Moving up? Intergenerational social class mobility in England and Wales and the impact of ethnicity, migration and religious affiliation

Transcription

1 Moving up? Intergenerational social class mobility in England and Wales and the impact of ethnicity, migration and religious affiliation Lucinda Platt, University of Essex Submitted for the Conference on Immigration: Impacts, integration and intergenerational issues to be held at UCL, March 2006 [word count = including abstract notes and references but excluding tables and figures] Contact Details Dr Lucinda Platt Department of Sociology University of Essex Wivenhoe Park Colchester Essex CO4 3SQ Tel: Fax: lplatt@essex.ac.uk 1

2 Moving up? Intergenerational social class mobility in England and Wales and the impact of ethnicity, migration and religious affiliation Lucinda Platt Abstract This paper offers a unique study of patterns of intergenerational social class mobility for children from different ethnic groups growing up in England and Wales in the 1960s to early 1980s. It explores the impact of ethnic and migration background on their outcomes as adults in It also investigates how examination of religious affiliation can modify or enhance our understanding of intergenerational processes. It stresses the advantages of using a truly longitudinal design to investigate these questions; but it also explores how such a design raises issues about those who are observed as children in 1971 or 1981 but not as adults in It concludes that patterns of intergenerational mobility vary by ethnic group, even when taking account of parental education and economic resources. For many minority groups, education provides the route to upward mobility from the working class at greater rates than for the white non-migrant majority. This is consistent with evidence from a range countries on the role of immigrants aspirations for their children and their commitment to education. However, education did not enable the Pakistanis and Bangladeshis in this study to reach a par with their white non-migrant counterparts in terms of occupational success. They remain severely disadvantaged in terms of social class outcomes, and this effect was enhanced rather than reduced after controlling for education. Exploring religious affiliation disrupted the picture of parity of intergenerational outcomes between the Indians and the white non-migrants. Instead it showed that there were sections of the Indian group who were achieving very high levels of upward mobility and those who were not doing so well. While these differences corresponded to differences in religious affiliation (Hindus and Christians doing well and Muslims and Sikhs doing less well), it is probably more helpful to think of them representing distinctive migration and settlement trajectories. Finally, analysis of selection effects for those whose outcomes were measured at 2001, showed that there was a correlation between remaining in the study up to 2001 and successful social class outcomes, but that the overall size of the effect was small. 1. Introduction and study design Britain has always been a country of immigration and some of its minority group populations have been here for an extremely long period. However, the current ethnic minority populations are largely the result of immigration in the post-war period from former colonies or commonwealth countries (Goulborne 1998; Mason 2000; Solomos 1989). The experience of these immigrant populations and their children and grandchildren has been a mixed one, and the different groups show great diversity in terms of income, employment, educational achievement and life chances (Mason 2003; Modood, Berthoud et al 1997; Platt 2002). This diversity is a consequence of factors associated with the processes of migration: the reason for migration; the different economic, social and human capital that the migrants brought with them; the stage of life and the point in time at which migration occurred and the opportunities available at that time. It is also shaped by the experience of the different groups 2

3 over time in Britain: the length of time spent in Britain; the characteristics of the areas of settlement and the possibilities for geographical mobility within Britain; the experience of racism, and responses to it; interaction with the education system and the labour market over time. Both migration histories and the experience of migrant groups and their subsequent generations in Britain have implications for patterns of intergenerational mobility: the extent to which the second generation does better (or worse) than the migrant generation and the ways in which advantage or particular motivations to succeed are transferred over time. In this paper, I attempt to shed light on the relative roles of factors associated with migration and those associated with living in England and Wales by examining the patterns of intergenerational mobility for two contiguous cohorts and the characteristics associated with particular outcomes. It has been argued that the migrant generation can be expected to differ from the succeeding generation, born and brought up in the country of immigration, in a number of (possibly conflicting) ways. Initially, lack of networks and familiarity with the host community can be expected to depress the occupational achievement of the migrants, relative to their skills and education. For example, Daniel (1968) drew attention to the extent of downward mobility experienced by immigrants following their entry into Britain. The lack of transparent congruence between qualifications and occupational class (Heath and Ridge 1983; Modood 1997b) is also taken as some evidence of downward mobility in the migrant generation. This depression of latent social class may be exacerbated if the migration was forced rather than voluntary. In this model, assimilation will lead to the second generation being much closer to their peers from the host community in educational and occupational terms (Park 1950 ; see also Alba and Nee 1997 and Gordon 1964); and in British accounts, we have seen a related argument for the recovery of the underlying class position of the first generation in the second generation (e.g. Modood 1997a). On the other hand, if migrant workers are being sought to fill, or have identified, particular occupational niches, these may be congruent with their skills and experience rather than representing downward mobility. For example, Smith (1977) argued that the concentrated employment of Pakistanis employed in the textile industry could be accounted for by prior experience. The role of geographical factors the influence of area or region of initial settlement of migrants has also been considered important in shaping or mediating outcomes for the second generation (Galster et al, 1999; see also Dorsett 1998). Such areas of settlement are themselves not independent of period of migration, opportunities for employment in different areas, and the match between those opportunities and the skills that the migrants bring with them. In so far as location of initial settlement and subsequent geographical mobility varies by migrant group, it is likely to result in systematic differences in the experiences of different ethnic minorities in the longer term. Moreover, migrants are acknowledged as a selected sample, who may have particular levels of motivation that caused them to migrate in the first place, but which may not be directly passed on to their children. This could mean that the second generation might be expected to fare less well than their parents in occupational terms, particularly if the children find that being born and bred in a country does not exempt them from racism and discrimination. The poor health status of certain minority groups in Britain, has, on occasion been discussed in these terms (see, for example, Marmot et al 1984; Nazroo 1997). On the other hand, there is substantial evidence that migrants place a great emphasis the education of their children, both expressing their own motivations through the next generation and gaining some compensation 3

