The family structures of ethnic minority groups in Britain

Размещено на http://www.allbest.ru/

Abstract

The family structures of ethnic minority groups in Britain are compared with those of the white population. Caribbeans' low rates of marriage and high rates of single parenthood are characterised as adopting `modern individualism'. South Asians' high rates of marriage are characterised as adhering to `old-fashioned values'. Even among Asians, Pakistanis and Bangladeshis stand out in terms of low rates of economic activity among women, and large family sizes. All groups are moving in the same direction: Caribbeans ahead of the white trend, South Asians behind it. Some of the patterns observed are associated with child poverty.

1. Interpreting diversity

The living arrangements of Britain's ethnic minorities have been of interest to researchers ever since the first wave of West Indians arrived in the 1950s. Early consideration was primarily focussed on the direct effects of a recent migration. In some migrant groups, for example, there was an initial preponderance of unattached men. In some areas, restricted access to accommodation led many recent arrivals to crowd together in the same house or flat, and so create large `households'. But these characteristics were temporary and in a sense artificial consequences of the movement of populations, and have not been considered as primary issues in more recent analyses.

All four of the national surveys of ethnic minorities conducted by the Policy Studies Institute at about ten-year intervals have collected data about family and household structures, and each of the reports has presented summaries of the evidence (Daniel 1968, Smith 1976, Brown 1984, Modood, Berthoud and others 1997). There has been extensive analysis by demographers of other large-scale sources, especially the 2001 Census (eg Coleman and Salt 1996) and the Labour Force Survey (eg Berrington 1994). It seems fair to say, though, that most of this work has been demographic and descriptive, rather than sociological or evaluative. The national surveys provided analyses of household structure in the form of background pictures of the population under study, before the authors moved on to the `real issues' of employment, housing, race relations and so on. Variations in household structure have been used as part of the narrative of housing issues such as levels of over-crowding, or as explanatory variables in the analysis of employment or income. But few quantitative studies have attempted to explain (as opposed to describe) the variations in family structure between ethnic groups; the outcomes have been presented as `given'. Still less have family formation and/or dissolution been discussed in a way which helps us to understand the social and economic processes at work in a multi-cultural society. Nor have the implications of particular patterns of living been much discussed in terms of their potential consequences for the people concerned.

This paper is mainly based on analyses of family formation derived from the Fourth National Survey of Ethnic Minorities and the Labour Force Survey, which have already been published. A note at the end of the paper briefly describes five papers from which the empirical findings are drawn. To avoid repetitive bibliographical citation of the same papers, the references are given in the text as EMiB (Ethnic Minorities in Britain), YCM (Young Caribbean Men and the Labour Market), IEM (The Incomes of Ethnic Minorities), EMCG (Ethnic minority children and their grandparents) and TBEMW (Teenage births to ethnic minority women).

The paper will be useful as a summary of some of the key findings for readers who are new to these issues. But the facts have been organised here in such a way as to highlight some of the issues surrounding family formation which ought, perhaps, to be the subject of more debate within and around minority communities. The presentation is not, itself, evaluative (at least, I hope, not in a tendentious way). But it focuses on some of the key questions. Why do some ethnic groups adopt particular family structures? What influences will affect stability or change in these patterns? What are the implications for the sense of identity of the minorities themselves? Do people's decisions about partnering and parenting affect their or their children's chances in other spheres of life such as education or employment? I do not claim to answer many of these questions; but I think they should be asked.

It is appropriate to frame those questions within a broader discussion of normative social values in a multi-cultural society. One extreme position is to argue that the behaviour of migrant groups should be judged exclusively by the standards of the host society. That tends to lead to a pathological interpretation of minority family patterns as deviant, rather then merely diverse. The other extreme position is to argue that minorities' behaviour should be judged exclusively according to the conventions of their own societies, in their countries of origin. That overlooks the essential fact that the minorities' current social structures are no longer in their countries of origin, but are located in multicultural Britain. Some of these issues are well discussed by Song and Edwards (1996) and in the concluding chapter of Beishon, Modood and Virdee (1998).

It is not the aim of this paper to judge anyone, by any set of standards. It is to raise some questions about what is happening within and across community boundaries, in a Britain where young people of black and Asian origin share classes and lectures with white boys and girls who are also beginning to think about their approach to `the family'; and where white social norms are changing too.

The first thing to say about ethnic minorities is that they are a plural, not a collective. All non-white minorities may face a common experience of racism and harassment; but in most other spheres, it is the diversity between minority groups that is their most striking characteristic. In the crucial linked fields of education, employment and income, for example, Indians and Chinese are, if anything, slightly better off than the white population, while Caribbeans, Bangladeshis and Pakistanis are to various degrees worse off than average. Nowhere is this diversity more apparent than in family structures.

