Published online by Cambridge University Press: 01 July 2009
How do majority group members in emerging multicultural and multiracial societies respond to the experience of living amidst ethnic diversity? Recent public opinion surveys are analysed to assess the contextual determinants of English whites’ opinions towards ethnic minorities and immigrants. Multilevel analyses reveal that whites’ racial hostility is affected by local ethnic context; however, the direction of this effect depends on which ethnic minority groups reside in the area. Consistent with the contact hypothesis, whites who live in neighbourhoods with relatively large black populations display lower levels of racial hostility than respondents with few black neighbours. However, in line with racial threat theory, residential proximity to Pakistanis and Bangladeshis is associated with more negative attitudes towards ethnic minorities.
1 As cited in Bill Smithies and Peter Fiddick, Enoch Powell on Immigration (London: Sphere, 1969), p. 37.
2 Throughout this article, the terms ‘non-white’, ‘ethnic minority’, ‘racial minority’ and their respective variants are used interchangeably. While ‘race’ and ‘ethnicity’ are often seen as analytically distinct phenomena, typical use in the British context does not readily distinguish between the two concepts. For instance, the standard British Census ‘ethnic group’ categories of ‘White’, ‘Black-Caribbean’, ‘Black-African’, ‘Black-Other’, ‘Indian’, ‘Pakistani’, ‘Bangladeshi’, ‘Chinese’, ‘Other-Asian’, ‘Other-Other’, and the various ‘Mixed’ groups are a somewhat haphazard combination of racial, national and ethnic labels. This convoluted classification scheme underscores that these categories are social constructions.
3 R. Robert Huckfeldt, ‘Variable Responses to Neighborhood Social Contexts: Assimilation, Conflict, and Tipping Points’, Political Behavior, 2 (1980), 231–57, p. 252.
4 R. Robert Huckfeldt, Politics in Context: Assimilation and Conflict in Urban Neighborhoods (New York: Agathon Press, 1986), chap. 4.
5 David Butler and Donald Stokes, Political Change in Britain: The Evolution of Electoral Choice (New York: St Martin’s Press, 1974), chap. 6.
6 Ron Johnston and Charles Pattie, ‘Putting Voters in Their Places: Local Context and Voting in England and Wales, 1997’, in Alan S. Zuckerman, ed., The Social Logic of Politics: Personal Networks as Contexts for Political Behavior (Philadelphia: Temple University Press, 2005).
7 Iain MacAllister, Ron Johnston, Charles Pattie, Helena Tunstall, Daniel Dorling and David Rossiter, ‘Class Dealignment and the Neighbourhood Effect: Miller Revisited’, British Journal of Political Science, 31 (2001), 41–59.
8 Ron Johnston, Charles Pattie, Daniel Dorling, Iain MacAllister, Helena Tunstall and David Rossiter, ‘Housing Tenure, Local Context, Scale and Voting in England and Wales, 1997’, Electoral Studies, 20 (2001), 195-216.
9 Gordon W. Allport, The Nature of Prejudice (Reading, Mass.: Addison-Wesley, 1979).
10 Yehuda Amir, ‘Contact Hypothesis in Ethnic Relations’, Psychological Bulletin, 71 (1969), 319–42; Thomas F. Pettigrew, ‘Intergroup Contact Theory’, Annual Review of Psychology, 49 (1998), 65–85.
11 H. D. Forbes, Ethnic Conflict: Commerce, Culture, and the Contact Hypothesis (New Haven, Conn.: Yale University Press, 1997); Pettigrew, ‘Intergroup Contact Theory’.
12 See, for example, Mary R. Jackman and Marie Crane, ‘ “Some of My Best Friends Are Black”: Interracial Friendship and Whites’ Racial Attitudes’, Public Opinion Quarterly, 50 (1986), 459–86; Jürgen Hamberger and Miles Hewstone, ‘Inter-Ethnic Contact as a Predictor of Blatant and Subtle Prejudice: Tests of a Model in Four West European Nations’, British Journal of Social Psychology, 36 (1997), 173–90; Robert M. Stein, Stephanie Shirley Post and Allison L. Rinden, ‘Reconciling Context and Contact Effects on Racial Attitudes’, Political Research Quarterly, 53 (2000), 285–303.
13 Forbes, Ethnic Conflict; Pettigrew, ‘Intergroup Contact Theory’.
14 Jackman and Crane, ‘ “Some of My Best Friends Are Black”’; Susan Welch, Lee Sigelman, Timothy Bledsoe and Michael Combs, Race and Place: Race Relations in an American City (Cambridge: Cambridge University Press, 2001).
15 Forbes, Ethnic Conflict.
16 Lincoln Quillian, ‘Prejudice as a Response to Perceived Group Threat: Population Composition and Anti-Immigrant and Racial Prejudice in Europe’, American Sociological Review, 60 (1995), 586–611.
17 James M. Glaser, ‘Back to the Black Belt: Racial Environment and White Racial Attitudes in the South’, Journal of Politics, 54 (1994), 21–41.
18 V. O. Key Jr, Southern Politics in State and Nation (New York: A. A. Knopf, 1949).
19 Michael W. Giles and Melanie A. Buckner, ‘David Duke and Black Threat: An Old Hypothesis Revisited’, Journal of Politics, 55 (1993), 702–13.
20 Lincoln Quillian, ‘Group Threat and Regional Change in Attitudes towards African-Americans’, American Journal of Sociology, 102 (1996), 816–60.
21 Quillian, ‘Prejudice as a Response to Perceived Group Threat’.
22 D. Stephen Voss, ‘Beyond Racial Threat: Failure of an Old Hypothesis in the New South’, Journal of Politics, 58 (1996), 1156–70.
23 J. Eric Oliver and Tali Mendelberg, ‘Reconsidering the Environmental Determinants of White Racial Attitudes’, American Journal of Political Science, 44 (2000), 574–89, p. 575.
24 J. Eric Oliver and Janelle Wong, ‘Intergroup Prejudice in Multiethnic Settings’, American Journal of Political Science, 47 (2003), 567–82.
25 Christopher Bagley, Social Structure and Prejudice in Five English Boroughs: A Report Prepared for the I.R.R. Survey of Race Relations in Britain (London: Institute of Race Relations, 1970); Clifford S. Hill, How Colour Prejudiced Is Britain? (London: Gollancz, 1965).
26 Richard T. Schaefer, ‘Regional Differences in Prejudice’, Regional Studies, 9 (1975), 1–14; E. J. B. Rose, Nicholas Deakin, Mark Abrams, Valerie Jackson, Maurice Peston, A. H. Vanags, Brian Cohen, Julia Gaitskell and Paul Ward, Colour and Citizenship: A Report on British Race Relations (London: Institute of Race Relations, 1969); Bagley, Social Structure and Prejudice in Five English Boroughs.
27 Hill, How Colour Prejudiced Is Britain? p. 12.
28 When asked about the ‘way of life’ in the immigrants’ countries of origin, more white respondents agreed that life in Britain was different from life in India or in Pakistan than from life in the West Indies (see Mark Abrams, ‘The Incidence of Race Prejudice in Britain’, in E. J. B. Rose, Nicholas Deakin, Mark Abrams, Valerie Jackson, Maurice Peston, A. H. Vanags, Brian Cohen, Julia Gaitskell and Paul Ward, Colour and Citizenship: A Report on British Race Relations (London: Institute of Race Relations, 1969), p. 568). At the same time, though, respondents who agreed with the assertion that members of these immigrant groups take ‘more out of the country than they put into it’ did not seem to distinguish between West Indians, Indians and Pakistanis (Abrams, ‘The Incidence of Race Prejudice in Britain’, p. 570).
29 Abrams, ‘The Incidence of Race Prejudice in Britain’, pp. 582–83.
30 Christian Dustmann and Ian Preston, ‘Attitudes to Ethnic Minorities, Ethnic Context and Location Decisions’, Economic Journal, 111 (2001), 353–73.
31 Commission for Racial Equality, ‘Stereotyping and Racism: Findings from Two Attitude Surveys’ (London: 1998).
32 Commission for Racial Equality, ‘Stereotyping and Racism’, p. 20.
33 Wards are the smallest geographical units available for the use of survey respondents and they roughly correspond to the concept of a neighbourhood. However, because ward boundaries change frequently, the wards in the surveys do not match up neatly with the wards in the Census. Moreover, the lowest geographic level available in the 1999 BSA was the postal district. Consequently, the Census statistics had to be recalculated to reflect the boundaries appropriate for each survey. The effect of these adjustments was to increase somewhat the average size of the contextual units. The population of the median ward in the combined sample was 8,118. These wards are nested within local authority districts (i.e., metropolitan districts, non-metropolitan districts, unitary authorities, the City of London and the various London boroughs). Districts range in population size from about 40,000 residents to just under 1,000,000 residents, with the median respondent in the pooled sample living in a district with a population of about 140,000.