4 for any downward mobility they have experienced through the success of the next generation (see, for example, the summary of these findings in Vallet, 2005). While British sociology is rich in studies of intergenerational social mobility generally (see, for example, Bottero and Prandy 2000; Goldthorpe et al 1987; Heath and Payne 2000; Marshall, Swift and Roberts 1997; Prandy 1998; Savage 2000), there are only a small number of studies ethnic differences in social mobility in Britain (Heath and Ridge 1983; Heath and McMahon 2005; Platt This is largely due to the difficulty of having sufficient sample sizes to measure differences between groups in intergenerational transitions. While some pooled samples have been used with studies containing retrospective questions on class background, it then becomes problematic to distinguish between occupational position in country of origin and occupational position in country of migration. Not only will these have different implications for the interpretation of results, but it is extremely difficult to maintain international equivalence of class. Attempts to understand groups relative positions in the course of the transition from immigrant groups to new ethnic minorities (Castles 2000) have also tended to shift with the way the first and second generations are defined, and according to which ethnic group is the principal focus of interest. On the other hand, translating insights into minority mobility processes to Britain from the studies that exist of immigrant social mobility in other countries is vexed by the particularity of their experiences of immigration, the institutional responses to migrants, the overall extent of fluidity (Erikson and Goldthorpe 1993; Vallet 2005) within the country, and the particular ways in which minorities are measured in different national sources. By using a genuinely longitudinal source, with sufficiently large samples from minority ethnic groups to allow analysis of individual groups, this paper makes a unique contribution to illuminating the patterns of social mobility within England and Wales and how these vary by ethnic group between the first and second generations. In doing this it builds on the author s previous work in this area (Platt 2005). By examining the role of resources within the migrant generation and these are translated into the outcomes of the second generation, it enables a consideration of what factors appear to be common across country studies and what is unique about the situation of some of Britain s minority ethnic groups in terms of their relative life chances. This paper aims to resolve some of the issues presented by the different migration histories of Britain s ethnic groups and the difficulty of the international comparison of class origins by considering a cohort who were all growing up in England and Wales at the same time. It measures the origins and destinations for this cohort at the same time points, and measures origins (parental class) as they are experienced within the English and Welsh labour market at that particular time. It observes, rather than asking for retrospective information on, parental social class. This removes the problems of recall error typically associated with intergenerational studies derived from cross-sectional samples; and such error is likely to be a greater source of bias for minority ethnic groups, where there may have been an abrupt occupational transition on migration. In addition, this paper excludes from analysis those of a minority ethnicity whose parents were born in the UK, and, similarly, distinguishes among those who claim a white ethnicity between those with both co-resident parents born in the UK (white non-migrants) and those with neither born in the UK (white migrants). It can thus more effectively test migrant effects. The data set which allows such a common framework to be established and observation of mobility across a thirty year period is the ONS Longitudinal Study. The ONS Longitudinal Study (LS) is a one per cent sample of the population of England and Wales that is followed 4

