AIC: Akaike information criterion; p < .05 in bold. 155
the association between attitudes and levels and changes in ethnic minority residents defined in terms of ethnic, religious and skill-based criteria. I provide evidence for a weak effect of local experience.
In this study I have tested whether the same mechanisms apply to differently defined groups, an issue which has been highlighted by Pottie-Sherman and Wilkes (2017) who note that findings on the impact of local diversity on attitudes towards immigrants depend on the groups chosen to represent immigrant groups. I find that ethnic diversity is more salient in determining attitudes toward immigration,
Table 4. Attitudes toward immigrants - random coefficient models with interaction.
Interaction (Model 4)
Fixed effects coef. se
Male 0.113 0.015 Age 0.037 0.003 Age squared 0 0.000 No qualifications 0.391 0.028 Risk of unemployment 0.08 0.025 Risk of poverty 0.184 0.020
Share with no qualifications 1.602 0.258
Population density 0.002 0.003
Region: (ref. North East)
North West 0.061 0.049
Yorkshire and the Humber 0.021 0.051
East Midlands 0.031 0.053 West Midlands 0.078 0.052 East of England 0.103 0.054 London 0.318 0.067 South East 0.111 0.053 South West 0.007 0.054 Wales 0.011 0.058
Segregation: white ethnic minority 0.295 0.188 Levels: white ethnic minority 0.751 0.834 Change: white ethnic minority 0.468 1.265 Interaction: levels and change white ethnic minority 12.005 13.799 Segregation: non-white ethnic minority 0.534 0.119
Levels: non-white ethnic minority 1.631 0.334
Change: non-white ethnic minority 0.401 0.343 Interaction: levels and change non-white ethnic minority 3.148 1.363
Intercept 1.286 0.107 Random effects Male 0.016 0.01 Age 0 0 No qualifications 0.043 0.02 Risk of unemployment 0.1 0.07 Risk of poverty 0.055 0.04 Intercept 0.015 0.01 Residual N 17,373 AIC 45,128.74
compared with religious diversity or presence of low-skilled non-white British residents. Recent work by Jeannet (2020) shows that national level migration from E8 countries to Western European countries can be linked to public scepticism about the consequences of immigration, but my finding suggests that when it comes to attitudes toward immigration, direct exposure to low skilled migrants from Central and Eastern European may be less important than local exposure to ethnic diversity.
Similar to the findings reported by Laurence and colleagues (2019), I confirmed that context with higher segregation levels between white British and non-white residents is associated with more negative attitudes toward immigration, even net of demographic levels and changes. Furthermore, I found that base levels and changes in levels of non-white population are both associated with attitudes. I confirmed that pre-existing levels are associated with more positive attitudes, while change fosters more negative attitudes. This is similar to the argument made by Meuleman and colleagues (2009), who observe that changes in attitudes are driven by changes in diversity and the threat perceived by natives. While stable levels may not affect changes in attitudes towards minority groups (or they may be even associated with reduction of prejudice over time due to contact), further increase in diversity may lead to more anti-immigration sentiment. Importantly, the interaction effect found in this study differs from predictions derived from the ethnic competition theory or the defended neighbourhood hypothesis. I find that changes are associated with more hostility, and this effect is more pronounced in constituencies with higher pre-existing levels of non-white population. A plausible explanation is that while areas with higher pre-existing levels of non-white population offer more opportunities for contact, the speed of change undermines natives’ ability to accommodate further changes. In other words, the prejudice-reducing mechanisms of interpersonal contact cannot catch up with the threat responses triggered by the rapid growth of the immigrant population. This result can be discussed in reference to Bowyer’s influential study (2009), where the author found that natives in England do not view different ethnic minority groups as equally threatening. Bowyer reported that while the share of black residents at local ward level was associated with lower prejudice, the share of East Asian residents was associated with increased racial prejudice. The author suggested that this may be because of religion – black residents are predominantly Christian, while South Asian residents are not. An alternative explanation that is in
Figure 2. The effect of changes on attitudes towards immigrants at different base levels; predictive
line with my findings has to do with the length of stay – black migration in Britain has a longer history and perhaps more time to habituate than the more recent migration from Pakistan or Bangladesh.