A theme of this paper is that Caribbean family patterns are more `modern' than those of whites, while South Asians adopt more `old-fashioned' practices. The words modern and old-fashioned may seem loaded, though each is ambiguous. The construct is not intended to imply an idea of `progress' from out-of-date to up-to-date patterns; nor does it imply a decline from traditional to untested values. The objective fact is that the two main groups are on either side of the white majority, and cannot be summarised together.

2. Diversity among white families

A key point is that diversity of family patterns is not a feature specifically of ethnic minorities. It is worth diverting from the focus on minority groups, for a page or two, to consider parallel changes affecting white communities in Britain and in Europe.

There has been a huge change in routes to family formation within Britain over the past generation. Young people nowadays face a set of choices which itself can be interpreted as a form of diversity. And there are wide differences between the conventional patterns accepted in neighbouring countries within Europe. These diversities - over time, between countries and among minorities - all help to show that there is no absolute ideal of family structure which needs to be defended against all rivals. The family is an organism whose successful evolution may be promoted by the introduction of diverse strains.

A generation or two ago, there was a standard trajectory of family formation in Britain whereby young people completed their education, got married and had children - and stayed that way until their own children left home. This was never an obligatory or universal path, but it was recognisably the standard thing to do. Chart A shows young women's family positions at each year of age in 1973. Almost all were single and living with their parents at the age of 16. The great majority were married, and had children, by the time they were 35. In those days, the route from family of origin to family or procreation was rather short. Very few women in their twenties were in an intermediate position, no longer with their parents, not (yet) with their husband-and-children.

Only a generation later, by 1996, the pattern had changed dramatically (Chart B). Young women were staying at home just as long. They were reaching the stage of `married with children' much later (if at all). A whole range of alternative or intermediate situations was now available - single but living away from their parents; one-parent families; cohabiting; or married but without children. A single statistic sums up this change: in 1973, more than two thirds (69 per cent) of women in their late twenties were (legally) married and had children. By 1996, the proportion had fallen to less than a third (31 per cent).

Indeed, the label `intermediate' for those who lived neither with their parents, nor with their husband-and-children, is no longer appropriate. An increasing proportion of women will remain in one of the alternative positions without ever moving to the destination that would once have been their primary expectation.

This increasing diversity of family form for men and women in their 20s and 30s suggests a much wider range of options than was available and socially acceptable in the past. For many, it will have offered a greater degree of flexibility and perhaps choice, though not everyone would necessarily have chosen the positions they found themselves in. It has also led to a substantial degree of economic polarisation, as the gap has widened between DINKYs at one extreme and non-working lone-parents at the other.

There is diversity, too, between countries in Europe (Chart C). Nearly half of women in their late 20s continue to live with their parents in Italy and Spain; hardly any do so in Denmark and the Netherlands. Among those who have left home, the great majority are married and with children in countries such as Portugal, but `intermediate' family forms predominate in Denmark and the Netherlands. A clear regional pattern has been identified (Iacovou 1998). `Southern and/or catholic' countries - Italy, Spain, Greece, Portugal and Ireland - have `traditional' patterns whereby young men and women live with their parents until they get married; and they have children soon after that. In `northern/protestant' countries - Denmark, the Netherlands, the UK, France, Luxembourg and Belgium - few people live with their parents, and many adopt intermediate living arrangements before (or perhaps instead of) getting married and starting a family.

These comparisons over time and between countries are offered simply to set the diversity between British ethnic groups in context. Family patterns are not fixed in time or across space. Every generation in every society has its own ideas about what young people ought to do, and wags its beard if individuals deviate too far from that expectation. The fact that norms vary does not, though, mean that they are irrelevant. They are changeable - perhaps more restrictive in some societies, more liberal in others. They are changing now in Britain. They are changing, and will no doubt change again, within the minority communities. We need to understand the directions and the implications of these changes, rather than idealise current patterns (on the one hand), or bemoan them (on the other).

The key feature of family life in the Caribbean community is the very low rate of marriage. This true across all age-groups, but can be seen most clearly by focusing on the late 20s (Table 1). Two thirds of white men and women in that age group had lived with a partner; little more than one third of Caribbeans had done so. Among those with a partner, three-quarters of whites, but only half of Caribbeans, were in a formal marriage. Among those who had married, the proportion who had separated or divorced was twice as high for Caribbeans as it was for whites (across all age-groups under 60).