5 over time. It was initially obtained by taking a sample of the 1971 Census, based on those born on one of four birth dates (day and month). Information from samples taken at each subsequent Census has been added to the study. Members are also added to the study between Censuses by linking information on births and immigrations using the same selection criteria. Data on events that occur to sample members: births of children to them, infant deaths, deaths of spouses and cancer registrations are also added. No more information is linked where study members have records indicating that they have died or left England and Wales (unless emigrants re-enter at a later date, in which case they are re-incorporated into the study). For this study, two cohorts of children aged between 4 and 15 at the point they were observed in the LS have been selected from both the 1971 and the 1981 records in the LS. Their parents and household characteristics are measured at that point to give information about their origins. Parents social class was grouped from the CASMIN scheme into three categories service, intermediate and working, with those who did not fit one of these classes (e.g. housewives ) allocated to an other category. The hierarchical approach meant that in the case of parents with different social class locations the higher of the two was taken for the family. The children s own characteristics were measured in 2001, when they were adults aged between 24 and 45, to given information about their destinations. Their social class at this point (their destinations) was grouped from the NS-SeC into three classes (professional/ managerial, intermediate and manual/routine non-manual) corresponding to the parental social classes, with unemployment forming a fourth outcome. i Those who did not fit one of these classes left in a residual other category. The hierarchical approach was again used to allocate class to those in a partnership. Those for whom there was insufficient information to allocate class at all were excluded from analysis. Other variables that were included in the analyses were economic status and educational level of the study sample s parents, along with the ethnic minority concentration in the ward in which they were living, and educational level, marital status and religion of the study sample. The two cohorts of children were pooled to allow sufficient sample sizes for analysis, but age and cohort have been controlled in the various models. There are three elements to the paper. First it examines patterns of intergenerational mobility and the extent to which there are ethnic differences in chances of occupational class success, controlling for a range of background characteristics. I also examine whether there are differences in chances of the full range of class outcomes measured. Next the paper considers how looking at religious affiliation within ethnic groups can reveal differences within ethnic groups, differences associated with different migration trajectories as well as group practices. It thus challenges the coherence of standard ethnic group classifications. Finally, the paper considers the impact of selection issues on the main findings of the paper. Do those who survive in the sample up to the point of measurement of class outcomes differ from those who are lost to follow up? And do any differences have implications for the substantive findings. The paper concludes by illustrating the impact that ethnicity has on class outcomes by class background and argues that there is no single explanatory framework that can adequately represent the second generation s subsequent life chances for all migrant groups. 2. Understanding patterns of intergenerational social mobility Tables 1 to 5 show the parent to child mobility patterns for the five largest ethnic groups in the data: white non-migrants (the majority group), white migrants, Indians, Caribbeans and Pakistanis. These reveal that the tendency of privileged origins to preserve advantage in the 5

6 second generation applies more to some groups (including the majority) than others. They also reveal both the higher risks of unemployment for the minority groups and that unemployment appears to follow a class gradient in relation to origins for the white nonmigrants and the Indians but not for the Caribbeans and the Pakistanis. Again, there would seem to be differential opportunities for more privileged backgrounds to protect against occupational risks according to ethnic group. [Tables 1-5 about here] To examine the role of different factors in these patterns of intergenerational mobility, individuals chances of ending up in the professional or managerial class were calculated holding an increasing succession of characteristics constant. The results of these nested models exploring the effect of different factors on probabilities of class success can be found in Table 6. The effects of origin class and minority group concentration in ward of origin were examined, alongside sex, age, cohort and partnership status of the sample member, as the starting point for looking at the relationship between origins and destinations (Model 1). The power of social class origins to influence future generations class outcomes both independently and indirectly through its influence on educational achievement has been extensively demonstrated (Goldthorpe et al 1987; Erikson and Goldthorpe 1993; Goldthorpe 1997; Heath and Payne 2000). Here, class showed a clear impact on social class outcomes across the sample, with more privileged origins increasing the probability of professional or managerial class outcomes. Area effects are also potentially important in influencing outcomes in the second generation. There are two issues here. First, there is the extent to which certain minority groups may be concentrated in areas that suffer from processes of de-industrialisation and the consequent more limited employment opportunities. While it has been argued that ethnic minority group members have adapted relatively well to deindustrialisation (Iganski and Payne, 1999), it still remains the case that minority groups are still heavily concentrated in some of the most deprived areas with the consequent impacts on life chances that have been stressed in discussions of social exclusion. The second way in which area may be important is that concentrations of the same ethnic group have been argued to be potential resources. Ethnic group concentrations in particular areas, may, it is argued, enhance social capital, enable some pooling of resources and capital and aid enterprise. Here an attempt to capture area effects uses a variable summarising the level of minority group concentration in the ward of origin. The impact of this variable appeared to illustrate that, for the majority at least, it was living in a ward with no minority group concentration constituted a penalty in relation to chances of upward mobility. [Table 6 about here] Next, origin economic variables and parental qualifications were included (Model 2). Educational levels of parents have also been shown to be important in facilitating the upward mobility or higher class retention of their children. Parental educational qualifications can also stand in for some sort of latent class position which may have been disguised by downward occupational mobility for the migrants on migration. Parental education is therefore included alongside parental class and two indicators of economic status in the household of origin car ownership and housing tenure to investigate the ways in which 6