When discussing the implications of these results, it is important to consider their limitations. My dependent variables rely on respondents’ evaluation of factual statements. Alternatively, one could ask respondents for their views about immigrants, but then I would run into a problem of desirability bias. I argue that the problem is to some degree alleviated because I created an index from three factual statements, referring to immigrants’ impact on the economy, culture and the welfare state. My measure of attitudes is also arguably better than that employed in another British study looking at the link between local levels and changes in diversity and attitudes towards immigrants (Kaufmann, 2017). The dependent variable used by Kaufmann is constructed from a question on whether there should be a reduction in immigration levels into the United Kingdom, the answers to which are skewed (81% of respondents thought immigration should be reduced) and may be too blunt a measure to capture attitudes towards immigrants.
A note of caution is also due in respect to making statements about the causal link between positive attitudes towards immigrants and diverse context. Dustmann and Preston (2001) have previously argued that ethnic composition should not be treated as an exogenous variable in attitudinal studies, especially if units of analysis are small. Finding that a higher share of immigrants is associated with more positive attitudes towards immigrants does not warrant a conclusion that the presence of immigrants positively influenced natives’ attitudes; an alternative interpretation would be that those with the most negative attitudes simply moved out. I do not have a direct way to ensure that this mechanism did not influence my results; however, I can cite a study by Kaufmann and Harris (2015), who showed that while it is true that white movers generally select whiter destinations than ethnic minorities, there is little difference in the diversity of the destinations of those who tend to be pro-immigration (that is, left leaning with a degree) and those who are usually against it (declaring an English identity).
Future research on contextual determinates of attitudes towards immigrants would benefit from further investigation of the dynamic nature of contractual effects, for instance by employing difference in difference design or panel data modelling. Nordic register data may provide a promising avenue for such research, if coupled with geocoded survey data. Lastly, let me reiterate that while this study shows there exists a weak contextual effect, the vast majority of variance in attitudes toward immigration is attributable to individual level differences. Future research should attempt to better understand what other factors, aside from local demographic realities, shape these attitudes.
Acknowledgements
Earlier versions of this paper were presented at the European Consortium for Sociological Research conference and the Institute for Analytical Sociology annual symposium. I am grateful to Michael Biggs, Jon Mellon, Anna Krusova, Tak Wing Chan and colleagues at the Institute for Analytical Sociology at Linko¨ping University for their feedback. I would also like to thank three anonymous reviewers for their valuable comments.
Funding
The author disclosed receipt of the following financial support for the research, authorship and/or publication of this article: The research was supported by the Economic and Social Research Council in the UK and Forte, Swedish Research Council for Health, Working Life (Grant Nos. 505175 and 2015-01200).
ORCID iD
Juta Kawalerowicz https://orcid.org/0000-0002-8473-8031
Notes
1. I selected England and Wales due to Census data availability. For the 2011 Census, these two countries had the widest coverage, including small-scale data on religion and ethnicity.
2. Although, see Kaufmann and Goodwin (2018) who argue that at very small geographical scales threat works more strongly than contact.
3. Some European cross-national studies that include Eastern European counties show that the actual or perceived size of the immigrant population at the country level is unrelated to attitudes toward immigrants (Hjerm, 2007; Sides and Citrin, 2007).
4. I used the NS-SEC scale as a proxy for skills. Those in class 5 (semi-routine and routine occupations) in the five-class division were considered low-skilled.
References
Allport GW (1954) The Nature of Prejudice. Cambridge, Mass: Addison-Wesley. Blalock HM (1967) Toward a Theory of Minority-Group Relations. New York: Wiley.
Blinder S (2015) Imagined immigration: The impact of different meanings of ‘Immigrants’ in public opinion and policy debates in Britain. Political Studies 63(1): 80–100.