As a result of these three factors - low partnership rates, low marriage rates and high separation rates - only 39 per cent of Caribbean adults under the age of sixty are in a formal marriage, compared with 60 per cent of white adults under sixty. Actually, most Caribbeans in the older half of that age range are married, or at least have been; it is among younger people that marriage rates are so low. That is partly an ageing effect (young people will eventually get married) but also partly a cohort effect (today's young people are less likely to marry than their predecessors were).

Of course most of those who migrated to this country as adults were married to (or lived with) a man or woman of their own ethnic group. But among British-born Caribbeans, half of men with a partner live with a white woman; a third of women with a partner live with a white man. The rate of mixed partnership is increasing rapidly (for men, but not so rapidly for women), and it is probable that the majority of Caribbean men starting a live - in relationship (or marrying) nowadays, have chosen a white woman (YCM, Table 22).

These mixed partnerships are much more common than is found, for example, among African Americans, and they have attracted some interest. Whereas worry about `miscegenation' has always been a central issue for white racists, many liberal commentators see cross-cultural relationships as a welcome sign of increasing mutual acceptance between the white and the black communities. Mixed marriages are widely accepted among Caribbeans - more than among whites, and much more than among Asians (EMiB Table 9.18 and 9.20). For some black people, though, there remains a suspicion that marrying into a white family is disloyal to one's ethnicity of origin, an opportunist move into white society (Modood and others 1994, Beishon and others 1998). In so far as objective measures can contribute to such issues, the indications are that the high rate of mixed marriage among Caribbean men is not associated with their social standing, in either direction: mixed relationships are about equally common among those with high and low levels of education, and among those with good and bad employment experiences (YCM, Table 22 and page 51).

From the point of view of our immediate interest in family formation, though, the number of mixed partnerships among Caribbeans, combined with the low rate of partnership and marriage in the first place, means that very few Caribbean men and women are married to each other. This has obvious implications, which will be discussed later, for the future structure of the Caribbean community. But another point of interpretation is that this helps to emphasise the relative unimportance of marriage to the Caribbean self-image (in contrast, as we will see, to the South Asian self-image). It is not in marriage, but in other activities, that one asserts membership of the Caribbean community.

Because such a small proportion of Caribbean men and women are living with each other in nuclear families, more detailed analysis of family structures has to start with women. Most surveys do not ask respondents to record all the children of which they are the father or mother, but only the children living with them at the time. It will be seen that a large proportion of Caribbean children do not live with their father. All we can do is describe the families of the women, and make inferences from that about the probable position of the men.

For the same reason, family formation can no longer be approached in the traditional order. It used to be accurate to ask, first, how many women were married, and second, which of the married women had children. We will ask these questions the other way round: who had children, and which of them were married (or had a partner)?

One third of Caribbean women in their early twenties have children; about half of those in their late twenties; three-quarters of those in their early thirties (YCM Table 13). These figures suggest that Caribbean women are slightly earlier into the child-rearing phase than white women, but the two ethnic groups are broadly similar in their fertility rates.

Again, it was on the question of partnership that the two groups diverged most widely. One in ten white women with children (and under the age of 35) were single (never - married) mothers (YCM, Table 13). No less than half of Caribbean mothers were single parents on this definition.

The patterns of single parenthood within each ethnic group are not easy to analyse, because all of the potential influences are inter-related. A multivariate logistic regression equation has been used to estimate the probability of a mother remaining single, depending on the age at which she first had children, the age of her oldest child, the year she passed her 20th birthday, whether she was a migrant or a non-migrant, and what her educational qualifications were. The analysis can be used to show what happens if each of these characteristics is varied one at a time, holding the other four constant (Table 2). For both whites and Caribbeans, women who had been young when their first child was born were much more likely to be single mothers at the time of their interview - in both groups, mothers who had started a family at 20 were about twice as likely to be in that state as those who had waited another 10 years. But Caribbean women retained six times as high a risk as white women, whatever their age of first birth.

The age of the oldest child can be seen as the number of years since the woman became a mother. Both groups of women were less likely to be single, the longer the period of their motherhood. This can be interpreted in terms of about half of white single mothers, and about one-third of Caribbean mothers, marrying (or forming a live-in partnership) later. That is, single motherhood is not necessarily a temporary stage prior to marriage; many Caribbean women have children single, and stay single.

The estimates in the first two panels of the table are all based on a notional cohort of women whose 20th birthday occurred in 1985. It is well known that there has been a substantial increase in the number of single parents over the years (Haskey 1998), and we would expect such a change in social expectations to be reflected in a cohort effect. The third panel confirms this: women who reached the potential child-bearing phase of their lives in the most recent period were much more likely to have been a single parent than their predecessors. In absolute terms this trend seemed strongest for the Caribbean women: the increase from 27 per cent to 59 per cent over 15 years represents one third of all Caribbean women with children. But in relative terms, the increase was even stronger for whites: the number of white single mothers multiplied more than 10-fold over 15 years.