7 parental capital (human and economic) is related to their children s outcomes. As each parent s educational level may be independently important both are included separately. This means that this variable also includes the possibility that there is no co-resident father/mother to have an educational level. Thus, these variables also incorporate family structure. The information on qualifications for 1971 and 1981 is limited by the form of the question asked in those censuses. The parental educational variable derived from these can only distinguish higher qualifications levels, which were only held by a small minority of the population in this period (only 9 per cent of the sample mothers and 15 per cent of the sample fathers had such qualifications). There is no way of distinguishing between those parents with no qualifications at all and those with some non-advanced qualifications. ii Nevertheless, mother s and father s qualifications and the presence of a mother were shown to be significant in affecting outcomes, on top of social class effects. Perhaps surprisingly, the absence of a father compared to the presence of a father without advanced qualifications was shown to have a positive effect on the chances of professional or managerial class outcomes. This is presumably the result of the fact that an absent father s educational, social and economic assets cannot register in this model, but that they can still produce benefits for his children. It was after the impact of these variables had been evaluated that ethnic group was added to test if it had any effect over and above these origins (Model 3). This showed that all the minority groups except the Pakistanis (and the Bangladeshis but the result for that group is not statistically significant), have a higher probability of professional / managerial outcomes than can be explained by their origins (as imperfectly measured here) alone. This is congruent with the very high proportions who started off with working class origins, as shown in Tables 2-5. It would also support the idea that these groups may have experienced downward mobility on migration and that this was not the experience of the Pakistanis (Daniel, 1968; Smith, 1977; Modood 1997b). Nevertheless, it is interesting to find this effect even when other factors relating to origins (parental qualifications and economic status) have been accounted for. The different story for Pakistanis and Bangladeshis indicates that we cannot think in terms of a single immigrant story that impacts on intergenerational transmission across groups. Instead particular locations geographic and social that are only captured partially in this analysis must be playing a part, alongside unobserved characteristics. The extreme situation of Pakistanis and Bangladeshis has often been highlighted in research on current differences between minority ethnic groups in Britain (see for example Berthoud 1997, 1998). Equally, however, this particular level of disadvantage is often attributed to the sorts of background factors (class and educational level of parents) that are controlled for here. One unobserved factor may be the effect of different family sizes with evidence that numbers of siblings impacts on possibilities of social class success (Vallet, 2005). Another contribution to these negative outcomes for Pakistanis might be found by considering the way that class has been constructed as a family class. That is, for married or cohabiting couples it is the higher social class of either partner that determines the family class. The differential labour market participation of Pakistani women and their greater tendency to be in routine / manual occupations, would mean that for Pakistanis there was less opportunity for women to boost the family class. And a comparison (not illustrated) of own class with family class showed that this boosting effect of marriage did not apply for Pakistanis to the extent that it did for other groups. Finally, sample members own educational qualifications were added (Model 4). Unsurprisingly, increasing levels of education are strongly associated with higher probabilities of professional / managerial destinations, and the impact of origins is correspondingly reduced. We can see therefore how privileged and more educated groups gain advantage for the next generation through ensuring that their children gain educational 7

8 qualifications an aspect of class mobility that has long been noted (Glass 1954; Halsey et al 1980). But the origin effects, though reduced, do not disappear. Once education was held constant the positive ethnic group effects disappeared. That is, the greater chances of access to professional / managerial destinations than their parental class origins would imply, which was observed before controlling for education, would appear to be achieved through education. The argument that upward mobility is achieved through education for some groups is then supported by this finding for all groups except the Pakistanis and Bangladeshis. When educational qualifications are held constant there is a further reduction in the relative chances of occupational success for Pakistanis, and Bangladeshis show a similar disadvantage. Education serves to increase chances of professional or managerial class outcomes across the sample. But for these two groups, education is not able to compensate for whatever it is about or associated with Pakistani or Bangladeshi ethnicity that results in relative disadvantage. Lower levels of educational success are not able to explain lower chances of professional or managerial class outcomes for these two groups; and they are not achieving the levels of occupational success that not only their origins but also their educational achievements should imply. This is a startling finding, even given the attempts above to explain relative Pakistani disadvantage. In some studies, the operation of discrimination would be typically be deduced from such a finding. But the processes by which such differentiated discrimination occurs and the size of its impact would need to be understood before such a claim could be confidently made. This final model was repeated to examine the role of origins and own characteristics on all outcomes. A multinomial logit iii compared the relative chances of ending up in any one of the other possible class positions (intermediate, routine and manual, unemployed, or other) rather than in the professional or managerial classes (Table 7). This showed that the story was somewhat more complex than the binomial logistic regression had revealed. For example, controlling for all the factors in model 4 of Table 6, the multinomial regression shows the greater propensity Caribbeans to end up in the intermediate class, and, more particularly unemployment, compared to their white non-migrant counterparts. The relative risks of unemployment (as an alternative to professional or managerial class destinations) are 74 per cent higher for Caribbeans than they are for white non-migrants. This demonstrates the importance of considering unemployment as a potential outcome alongside class destinations when examining ethnic group differences in mobility. And it is important to remember that these increased risks of unemployment are over and above the effects of age, level of qualifications and so on. Unemployment is also a much greater risk for Bangladeshis: they are over 4 times as likely as white non-migrants to face unemployment compared with achieving service class outcomes, other things being equal. But the most striking results arise again in relation to the Pakistanis. Controlling for characteristics, they are more likely than their white non-migrant counterparts to fall into all other destinations compared to the chances of being in the professional / managerial classes: and they are over five times as likely as white non-migrants to be unemployed relative to being in the professional managerial classes. [Table 7 about here] 3. How can religious affiliation increase or complicate our understanding of intergenerational processes among immigrant groups? 8