Blinder S and Allen W (2016) UK Public Opinion toward Immigration: Overall Attitudes and Level of Concern. Migration Observatory Briefing, COMPAS, University of Oxford.
Bowyer BT (2009) The contextual determinants of whites’ racial attitudes in England. British Journal of Political Science 39(03): 559–586.
Burnham KP and Anderson DR (2004) Multimodel inference understanding AIC and BIC in model selection. Sociological Methods & Research 33(2): 261–304.
Card D, Dustmann C and Preston I (2005) Understanding attitudes to immigration: The migration and minority module of the first European social survey. IDEAS Working Paper Series from RePEc, IDEAS Working Paper Series from RePEc.
Card D, Dustmann C and Preston I (2012) Immigration, wages, and compositional amenities. Journal of the European Economic Association 10(1): 78–119.
Dixon JC and Rosenbaum MS (2004) Nice to know you? Testing contact, cultural, and group threat theories of anti-black and anti-Hispanic stereotypes. Social Science Quarterly 85(2): 257–280. Dustmann C and Preston I (2001) Attitudes to ethnic minorities, ethnic context and location decisions.
The Economic Journal 111(470): 353–373.
Ellison CG, Shin H and Leal DL (2011) The contact hypothesis and attitudes toward Latinos in the United States. Social Science Quarterly 92(4): 938–958.
Evans G and Mellon J (2019) Immigration, Euroscepticism, and the rise and fall of UKIP. Party Politics 25(1): 76–87.
Evans G and Need A (2002) Explaining ethnic polarization over attitudes towards minority rights in Eastern Europe: A multilevel analysis. Social Science Research 31(4): 653–680.
Fieldhouse E, Green J and Evans G (2019) Electoral Shocks: The Volatile Voter in a Turbulent World. Oxford: Oxford University Press.
Ford R (2011) Acceptable and unacceptable immigrants: How opposition to immigration in Britain is affected by migrants’ region of origin. Journal of Ethnic and Migration Studies 37(7): 1017–1037. Fox C (2004) The changing color of welfare? How whites’ attitudes toward Latinos influence support for
welfare. American Journal of Sociology 110(3): 580–625.
Gorodzeisky A (2011) Who are the Europeans that Europeans prefer? Economic conditions and exclu-sionary views toward European immigrants. International Journal of Comparative Sociology 52(1–2): 100–113.
Green DP, Strolovitch DZ and Wong JS (1998) Defended neighborhoods,Integration, and racially motivated crime. American Journal of Sociology 104(2): 372–403.
Ha SE (2010) The consequences of multiracial contexts on public attitudes toward immigration. Polit-ical Research Quarterly 63(1): 29–42.
Hainmueller J and Hiscox MJ (2010) Attitudes toward highly skilled and low-skilled immigration: Evidence from a survey experiment. American Political Science Review 104(01): 61–84.
Hainmueller J and Hopkins DJ (2014) Public attitudes toward immigration. Annual Review of Political Science 17(1): 225–249.
Hjerm M (2007) Do numbers really count? Group threat theory revisited. Journal of Ethnic and Migra-tion Studies 33(8): 1253–1275.
Hjerm M (2009) Anti-immigrant attitudes and cross-municipal variation in the proportion of immi-grants. Acta Sociologica 52(1): 47–62.
Hobolt SB (2016) The Brexit vote: A divided nation, a divided continent. Journal of European Public Policy 23(9): 1259–1277.
Hopkins DJ (2010) Politicised places: Explaining where and when immigrants provoke local opposition. American Political Science Review 104(01): 40–60.
Jeannet AM (2020) A threat from within? Perceptions of immigration in an enlarging European Union. Acta Sociologica 63(4): 343–360.
Kaufmann E (2017) Levels or changes? Ethnic context, immigration and the UK Independence Party vote. Electoral Studies 46: 57–69.
Kaufmann E and Goodwin M (2018) The diversity Wave: A meta-analysis of the native-born white response to ethnic diversity. Social Science Research 76: 120–131.