Much has been written about single mothers of Caribbean origin (Dench 1996, Song and Edwards 1996). The practice of living independently of the children's father can be traced to West Indian social and economic traditions - themselves, perhaps, a hangover from the days of slavery when husbands and wives might be sold to separate plantations. What is striking about these findings, though, is that behaviour said to be derived from Caribbean cultural values is becoming more, not less, common over the years in Britain. The same conclusion is supported by another result from the logistic regression equation: after controlling for the age and cohort effects shown in Table 2, the proportion of Caribbean mothers who were single was 48 per cent among those who had been born in Britain or had arrived as children; it fell to only 24 per cent among women who had come to Britain at the age of 16 or later. Thus it is the second generation which has adopted the `West Indian' tradition, rather than the migrants themselves.

Research among white women has tended to show that those with lower levels of educational qualifications tend to have children younger than their better-educated counterparts, and that women who have children at an early age are often single mothers (Rowlingson and McKay 1998). So Table 3 distinguishes the two components of single motherhood - whether women had children, and whether those with children were single. Less than a third of women with degrees had children (at the `standard' age of 30); four fifths of women with no qualifications were mothers. The strong and consistent effect of qualifications on the proportion of women who had children was almost identical for both whites and Caribbeans. In both groups there was a parallel tendency for less-qualified women with children to have remained single. The relative effect was stronger among white women (a ratio of 3 to 1 between unqualified women and graduates); but education was still an important source of variation in the much higher rate of single parenthood among Caribbeans. Putting the two sets of findings together, the analysis suggests that nearly half of all unqualified (30-year-old) Caribbean women were single mothers, compared with just 2 per cent of white women with degrees.

What does the high rate of single parenthood among Caribbean women imply for the family positions of Caribbean men? The fact that a very large proportion of young Caribbean women with relatively low educational qualifications are single parents suggests that young Caribbean men with relatively low qualifications are unlikely to be living in the standard partner-plus-children family relationship. Existing survey data shows only who lives with each person in the same household. It can be assumed that many Caribbean men are fathers in the genetic sense of the word. There is qualitative evidence that many Caribbean couples have retained the `visiting partner' relationships that are common in the West Indies (Beishon and others 1998), and that this often involves the man in financial and other paternal responsibilities. There are, though, no quantitative estimates of the proportion of non-resident fathers who are involved in family life without living with their partner and children.

Nearly two-thirds (63 per cent) of young adult Caribbean men (aged 22 to 35) are `unattached' - that is, they have neither a live-in partner nor live-in children. The equivalent figure for white men is less than half (47 per cent). As before, it is difficult to sort out the influences of a range of potentially overlapping influences on men's family status, and a logistic regression equation has been used to measure the effect of each, independently of all the others. The first panel of Table 4 shows the expected tendency for the proportion remaining unattached to fall as young men grew older, and illustrates the clear divergence between white men and Caribbean men by age.

The comparison between the cohorts of young men who entered adulthood at various periods over the past two decades is more important. For both white and Caribbean men, the proportion of men estimated to remain unattached at the age of 30 doubled in 15 years. There appears to have been what amounts to a transformation in young men's family relationships. Most young men in the 1975 cohort lived with a partner and/or children by the age of 30. By the 1990 cohort, most white and black young men remained single. Caribbeans were running ahead of whites throughout the period, but the trend was equally strong for both groups.

Among Caribbeans, men who had been born in Britain (or migrated as children) were more likely to remain unattached (60 per cent) than those who had arrived age 16 or more (43 per cent).

The high proportion of young black men who remain unattached may be related to their poor employment prospects. This hypothesis, which has been very influential in the United States (Wilson 1987), suggests that black women are reluctant to marry (or live with) a man whose chances of getting or keeping a good job make him an unreliable source of income for themselves and their children. Better to remain single (and claim social assistance if necessary). Young men's family position has therefore been analysed by some of the factors which might make a difference to their employability. Comparison across educational qualifications is complicated by the fact that better qualified people tend to form partnerships later in their 20s, simply because they spend longer in the education system. So Table 5 shows how long it took men to get married or cohabit, in terms of the duration of the period since they completed their studies. For both white men and Caribbean men, those with lower than average levels of qualification took longer than others to form partnerships. The relationship was strongest for Caribbeans - in the extreme, a Caribbean man with no qualifications would expect to live without a partner for 20 years after leaving school.