9 It has been posited by some commentators religion may be a more telling differentiator between groups, whether because it better captures the specific histories and experiences of particular groups, or because it is a greater source, as some have argued, of group-targeted discrimination (see, for example, Modood 1997c). In this section, I examine the potential of religious affiliation to reveal heterogeneity across groups and to draw attention to the limited power of ethnic group categories in themselves to explain outcomes. To this end the differences in outcomes for those of Indian ethnicity according to their religious affiliation are explored. This has the potential to draw attention to what distinguishes the specific migration experiences of subgroups within particular ethnic categories. This itself may help us to understand some of the differences between the observed ethnic group categories that have been noted. The Indian group has been selected for this analysis as it has substantial proportions with different religious affiliations, as Figure 1 illustrates. [Figure 1 about here] Table 8 shows the probabilities of professional or managerial class success for the Indian group controlling for the same background factors and own educational level as employed in Tables 6 and 7 above, and examining the independent effect of religious affiliation. iv This model shows that both Sikhs and Muslims had significantly lower chances of achieving professional/ managerial class outcomes than the other religious groups. This effect of religious affiliation was compounded by lower probabilities of other characteristics associated with professional or managerial outcomes, so that predicted probabilities by religion demonstrated a clear distinction between Sikhs and Muslims on the one hand and Christians and Indians on the other. [Table 8 about here] That there are such clear distinctions within one ethnic group problematises the attempt to understand impacts of migration and experiences within the country of settlement in relation to particular minority ethnic groups. Within the Indian group there would appear to be sections who are following the path of swift upward migration associated with the motivation and commitment to education for their children of immigrants, and resulting in them outpacing the majority population in both educational and occupational success. On the other hand, this same group would appear to have sections that are faring less well in terms of intergenerational mobility and educational achievement than the majority. Factors are likely to be relationships with country of origin (those who are expelled, such as the East African Asians, are more likely to see their future and that of their children in the country of migration), routes of migration into particular occupations and locations, and differential experiences of (or responses to) discrimination. What it means to be Indian then become a further question, once it is disassociated from a particular pattern of life chances that can justify the retention of albeit clumsy ethnic group categories for analysis. These issues deserve further consideration, but the ability to explore them with the data available is limited. 4. Selection effects Up to this point, I have assumed that the results presented are not affected by differential attrition in the study sample. Of those whose origins were observed in 1971 and 1981, approximately 20 per cent were not observed in Some of these emigrated, but most of them were simply lost to follow up, and a very few died in this period. As Table 9 illustrates, the characteristics of these differed from those who were not lost from the sample 9

10 in this way. It therefore becomes worth considering what the possible impact of this differential loss to follow up was and whether it materially effects the conclusions that we are able to draw from the intergenerational analysis of those who continue to be observed up to This is an opportunity that the genuinely longitudinal data used here make possible. In the sociological literature on social mobility, missing data issues and difficulties with extrapolating from an observed sample from whom retrospective information is collected receive relatively scant attention. Given that for many of those not observed in 2001, ethnic group information (first collected in 1991) was not available, an ethnic group proxy, based on parents country of birth was constructed, and was validated against those for whom ethnic group was known. Similarly own education was not available, by definition, for those not observed in [Table 9 about here] A Heckman selection probit (Van de Ven and Van Pragg 1981) was then run to test for such selection effects, using the ethnic group proxy in the model for presence at 2001 and elective ethnic group for the chances of professional or managerial class. Results from this model (Table 10) indicated that unobserved characteristics associated with probability of being observed at 2001 were positively and significantly associated with unobserved characteristics associated with the probability of having a professional or managerial outcome at that point. [Table 10 about here] That is, the patterns of upward mobility that we observed in Tables 1-6 are relatively inflated, compared to those we would have observed had there been no attrition of the sample. The effects are small, however, and make no substantial difference to the relative position of different ethnic groups. This small scale impact is illustrated in Figure 2, which shows predicted probabilities of professional or managerial outcomes for certain ethnic groups and controlling for background characteristics based both on the selection model and on a simple probit. It shows the inflationary effect of not taking account of selection issues alongside the constant relative position of the different ethnic groups. [Figure 2 about here] 5. Conclusion This paper has considered intergenerational social class outcomes between migrant and second generations within Britain and for a comparable cohort of white non-migrants. It has given general support in the British context to the evidence for quantitative sociological studies in other countries of the important role of migrants motivation and commitment to education in explaining outcomes in the second generation. It shows that for the majority social class background both influences educational attainment and has an independent effect on successful class outcomes. For the minorities, class background would appear to play a less important role v but education is critical in evening out life chances and in some cases compensating for initial class position. However, there is substantial diversity between groups and this aggregate picture is not borne out for the Pakistanis and Bangladeshis. They are not achieving upward mobility on a par with other migrant groups or with the white majority. But this cannot be attributed simply to a lesser commitment to education since their 10