Kaufmann E and Harris G (2015) ‘White flight’ or positive contact? Local diversity and attitudes to immigration in Britain. Comparative Political Studies 48(12): 1563–1590.
Lancee B and Pardos-Prado S (2013) Group conflict theory in a longitudinal perspective: Analyzing the dynamic side of ethnic competition. International Migration Review 47(1): 106–131.
Laurence J, Schmid K and Rae JR (2019) Prejudice, contact, and threat at the diversity-segregation nexus: A cross-sectional and longitudinal analysis of how ethnic out-group size and segregation interrelate for inter-group relations. Social Forces 97(3): 1029–1066.
McLaren LM (2003) Anti-immigrant prejudice in Europe: Contact, threat perception, and preferences for the exclusion of migrants. Social Forces 81(3): 909–936.
McLaren L and Johnson M (2007) Resources, group conflict and symbols: Explaining anti-immigration hostility in Britain. Political Studies 55(4): 709–732.
Markaki Y (2014) Public support for immigration restriction in the United Kingdom: Resource scarcity, ethnicity or poor origins? National Institute Economic Review 229(1): R31–52.
Markaki Y and Longhi S (2013) What determines attitudes to immigration in European countries? An analysis at the regional level. Migration Studies 1(3): 311–337.
Meuleman B, Davidov E and Billiet J (2009) Changing attitudes toward immigration in Europe, 2002–2007: A dynamic group conflict theory approach. Social Science Research 38(2): 352–365. Newman BJ (2013) Acculturating contexts and Anglo opposition to immigration in the United States.
American Journal of Political Science 57(2): 374–390.
Oliver J and Wong J (2003) Intergroup prejudice in multiethnic settings. American Journal of Political Science 47(4): 567–582.
Olzak S (1992) The Dynamics of Ethnic Competition and Conflict. Stanford: Stanford University Press. Pettigrew TF (1998) Reactions toward the new minorities of Western Europe. Annual Review of
Sociol-ogy 24: 77–103.
Pottie-Sherman Y and Wilkes R (2017) Does size really matter? On the relationship between immigrant group size and anti-immigrant prejudice. International Migration Review 51(1): 218–250.
Quillian L (1995) Prejudice as a response to perceived group threat: Population composition and anti-immigrant and racial Prejudice in Europe. American Sociological Review 60(4): 586–611. Quillian L (1996) Group threat and regional change in attitudes toward African-Americans. American
Journal of Sociology 102(3): 816–860.
Rabe-Hesketh S and Skrondal A (2008) Multilevel and Longitudinal Modeling Using Stata. Texas: Stata Press.
Rink N, Phalet K and Swyngedouw M (2009) The effects of immigrant population size, unemployment, and individual characteristics on voting for the vlaams blok in flanders 1991–1999. European Socio-logical Review 25(4): 411–424.
Rocha RR and Espino R (2009) Racial threat, residential segregation and the policy attitudes of Anglos. Political Research Quarterly 62(2): 415–426.
Rose EJB (1969) Colour and Citizenship: A Report on British Race Relations. London: Oxford Uni-versity Press.
Scheepers P, Gijsberts M and Coenders M (2002) Ethnic exclusionism in European countries: Public opposition to civil rights for legal migrants as a response to perceived ethnic threat. European Sociological Review 18(1): 17–34.
Schlueter E and Scheepers P (2010) The relationship between outgroup size and anti-outgroup attitudes: A theoretical synthesis and empirical test of group threat- and intergroup contact theory. Social Science Research 39(2): 285–295.
Schneider SL (2008) Anti-immigrant attitudes in Europe: Outgroup size and perceived ethnic threat. European Sociological Review 24(1): 53–67.
Semyonov M, Raijman R and Gorodzeisky A (2006) The rise of anti-foreigner sentiment in European societies, 1988–2000. American Sociological Review 71(3): 426–449.
Semyonov M, Raijman R and Gorodzeisky A (2008) Foreigners’ impact on European societies public views and perceptions in a cross-national comparative perspective. International Journal of Com-parative Sociology 49(1): 5–29.