It was also found that Caribbean men living in regions with high levels of unemployment had lower than average partnering rates (LFS, new analysis). The relationship was much weaker for white men. This provides further evidence of a link between Caribbean men's job prospects and their marriage prospects.

The two most striking aspects of Caribbean family formation are the large number of men and women who live without a partner; and the high proportion of those, with a partner, whose partner is white. Both findings are summarised in Table 6, for families with children. Only a quarter of `Caribbean' children live with two black parents.

Most commentators seem to welcome the increase in the number of mixed relationships, as an indicator of reducing cultural, social and economic barriers between ethnic groups. Perhaps it is a sign of declining racism within the white community. The fact that some black people are anxious about the trend nevertheless needs to be recognised. There is little evidence that men and women in mixed marriages are any better off or worse off than those who have chosen partners within their own ethnic group. There are nevertheless some consequences of mixed relationships which need to be taken into account in the analysis of multi-cultural Britain.

The first, and obvious one, is that a significant proportion of the children of Caribbeans will be of mixed origin. It is not yet possible to say exactly how many, because we do not know how many Caribbean single mothers had had relationships with white men; nor how many white single mothers had had relationships with Caribbean men. Whatever the precise figures, mixed parentage means an increase in the number of members of the next generation who are in some way Caribbean, and a reduction in the number who are unambiguously Caribbean.

None of the quantitative surveys so far available provides accurate information about people of mixed origin, though the 2001 Census will provide a much more effective classification. In the meantime, it is rather assumed that people of mixed origin perceive themselves, and are perceived by others, to be black. It seems likely that they have a more complex identity than that, reluctant to throw over one half of their heritage. This is a major issue for further research.

From the point of view of Caribbeans as a community, though, the trend may lead either to the decline, or the increasing isolation, of blackness as an independent identity. Whereas early commentators were eager for minorities to be assimilated into the majority culture, the essence of multiculturalism is that Britain should learn to value the diversity of communities making their various contributions to our national life. In that context, a long term reduction in the number of people firmly rooted in the Caribbean tradition might be seen to impoverish us all.

The other main feature of Caribbean family formation is the number of single mothers, on the one hand, and of unattached men, on the other. It was emphasised at the beginning of this paper that the proportion of young people in `intermediate' family forms is much higher in `northern/protestant' European countries than in `southern/Catholic' ones, and has risen rapidly in Britain over the last quarter of a century. Caribbeans could therefore be considered to be very `northern/protestant'; or very modern. In this light, they are ahead of the trend (while South Asians are behind the trend). But the scarcity of coresident partnerships during the child-rearing years is so far ahead of the national trend as to represent a real difference of social convention, rather than the kind of variation between groups which might be expected in any society. While the different convention is often explained in terms of the matriarchal family structures common in the West Indies, it is important to take account of the fact that Caribbeans are not converging on the European norm - the evidence clearly points to an increase in the number of unpartnered parents from generation to generation and from year to year.

Qualitative evidence suggests that Caribbeans' own views of these trends may be ambivalent. My former colleagues Beishon, Modood and Virdee (1998) found three potentially conflicting points of view.

• `Some Caribbeans… emphasised an individualism - the importance of individual choice, the value of commitments generated by the quality of relationships rather than custom, duty or marriage certificate, and independence… Marriage has come to be regarded as just a lifestyle option, which one may or not wish to choose, depending on individual circumstances'. Thus there is a theoretical and principled preference for the less committing family forms that can be labelled modern individualism.

• On the other hand, Caribbeans said that `their own communities were more family - orientated than the white British, and instilled more discipline and respect for age than their white contemporaries. Indeed, there was unity among the minorities in their criticism of what they perceived as a lack of commitment to parenting amongst whites, and they talked of white children as not having respect for their elders, and as being out of control.' This apparently older-fashioned view is not necessarily inconsistent with a preference for lone-parenthood, since it might be argued that a mother (and a grandmother) could exert a firmer discipline on children than a mother and father together. `Fathers were in practice dispensable; there was no particular role or responsibility that fathers performed which could not be performed in their absence by mothers.'

• Nevertheless `it was noticeable that the ideal of marriage and joint parenting between resident fathers and mothers still exerted a considerable appeal among the Caribbeans… For the majority… marriage is an ideal which unfortunately only some achieve - mainly because, in the view of women, men lack commitment.' On this latter view, single motherhood may be accepted by Caribbean women, but it is the men who should bear the responsibility for their position.

There is a large literature on the question of whether children brought up by one parent experience worse outcomes than those with two (Kiernan, Land and Lewis, 1998). Having only one resident parent may be a disadvantage, but the primary mechanism may be the poverty experienced by most such families, rather than the lack of a father as such. Whatever the overall conclusion, little of the research can be applied directly to Caribbean families, where single parenthood is now a conventional rather than a deviant family form and where some non-resident fathers may have more contact with their children than is common in the white community.