11 class disadvantage in the second generation is heightened once account is taken of their educational qualifications. Figure 3 illustrates this point by showing the predicted probabilities for a set of ethnic groups of professional managerial outcomes according to their own characteristics and from either the service or working class, but attributing to them either the Pakistani or the white non-migrant coefficient. Why education does not appear to work for these groups in line with the conventional model is, therefore, an important question for future research. [Figure 3 about here] The paper also highlighted diversity within groups. From the initial analysis, the Indians appeared to offer a clear example of a migrant group which, following initial downward mobility, equalised their occupational chances with the majority in the second generation through an emphasis on education. However, examining religious sub-groups within this single ethnic category showed that the picture was more complicated. Extremely successful sections of this ethnic group were balanced by less successful sections, so that the picture of comparable achievement with the majority was in fact not true for either category. Relative disadvantage for the Sikh and Muslim Indians, alongside the notable success of the Hindu and Christian Indians invites reconsideration of the extent to which standard accounts of intergenerational processes among immigrant communities can inform the diversity of experiences among Britain s minority ethnic groups. The challenge is to draw on the existing sociological literature to derive an account of the observed differentiated patterns of intergenerational mobility that allows for the extent of the variation within an understanding of common processes. Notes The permission of the Office for National Statistics to use the Longitudinal Study is gratefully acknowledged, as is the help provided by staff of the Centre for Longitudinal Study Information & User Support (CeLSIUS), in particular Julian Buxton. The author, however, retains responsibility for the interpretation of the data. i It is considered crucial to measure unemployment as an outcome given that Britain s ethnic minorities are widely shown to suffer excess unemployment, even when account is taken of other factors such as age and qualifications. ii This is aside from important issue of comparability of qualifications across minority groups, and the extent to which variation in the meaning or exchange value of qualifications obtained in different countries or different types of institution may influence observed patterns and relationships. iii An ordered logit was also tried but the comparison of goodness of fit indicated that it was not a sufficient improvement on the multinomial logit to justify being used in its stead. iv Religious affiliation was asked in the Census for England and Wales for the first time in Though the only non-compulsory question in the Census it was answered by the vast majority of respondents. v subsequent analysis of individual ethnic groups has confirmed this point. Class background is still relevant for the Indian group but not for the other minority groups. 11

12 References Alba, R. and Nee, V. (1997) Rethinking assimilation theory for a new era of immigration, International Migration Review 31(4): Borjas, G. (1992) Ethnic capital and intergenerational mobility, Quarterly Journal of Economics 107, Berthoud, R. (1998). The Incomes of Ethnic Minorities. Institute for Social and Economic Research, Occasional Paper. Colchester: University of Essex. Bottero, W. and Prandy, K. (2000) Social reproduction and mobility in Britain and Ireland in the nineteenth and early twentieth centuries, Sociology, 34: Castles, S. (2000) Ethnicity and globalization. London: Sage. Daniel, W.W. (1968( Racial Discrimination in England. Harmondsworth: Penguin. Dorsett, R. (1998) Ethnic minorities in the inner city. Bristol: The Policy Press Erikson, R and Goldthorpe, J.H. (1993) The Constant Flux: A Study of Class Mobility in Industrial Societies, Oxford: Clarendon. Galster, G.C., Metzger, K and Waite, R. (1999) Neighbourhood opportunity structures and immigrants socioeconomic advancement, Journal of Housing Research 10(1): Goldthorpe, J. (1997) Problems of meritocracy in Halsey et al (eds.) Education: Culture, Economy, Society, Oxford: Oxford University Press, pp Goldthorpe, J.H. with Llewellyn, C. and Payne, C. (1987) Social Mobility and Class Structure in Modern Britain. Second Edition, Oxford: Clarendon Press. Gordon, M. (1964) Assimilation in American life: the role of race, religion and national origins. Oxford: Oxford University Press. Goulbourne, H. (1998) Race Relations in Britain since 1945, Basingstoke: MacMillan. Heath, A. and Payne, C. (2000) Social mobility in A.H. Halsey with J. Webb (eds.) Twentieth Century British Social Trends. Basingstoke: Macmillan, pp Heath, A. and McMahon, D. (2005) Social mobility of ethnic minorities in G.C. Loury, T. Modood and S. Teles (eds.) Ethnicity, Social Mobility and Public Policy: Comparing the US and UK. Cambridge: Cambridge University Press: pp Heath, A. and Ridge, J. (1983) Social mobility of ethnic minorities, Journal of Biosocial Science Supplement 8: Iganski, P. and Payne, G. (1999) Socio-economic re-structuring and employment British Journal of Sociology, Marmot, MG., Adelstein, A.M. and Bulusu, L. (1984) Immigrant mortality in England and Wales : causes of death by country of birth. London: HMSO Marshall, G., Swift, A. and Roberts, S. (1997) Against the Odds? Social Class and Social Justice in Industrial Societies, Oxford: Clarendon Press. Mason, D. (2000) Race and Ethnicity in Modern Britain 2 nd Edition, Oxford: Oxford University Press. Mason, D. (2003) Explaining Ethnic Differences: Changing Patterns of Disadvantage in Britain, Bristol: The Policy Press. Modood, T. (1997a) Employment in T. Modood, R. Berthoud and others, Ethnic minorities in Britain: diversity and disadvantage. London: PSI, pp Modood, T. (1997b) Qualifications and English Language in T. Modood, R. Berthoud and others, Ethnic minorities in Britain: diversity and disadvantage. London: PSI, pp Modood, T. (1997c) Culture and Identity in T. Modood, R. Berthoud and others, Ethnic minorities in Britain: diversity and disadvantage. London: PSI, pp Modood, T., and Berthoud, R. and others (1997) Ethnic Minorities in Britain: Diversity and Disadvantage. London: Policy Studies Institute. Nazroo, J. (1997) Health and health services in T. Modood, R. Berthoud and others, Ethnic minorities in Britain: diversity and disadvantage. London: PSI, pp