Semyonov M, Raijman R and Tov AY (2004) Population size, perceived threat, and exclusion: A multiple-indicators analysis of attitudes toward foreigners in Germany. Social Science Research 33(4): 681–701.
Sides J and Citrin J (2007) European opinion about immigration: The role of identities, interests and information. British Journal of Political Science 37(3): 477–504.
Taylor MC (1998) How white attitudes vary with the racial composition of local populations: Numbers count. American Sociological Review 63(4): 512–535.
Van Gent WPC, Jansen EF and Smits JHF (2014) Right-wing radical populism in city and suburbs: an electoral geography of the Partij Voor de Vrijheid in the Netherlands. Urban Studies 51(9): 1775–1794. Van Heerden S and Ruedin D (2019) How attitudes towards immigrants and shaped by residential context:
The role of ethnic diversity dynamics and immigrant visibility. Urban Studies 56(2): 317–334. Wagner U, van Dick R and Pettigrew TF (2003) Ethnic prejudice in East and West Germany: The
explanatory power of intergroup contact. Group Processes & Intergroup Relations 6(1): 22–36. Winship C and Radbill L (1994) Sampling weights and regression analysis. Sociological Methods &
Research 23(2): 230–257.
Author biography
Juta Kawalerowicz received her doctorate from Nuffield College at the University of Oxford. She is a researcher at the Department of Human Geography at Stockholm University in Sweden. Her research focuses on urban unrest, radical right wing politics and segregation.
Appendix
Table A1. Attitudes toward immigrants – random intercept models. Fixed effects Base Ethnicity Ethnicity þ religion Ethnicity þ skills coef. se coef. se coef. se coef. Male 0.112 0.014 0.112 0.014 0.112 0.014 0.112 Age 0.038 0.003 0.038 0.003 0.038 0.003 0.038 Age squared 0 0.000 0 0.000 0 0.000 0 No qualifications 0.388 0.025 0.39 0.025 0.39 0.025 0.39 Risk of unemployment 0.086 0.020 0.085 0.020 0.086 0.020 0.086 Risk of poverty 0.183 0.017 0.184 0.017 0.185 0.017 0.184 Share with no qualifications 2 0.215 1.605 0.244 1.425 0.217 1.518 Population density 0.006 0.003 0 0.000 0.002 0.003 0.001 Region: (ref. North East) North West 0.054 0.051 0.064 0.048 0.06 0.048 0.062 Yorkshire and the Humber 0.019 0.054 0.023 0.050 0.028 0.050 0.017 East Midlands 0.018 0.056 0.034 0.054 0.037 0.053 0.005 West Midlands 0.017 0.052 0.066 0.051 0.094 0.051 0.069 East of England 0.057 0.053 0.114 0.052 0.124 0.052 0.098 London 0.026 0.055 0.285 0.066 0.364 0.069 0.295 South East 0.076 0.052 0.126 0.050 0.127 0.050 0.107 South West 0.004 0.050 0.045 0.052 0.024 0.051 0.016 Wales 0.036 0.058 0.009 0.056 0.001 0.100 0.009 Segregation: white ethnic minority 0.226 0.171 Levels: white ethnic minority 1.305 0.489 Change: white ethnic minority 1.579 0.752 Segregation: non-white ethnic minority 0.491 0.112 Levels: non-white ethnic minority 0.904 0.190 Change: non-white ethnic minority 0.695 0.286 Segregation: non-Muslim ethnic minority 0.153 0.163 Levels: non-Muslim ethnic minority 1.271 0.209 Change: non-Muslim ethnic minority 0.717 0.341 (continued) 162Table A1. (continued) Fixed effects Base Ethnicity Ethnicity þ religion Ethnicity þ skills coef. se coef. se coef. se coef. Segregation: Muslim ethnic minority 0.114 0.103 Levels: Muslim ethnic minority 0.897 0.380 Change: Muslim ethnic minority -2.437 0.587 Segregation: skilled ethnic minority 0.507 Levels: skilled ethnic minority 1.297 Change: skilled ethnic minority 0.635 Segregation: unskilled ethnic minority 0.011 Levels: unskilled ethnic minority 1.283 Change: unskilled ethnic minority 0.747 Intercept 1.435 0.094 1.327 0.103 1.249 0.098 1.335 Random effects Intercept 0.028 0.02 0.022 0.02 0.022 0.02 0.025 Residual 0.778 0.76 0.778 0.76 0.778 0.76 0.778 N 17,373 17,373 17,373 17,373 AIC 45,335.79 45,277.10 45,280.78 45,306.57 coef: coefficient; se standard error; AIC: Akaike information criterion; p < .05 in bold. 163
Table A2. Attitudes towards immigrants – random intercept, without weighting. Fixed effects Base Ethnicity Ethnicity þ religion Ethnicity þ skills coef. se coef. se coef. se coef. Male 0.067 0.014 0.068 0.014 0.068 0.014 0.067 Age 0.039 0.003 0.039 0.003 0.039 0.003 0.039 Age squared 0 0.000 0 0.000 0 0.000 0 No qualifications 0.439 0.026 0.44 0.026 0.439 0.026 0.44 Risk of unemployment 0.045 0.021 0.043 0.021 0.044 0.021 0.045 Risk of poverty 0.206 0.018 0.208 0.018 0.207 0.018 0.208 Share with no qualifications 2.348 0.195 1.834 0.212 1.793 0.191 1.925 Population density 0.005 0.002 0 0.000 0.002 0.002 0.001 Region: (ref. North East) North West 0.054 0.048 0.067 0.044 0.064 0.045 0.067 Yorkshire and the Humber 0.011 0.048 0.023 0.045 0.025 0.045 0.012 East Midlands 0.018 0.050 0.04 0.048 0.032 0.048 0.008 West Midlands 0.01 0.050 0.063 0.047 0.087 0.047 0.071 East of England 0.078 0.049 0.145 0.046 0.14 0.047 0.121 London 0.117 0.051 0.181 0.058 0.235 0.061 0.162 South East 0.089 0.048 0.144 0.045 0.136 0.045 0.123 South West 0.026 0.050 0.071 0.046 0.041 0.046 0.045 Wales 0.162 0.050 0.133 0.045 0.128 0.045 0.134 Segregation: white ethnic minority 0.084 0.150 Levels: white ethnic minority 1.595 0.411 Change: white ethnic minority 2.215 0.650 Segregation: non-white ethnic minority 0.368 0.099 Levels: non-white ethnic minority 0.973 0.166 Change: non-white ethnic minority 0.885 0.249 Segregation: non-Muslim ethnic minority 0.039 0.139 Levels: non-Muslim ethnic minority 1.384 0.179 Change: non-Muslim ethnic minority 0.38 0.302 Segregation: Muslim ethnic minority 0.182 0.091 Levels: Muslim ethnic minority 1.087 0.333 (continued) 164
Table A2. (continued) Fixed effects Base Ethnicity Ethnicity þ religion Ethnicity þ skills coef. se coef. se coef. se coef. Change: Muslim ethnic minority -2.287 0.525 Segregation: skilled ethnic minority 0.656 Levels: skilled ethnic minority 0.904 Change: skilled ethnic minority 0.656 Segregation: unskilled ethnic minority 0.101 Levels: unskilled ethnic minority 2.682 Change: unskilled ethnic minority -1.084 Intercept 1.66 0.091 1.507 0.096 1.487 0.093 1.586 Random effects Intercept 0.016 0.01 0.008 0.01 0.009 0.01 0.011 Residual 0.839 0.82 0.839 0.82 0.839 0.82 0.839 N 17,373 17,373 17,373 17,373 AIC 46,554.67 46,454.99 46,470.78 46,499.30 coef: coefficient; se standard error; AIC: Akaike information criterion; p < .05 in bold. 165