34 None of these measures is a perfect measure of racial hostility, and the five questions appear to tap into different dimensions of the overarching concept. The first item, the self-reported prejudice question, is especially likely to be subject to social desirability biases: many people who are indeed prejudiced against ethnic minorities will not admit this when asked so directly. Consequently, I interpret affirmative responses to this question not so much as an ideal measure of prejudice but more as an expression of negative affect towards racial minorities. Responses to this question have been shown to be good predictors of responses to other measures of racial affect – see, for example, Geoffrey Evans, ‘In Search of Tolerance’, in A. Park, J. Curtice, K. Thomson, L. Jarvis and C. Bromley, eds, British Social Attitudes: The 19th Report (London: Sage, 2002), pp. 213–30. The second item, the equal opportunities question, seems to tap into a sense of ‘racial resentment’ that ethnic minority groups receive more than they deserve from the government – see Donald R. Kinder and Lynn M. Sanders, Divided by Color: Racial Politics and Democratic Ideals (Chicago: University of Chicago Press, 1996), chap. 5. The three final items refer to perceptions of the consequences of immigration: one of these does so in racially specific terms, the other two do not. Because the five questions were not all asked in the same years of the surveys, they cannot be combined into a factor score. Instead, the analytical approach here is to treat them all as imperfect indicators of the same overarching concept and to look for patterns that hold true across the five measures. Also, to test the racial threat and contact hypotheses, it would be better to analyse questions relating to specific minority groups rather than vague categories like ‘people of other races’, ‘black people and Asians’, or ‘immigrants’. Nonetheless, these are the best measures available in these surveys, and these surveys are valuable for testing these theories, because they provide large national samples of the contextual units needed for the multilevel analyses.
35 In deciding whether to combine the various ‘ethnic’ categories listed in the Census or to treat them as separate groups, the most important issue here is whether whites will react differently to members of the various ethnic groups. Accordingly, the decision to combine all of the various ‘black’ groups was based on (perceived) racial similarities between them. Following the same logic, it might also make sense to combine the various South Asian groups into one category. However, it is thought that whites’ reactions to members of this group might be affected not only by perceptions of racial difference but also by perceptions of cultural difference. The Pakistani and Bangladeshi groups are similar to one another, and distinct from Indians and all other ethnic groups in Britain, in terms of average household size, age distribution and female economic inactivity rates. Even more notably, the overwhelming majority (greater than 90 per cent) of both the ‘Pakistani’ and ‘Bangladeshi’ groups identify as Muslim, while nearly three quarters of the ‘Indian’ group identify as Hindu or Sikh. This might suggest that a more appropriate measure of ethnic context would rely on the Census question on religious affiliation. However, even if it is possible to separate ‘religious denomination’ and ‘ethnic group’ at a conceptual level, it is virtually impossible to do so at an empirical level. The percentage of a ward’s population that identifies as Muslim on the Census’s religion question is strongly correlated with the percentage that identifies as Pakistani or Bangladeshi on the ethnicity question (r = 0.93). Consequently, it is not possible to isolate the effects of ethnic difference from those of religious difference with aggregate-level statistics. Moreover, relying on the religion variable as a measure of ethnicity would not allow one to distinguish between the white and black groups, as the majority of both groups identify with a Christian denomination. Finally, a significant proportion (8 per cent) of the English population refused to answer the religion question on the Census (even declining to state ‘no religion’), making this item less useful than the ethnicity measure.
36 Stanley Lieberson, A Piece of the Pie: Blacks and White Immigrants Since 1880 (Berkeley: University of California Press, 1980), chap. 9; Stanley Lieberson, ‘An Asymmetrical Approach to Segregation’, in Ceri Peach, Vaughan Robinson and Susan Smith, eds, Ethnic Segregation in Cities (Athens: University of Georgia Press, 1981), pp. 61–82; Douglas S. Massey and Nancy A. Denton, ‘The Dimensions of Residential Segregation’, Social Forces, 67 (1988), 281–315.
37 Formally, the exposure of members of group X to members of group Y (xP*y) is calculated using the formula
where X is the total number of members of group X in the ward, xi is the number of group X members in output area i, yi is the number of group Y members in the output area, and ti is the total population of the output area. For example, exposure of whites to blacks is calculated by summing across all the output areas in a ward the product of the proportion of the ward’s white population that lives in the output area and the proportion of the output area’s population that is black. The exposure indices were calculated in stata 8, using the ‘seg’ module written by Sean Reardon.
38 Andrew Geddes, ‘Asian and Afro-Caribbean Representation in Local Government in England and Wales’, New Community, 20 (1993), 43–57.
39 Geoff Dench, Kate Gavron and Michael Young. The New East End: Kinship, Race and Conflict (London: Profile Books, 2006).
40 Herman Ouseley, Community Pride, not Prejudice: Making Diversity Work in Bradford (Bradford: Bradford Vision, 2001).
41 The ethnic group questions used in the Census of Population changed between 1991 and 2001, making direct comparisons between the two censuses difficult. Most notably, the censuses differed in their instructions for people who considered themselves to belong to more than one ethnic group. The 1991 Census asked such respondents to select one group only (including the residual ‘Any other ethnic group’ category), while the 2001 Census allowed respondents to select from a variety of ‘mixed’ categories (e.g., ‘White and Black Caribbean’). In order to match the 1991 categories most closely, these mixed ethnicity people are not included in the calculation of the changes in the relative size of the ethnic minority groups. The statistics for the 1991 Census were calculated according to 2001 district boundaries, ensuring that the geographic units were the same in both years.