13 Park, R.E. (1950) Race and culture. Glencoe, Ill: Free Press. Platt, L. (2002) Parallel Lives? Poverty among Ethnic Minority Groups in Britain London: CPAG. Platt, L. (2005) The intergenerational social mobility of minority ethnic groups, Sociology, Prandy, K. (1998) Class and continuity in the analysis of social reproduction, Sociological Review 46: Savage, M. (2000) Class Analysis and Social Transformation, Buckingham: Open University Press. Smith, D.J. (1977) Racial Disadvantage in Britain: The PEP Report. Harmondsworth: Penguin. Solomos, J. (1989) Race and Racism in Contemporary Britain, Basingstoke: Macmillan. Vallet, L-A (2005) What can we do to improve the education of children from disadvantaged backgrounds? Paper prepared for the ECSR Conference Comparative European Studies: Assessing ten years of sociological research November 2005 Institut d Études Politiques Paris. Van de Ven, W.P.M.M. and Van Pragg, B.M.S. (1981) The demand for deductibles in private health insurance: a probit model with sample selection, Journal of Econometrics 17:

14 Tables and Figures Table 1: Social class destinations 2001 according to social class origins (row percentages): white non-migrants Destinations (2001) Total (N) (column Managerial / Professional Intermediate Routine / Manual Unemployed %) Origins Service (29.3) (1971/ Intermediate (19.7) 1981) Working (51.0) Total (100 Table 2: Social class destinations 2001 according to social class origins (row percentages): Caribbeans Destinations (2001) Total (N) (column Managerial / Professional Intermediate Routine / Manual Unemployed %) Origins Service (12.8) (1971/ Intermediate (9.0) 1981) Working (78.2) Total (100) Table 3: Social class destinations 2001 according to social class origins (row percentages): Indians Destinations (2001) Total (N) (column Managerial / Professional Intermediate Routine / Manual Unemployed %) Origins Service (13.6) (1971/ Intermediate (13.3) 1981) Working (73.1) Total (100) Table 4: Social class destinations 2001 according to social class origins (row percentages): Pakistanis Destinations (2001) Total (N) (column Managerial / Professional Intermediate Routine / Manual Unemployed/ Other** %) Origins Service (7.2) (1971/ Intermediate (16.3) 1981) Working (76.5) Total (100) ** Note: as a result of small cell sizes, unemployed had to be combined with other for this table, which means this column and the overall row percentages are not directly comparable with the preceding ones. 14