There is little doubt, though, about the financial position of Caribbean one parent families. It is often remarked that Caribbean lone parents are more likely to support themselves through work than their white equivalents. The Fourth National Survey reported that only 27 per cent of white lone parents living in an independent household had any employment; the figure for Caribbeans was 42 per cent (EMiB, page 152). The Family Resources Survey, analysing lone parents regardless of their household position, recorded 37 per cent of white lone parents and 43 per cent of Caribbean lone parents to have any earnings (IEM, page 11). Caribbeans may be ahead of whites on this measure, but still, more than half of Caribbean lone parents have no job. There is no evidence to suggest that many of them receive the substantial and regular maintenance payments which would be needed to float them and their children above the poverty line. As Table 7 shows, the majority of one-parent families - whether white or Caribbean - have to

claim Income Support. It is difficult to avoid the conclusion that it is the number of lone parents in the Caribbean community, rather than high levels of unemployment, which confines more than a third of all families with children to the safety net. This may be regarded as unfair on the social security system (which was never designed to shoulder the primary long-term responsibility for large numbers of families). And it is unfair on the mothers and children, who are obliged to accept the poverty and the sense of dependence with which Income Support is inevitably associated.

This analysis may conjure up a picture of prosperous Caribbean men selfishly keeping their earnings to themselves. It is true that an exceptionally high proportion of young men of Caribbean origin are unattached. On the other hand, the analysis in Table 5 suggested that many of these bachelors have limited educational qualifications, and poor prospects in the labour market. Moreover, marital status itself has an independent effect on men's employment rates: for both whites and Caribbeans, a young man living on his own is about twice as likely to be unemployed as a young man who lives with a partner, holding other characteristics constant (YCM page 50). So the low rate of partnership in the Caribbean community may create poverty among men as well as among women and children.

As many as 17 per cent of Caribbean men in their 20s have virtually no qualifications, no family and no job - and are not students either (YCM page 79). This is nearly three times as many as among young white men. Thus there is a relatively large group of young Caribbean men with so few bonds or formal commitments that they must be at risk of alienation from conventional standards. That might lead in turn to social conflict, resentment and trouble with the law (Smith 1991). On this interpretation, the loosening of family ties within the Caribbean community may turn out to be a problem for them, and a problem for race relations.

3. South Asian families in Britain: `old-fashioned family values?'

All four of the South Asian groups identified in the Fourth National Survey had features of family life which were similar to each other, different from whites and even more different from Caribbeans. Depending on the measure used, there were some variations between the four South Asian groups, with Bangladeshis and Pakistanis showing the most characteristic Asian pattern while Indians, and especially African Asians, sometimes appearing closer to the white position.

In contrast to Caribbeans, the key feature of family life in South Asian communities is the very high rate of marriage. Around three-quarters of Pakistanis and Bangladeshi women are in partnerships by the age of 25, compared with about two-thirds of Indian women and just over half of African Asian and white women (Table 8). South Asian men tend to be rather older than their wives, so the differences between ethnic groups would be smaller if the analysis focused on men. But the important point is that virtually all South Asians with a partner are in a formal marriage. And the proportion who have separated or divorced is less than half that recorded among whites.

So for South Asians, the key questions are not whether they are married but how they married and who they chose. It is well-known that young men and women in India, Pakistan and Bangladesh commonly have their marriage partners chosen for them by their parents or other family elders, and that this practice continues, at least in part, among the communities that have migrated to Britain. Qualitative research (Stopes-Roe and Cochrane 1990, Anwar 1994) suggests that the attitudes of young South Asians on `arranged' marriages have been moving away from those of their parents, though this seemed to be much less true among Muslims than among Sikhs and Hindus. It is not necessarily a clean split between those who think it `my' decision and those who think it `their' decision, since an amicable negotiation might take account of both points of view. The FNS showed that the majority of South Asians who came to Britain at the age of 25 or more (most of whom are assumed to have been married before they came), reported that their parents had decided on their partner (Table 9). This was especially true among women; it was more true among Muslims and Sikhs than among Hindus. For South Asians who were born here or arrived as young children (who must have married some years after they arrived) less than half of Muslims and Sikhs, and only a very small proportion of Hindus, felt that their partner had been chosen by the family, rather than by themselves. Interestingly, there was no apparent difference between men and women in the second generation.