43 Stephen W. Raudenbush and Anthony S. Bryk, Hierarchical Linear Models: Applications and Data Analysis Methods (Thousand Oaks, Calif.: Sage, 2002); Joop Hox, Multilevel Analysis: Techniques and Applications (Mahwah, N.J.: Lawrence Erlbaum, 2002).
44 Raudenbush and Bryk, Hierarchical Linear Models.
45 All of the models were estimated using the hlm 6 software application. The self-reported prejudice question is a binary variable and all of the analyses of this variable use a Bernoulli (logit) link function. The other four dependent variables are ordinal variables, so ordered logit link functions are used. All of the independent variables are centred around their sample mean prior to estimation.
46 The effects of the full set of individual-level control variables and year of survey dummy variables are not displayed, but are estimated in all of the analyses. These results are available from the author upon request.
47 The Wald test that two estimated coefficients (β 1 and β 2) are equal is calculated using the formula
The W statistic is distributed as chi-square with 1 degree of freedom. See J. Scott Long, Regression Models for Categorical and Limited Dependent Variables (Thousand Oaks, Calif.: Sage, 1997), p. 93.
48 More specifically, the predicted probabilities are calculated for a hypothetical female respondent born between 1950 and 1959, who owns the home in which she lives, is an economically active member of the routine non-manual class, left school at age 16 or 17 without a degree, does not identify with any religious denomination, never attends religious services, does not belong to a trade union or staff association, and does not rely on a state benefit as the main source of family income. For each dependent variable, the probabilities are calculated for the first year when the question was asked on the surveys (i.e., the baseline respondent for the self-reported prejudice item participated in the 1996 BSA, the baseline for the questions about equal opportunities initiatives and the overall effect of immigration is the 1997 BGES, and for the items about the economic and cultural effects of immigrants it is the 1999 BSA). The other ward-level variables are held constant at their means (e.g., in calculating the effect of variation in percentage black on predicted racial hostility, the percentage Indian and percentage Pakistani/Bangladeshi in the ward are set to their respective means).
49 Hamberger and Hewstone, ‘Inter-Ethnic Contact as a Predictor of Blatant and Subtle Prejudice’.
50 Dustmann and Preston, ‘Attitudes to Ethnic Minorities, Ethnic Context and Location Decisions’.
51 R. Johnston, J. Forrest and M. Poulsen, ‘Are There Ethnic Enclaves/Ghettos in English Cities?’ Urban Studies, 39 (2002), 591–618; Ceri Peach and David Rossiter, ‘Level and Nature of Spatial Concentration and Segregation of Minority Ethnic Populations in Great Britain, 1991’, in Peter Ratcliffe, ed., Social Geography and Ethnicity in Britain: Geographical Spread, Spatial Concentration and Internal Migration, Ethnicity in the 1991 Census (London: HMSO, 1996).
52 Richard Dorsett, Ethnic Minorities in the Inner City (Bristol: Policy Press, 1998); Ceri Peach, ‘South Asian and Caribbean Ethnic Minority Housing Choice in Britain’, Urban Studies, 35 (1998), 1657–80; Vaughan Robinson, ‘Asians in Britain: A Study in Encapsulation and Marginality’, in Colin G. Clarke, David Ley, Ceri Peach and Paul Paget, eds, Geography & Ethnic Pluralism (Boston, Mass.: G. Allen & Unwin, 1984).
53 Ron Johnston, Carol Propper, Simon Burgess, Rebecca Sarker, Anne Bolster and Kelvyn Jones, ‘Spatial Scale and the Neighbourhood Effect: Multinomial Models of Voting at Two Recent British General Elections’, British Journal of Political Science, 35 (2005), 487–514.
54 There are some limits to the comparability of this study to that of Johnston et al. For one thing, their analysis is not limited to white respondents and they do not include race/ethnicity as an individual-level control variable. Consequently, some of the estimated effects of the ethnic make-up of a neighbourhood could reflect compositional effects (e.g., if non-white survey respondents were more supportive of the Labour party than their white counterparts). Additionally, their measure of ethnic context is based on a factor score that includes the relative sizes of the black, Indian and Pakistani/Bangladeshi groups as well as the percentage of an area’s households that live at high densities per room.
55 Paul M. Sniderman, Louk Hagendoorn and Markus Prior, ‘Predisposing Factors and Situational Triggers: Exclusionary Reactions to Immigrant Minorities’, American Political Science Review, 98 (2004), 35–49; John Sides and Jack Citrin, ‘European Opinion about Immigration: The Role of Identities, Interests and Information’, British Journal of Political Science, 37 (2007), 477–504.