15 Table 5: Social class destinations 2001 according to social class origins (row percentages): white migrants Destinations (2001) Total (N) (column Managerial / Professional Intermediate Routine / Manual Unemployed %) Origins Service (18.6) (1971/ Intermediate (18.9) 1981) Working (62.5) Total (100) Table 6: Logistic regressions of probability of professional/ managerial destination in 2001, controlling for individual and background characteristics Model 1 Coefficients (SE) Model 2 Coefficients (SE) Model 3 Coefficients (SE) Model 4 Coefficients (SE) Cohort (baseline is (.011) (.012) (.012) (.013) cohort) Age (base is ) Agegroup (.013).011 (.013).013 (.013) (.014) Agegroup (.014).041 (.014).041 (.014) (.015) Male.022 (.012).015 (.013).015 (.012).076 (.014) Partnered.986 (.014) (.014) (.014) (.016) Area concentration of minorities (baseline 0%) Up to 1%.213 (.020).197 (.021).192 (.021).191 (.023) 1 to 5%.330 (.023).326 (.024).301 (.024).349 (.026) 5-10%.210 (.033).241 (.034).189 (.034).268 (.037) More than 10%.221 (.032).240 (.032).154 (.035).266 (.038) Origin class: base is working Service class (.015).534 (.017).539 (.017).322 (.019) Intermediate.248 (.016).060 (.017).061 (.017).019 (.018) Other (.028) (.033) (.033) (.036) Mother s qualifications (base no qualifications No co-resident mother (.045) (.045) (.050) Mother with qualifications.432 (.025).420 (.025).115 (.027) Father s qualifications (base no qualifications No co-resident father.224 (.028).226 (.028).135 (.031) Father with qualifications.535 (.022).529 (.022).215 (.023) Tenure at origin (base is owner occupation) Local authority (.015) (.015) (.016) Private rented (.021) (.021) (.023) Car ownership at origin (baseline is no cars) 1 car.265 (.015).274 (.015).173 (.017) 2 or more cars.399 (.021).408 (.022).290 (.023) 15

16 Ethnic group (baseline is white non-migrant) Caribbean.197 (.068) (.073) Black African.557 (.219).050 (.232) Indian.445 (.062).078 (.062) Pakistani (.093) (.098) Bangladeshi (.223) (.219) Chinese and other.499 (.100).095 (.103) White migrant.257 (.041).070 (.044) Sample member s qualifications (base is 0) Lower 1.02 (.026) Middle 1.48 (.027) Further 2.78 (.028) Car ownership at destination (base is 0) 1 car 2 or more cars Tenure at destination (base is owner occupation) Local Authority Private Rented Constant (.025) (.029) (.030) (.038) N Chi2 Change (df) 1252 (6) 205 (8) 14004(4) Notes: Statistically significant results at a least the 0.05 level are highlighted in bold Standard errors are adjusted for repeat observations on persons The regression models were run both using dummies to represent missing cases and excluding all cases with missing values. For brevity the coefficients for the dummies are not given in this or subsequent Tables. Table 7: Multinomial logistic regression of destinations at 2001 controlling for individual and background variables Coefficient Coefficient Coefficient for Coefficient for for manual unemployment for other intermediate / routine (SE) (SE) class (SE) class (SE) Cohort (baseline is (.016).374 (.016).495 (.038).319 (.026) cohort) Age (baseline is 12-15) Agegroup (.018).279 (.018).308 (.040).194 (.028) Agegroup (.019).089 (.019).078 (.044).078 (.030) Male (.017).062 (.017).232 (.038) (.028) Partnered (.020) (.019) (.042) (.036) Area concentration of minorities (baseline is 0%) Up to 1% (.027) (.027) (.067) (.044) 1 to 5% (.032) (.032) (.077) (.051) 5-10% (.045) (.046) (.097) (.069) 16

17 More than 10% (.046) (.046) (.096) (.068) Class of origin (baseline is working class) Service class (.023) (.024) (.057) (.040) Intermediate.129 (.021) (.022) (.053) (.036) Other (.047) (.043).400 (.080).476 (.058) Mother s qualifications (base no qualifications No co-resident mother.113 (.062).061 (.060).342 (.107).271 (.083) Mother with qualifications (.033) (.038) (.086).124 (.057) Father s qualifications (base no qualifications No co-resident father (.038) (.037) (.076) (.054) Father with qualifications (.029) (.033).011 (.074) (.052) Tenure at origin (baseline is owner occupation) Local authority.098 (.020).377 (.020).492 (.044).482 (.031) Private rented.111 (.028).205 (.028) (.065) (.046) Car ownership at origin (base is 0) 1 car (.021) (.020) (.043) (.030) 2 or more cars (.028) (.030) (.069) (.047) Ethnic group (baseline is white non-migrant) Caribbean.185 (.088) (.094).556 (.137) (.117) Black African.090 (.276) (.299).0457 (.477) (.401) Indian.009 (.078) (.082).062 (.161) (.128) Pakistani.777 (.116).598 (.127) (.178) 1.545(.145) Bangladeshi.361 (.285).412 (.256) (.376).712 (.364) Chinese and other.024 (.129) (.148).295 (.211) (.193) White migrant.029 (.053) (.057).052 (.109).063 (.081) Sample member s qualifications (base is 0) Lower (.033) (.029) (.055) (.040) Middle (.033) (.030) (.059) (.043) Further (.035) (.034) (.063) (.047) Constant.671 (.047) 1.94 (.045).256 (.092) 2.198(.063) N Wald chi2 (df) (144) Notes: Statistically significant results at a least the 0.05 level are highlighted in bold Standard errors are adjusted for repeat observations on persons 17