Opinion questions in the Fourth National Survey suggested that South Asians were a good deal less likely to accept mixed marriages than Caribbeans were - indeed, less likely than white people. Well over half of Indians and Pakistanis felt that most members of their group would mind if a close relative were to marry a white person. It was not uncommon for respondents to claim that they themselves would not mind (even though most members of their group would). Both of the primary sources show that mixed marriages are much less common among South Asians than among Caribbeans. About one in five British born men of Indian or African Asian origin have a white wife; the equivalent figure for women is about one in ten. Very few Bangladeshis and Pakistanis have entered mixed relationships. It stands to reason that in communities where arranged marriages are common, mixed marriages are rare; in fact new analysis of the FNS shows that very few Indian Muslims or Sikhs had a white partner - most of the mixed marriages involved Hindus or the small number of Indian Christians.

Asians may also be `old-fashioned' in the nature of the relationships adopted between men and women within marriage. As Table 10 shows, Indian women are rather more likely to be a full-time homemaker than white women; but a clear majority of Bangladeshi and Pakistani women report their primary activity to be looking after the house and family. These are the sort of proportions that would have been observed among white women in the 1950s and 1960s. The fact (see below) that Bangladeshi and Pakistani women have significantly more children than members of other groups provides a partial explanation for their low rates of economic activity, but differences remain after taking family structure into account (EMiB page 87).

There are some signs, though, that the situation may be changing. Muslim women may be allowed to take professional level jobs, even though they are still discouraged from accepting low-prestige occupations. Bangladeshi and Pakistani girls are holding their own in the British education system (YCM, Table 5). The Labour Force Survey shows that qualifications can make a big difference. Among married Bangladeshi and Pakistani women born or brought up in Britain, the following proportions are full-time homemakers:

• 74 per cent of those with no qualifications;

• 52 per cent of those with qualifications up to GCSE level;

• 22 per cent of those with qualifications above GCSE level.

The strong implication of these findings is that more Muslim women will find their way into the labour market as more of them obtain educational qualifications.

Beishon, Modood and Virdee (1998) report that South Asian men retain family authority over their wives, even in households where the wife has a job. Very little data is available with which to measure variation between communities in the quality of relationships within families. A question in the FNS asked whether it would be the husband or the wife who had the final say in big financial decisions. Few families in any ethnic group said that the wife was the prime decision maker - the issue was whether it was husbands on their own or couples as a partnership who decided these things. Indian and African Asian couples were rather more patriarchal, on this measure, than white couples. Bangladeshis and Pakistanis were substantially more likely to assign responsibility to the husband, though even in these Muslim communities, power-sharing was the most commonly reported arrangement. More worrying, perhaps, for Bangladeshi and Pakistani men's sense of authority, their wives were much less likely to attribute power to them, than they were to claim it (EMiB page 179).

Three-quarters (76 per cent) of white women have had children by the time they are in their early 30s (YCM, Table 13). The figures are rather higher for Indian women (83 per cent) and very high indeed for Bangladeshi and Pakistani women (92 per cent). In fact Indian women's fertility is very similar to white women's, right across the relevant age-ranges (Chart D). The birth rate is substantially higher in the Pakistani community, and higher again among Bangladeshis, whose fertility is about double that of white and Indian women's at all stages. A striking point is the number of births to Bangladeshi teenagers - but, unlike white teenage mothers, the Bangladeshi women were already married. Another striking point is the size of Bangladeshi and Pakistani families. Large families were a significant area of British social policy interest until the 1970s (Land 1969), but are so rare now that they have virtually disappeared from the research agenda. They are not rare in Asian communities though: more than half of Bangladeshi and Pakistani women in their late thirties have four or more children (EMiB, page 41).

There are some signs, though, that these very high levels of fertility have also been reducing over recent decades. For example the annual rate of teenage motherhood among Bangladeshi women is estimated to have fallen from 61 per thousand in the mid 1980s to 38 per thousand in the mid-1990s (TBEMW). Table 11 compares successive cohorts of women age for age and suggests a decline in family sizes. Thus there are some indications of convergence on white fertility rates.

Another well-known feature of Asian family life is the number of three-generation households, where couples continue to live with their parents after they have started a family of their own. Detailed analysis both confirms and contradicts this stereotype (EMCG). Take Indian families as an example, and focus on the paternal grandmother. (She is the most likely candidate to live with the family, because young Asian families live more often with the young man's than the young woman's family; and because she is likely to outlive her husband, the paternal grandfather.)

• 70 per cent of Indian paternal grandmothers are still alive.

• 51 per cent of the surviving paternal grandmothers live in Britain.

• 67 per cent of British-resident Indian elders live with one of their adult children. This is a very high proportion, compared with only 15 per cent of white elders.

• 36 per cent of Indian families, whose paternal grandmother lives in Britain, include her in their household. This is not much more than half of the number of elders living with their children, presumably because some grandparents were living with other children, the siblings of the family under consideration. This 36 per cent is also a very high figure: only 1 per cent of white families live with their paternal grandmother.

• So, because a large number of grandmothers are dead, or in India, or living with another sibling, the proportion of all Indian families who live with their paternal grandmother is only 13 per cent.

The outcome is that Indian families have a very strong tendency to live with the father's parents if those parents are available. But in most cases they are not available, so it is quite wrong to imply that most families have paternal grandparents living with them.

These figures all refer to Indian families. The four South Asian groups are compared with each other, and with white families, in Chart E. It is immediately clear that most white grandmothers are alive, living in Britain, but not living with their children. Among South Asians:

• A large proportion of Bangladeshi grandmothers have died. This is probably not caused so much by high mortality in Bangladesh as by the fact that Bangladeshi men do not have children till relatively late in life. Their mothers are/were relatively old.

• Many Pakistani grandparents are alive, but still in Pakistan. On the other hand, most African Asian grandparents are in Britain; probably because the whole family came to Britain together from Kenya or Uganda.

• The four Asian groups have similarly high rates of co-residence among those where a grandmother is available in this country.

• But because many grandparents are dead, or living abroad, the overall proportion of children living with their paternal grandmother is reduced, to 17 per cent for African Asians, 13 per cent for Indians, 9 per cent for Pakistanis and only 4 per cent for Bangladeshis.

It is difficult to predict which way these figures will move over time. It is possible that a preference for three-generation families may reduce over the years, as Asian family patterns are increasingly influenced by white norms. On the other hand, the supply of grandparents available in Britain is bound to rise, as the current generation of children start to have families of their own. It is therefore possible that the number of families living with grandparents will hold up or even increase.

To sum up, all four of the South Asian communities in Britain have some elements in common, and distinct from the current white position. The key characteristic is the continued importance of marriage. All four groups provide evidence of distinctive South Asian forms, such as arranged marriages, a reluctance to marry out, and a preference for living with paternal grandparents. In some respects, Bangladeshis and Pakistanis are distinctive again, especially in the number of women spending their whole time looking after the house and family, and in the number of children they have.

The Asian patterns can clearly be labelled `old-fashioned' in the sense that many of their characteristics now could have been observed among white families in the past. They are `old-fashioned', too, in the sense that loyalty to their own communities' histories and traditions is one of the driving forces behind the preservation of these cultural patterns.

`Old-fashioned' can be a value-laden term, but it is one of those adjectives which could be complimentary or critical, depending on the context. It could mean tried and trusted; or it could mean out-of-date and inappropriate. Many of the old-fashioned characteristics of Asian family life are likely to resonate with a substantial body of white opinion, which favours a commitment to marriage and tight-knit families in preference to the current trends towards open relationships and loose loyalties. Other features of Asian family life, though, have now been rejected decisively within white families: these include arranged marriages, husbands' authority over wives, women's obligation to keep house, and large numbers of children. These are the areas where future generations of British-educated Asians (perhaps especially the girls) are most likely to question traditional patterns and face conflict with their parents.

It is also appropriate to consider the possible effect of family patterns on economic prosperity. Table 12 makes use of data and concepts established by the Department of Social Security for the measurement of poverty in Britain (DSS 1997). As one might expect, households where no-one has a job are much more likely to be poor than those with some earnings, and the distribution of poverty between ethnic groups is partly a simple function of the distribution of employment.

Although a number of data sources indicate that Indians have high rates of employment, and occupations and earnings which are on a par with those of white people, household - based measures tend to show that more Indian than white households have no earner; and that both working and non-working households have higher poverty rates than their white equivalents. It is not quite clear why these things are so, but the variations between whites and Indians are not large, and there is no particular reason to attribute them to the family patterns which are the subject of this paper.

In the Bangladeshi and Pakistani communities, though, the proportion of households containing no worker (and therefore living on social security) is far higher than in the other two groups shown in the table. Again, this is more likely to be a function of problems in the labour market, rather than any specific aspect of family structures. Nonworking Bangladeshi and Pakistani households have very high poverty rates - higher even than non-working whites or Indians. But it is when we turn to working households that family factors may have an important part to play. Astonishingly, as many as half of Bangladeshi and Pakistanis working households are below the poverty line. Their earnings simply do not stretch far enough - in spite of in-work social security benefits which account for a fifth of their income (IEM page 15). This is partly due to low pay. But it is also, undoubtedly, due to the large number of children in each family, and to the fact that so few wives and mothers have jobs of their own.