READING GUIDE TO: Platt, L.  (2005) ‘The Intergenerational Social Mobility of Minority Ethnic Groups’, in Sociology, 39 (3): 445 -61.

Studies of ethnic minorities and their mobility had problems in the past, because their class position was not very stable—it reflected their migrant status.  This study uses data from the ONS Longitudinal Study, based on a 1% sample of census data, from 1971 to 1991.  The study begins with children are aged 8 to 15 on census day in 1971.  The data is grouped into three ethnic groups [this seems to be traditional in these mobility studies]: Indians, Caribbeans, and white non migrants.  It is clear that ethnicity will interacts with gender and class origin.  The class position reflects the usual Goldthorpe findings that there has been quite a lot of upward mobility into the service class because of the expansion of the class, although the odds favour people with parents in that class already.  [Note that one study cited was based on the Nuffield data: Heath A & Ridge J (1983) ‘The social mobility of ethnic minorities’, in the Journal of Biosocial Science Supplement, 8:169-84].

Earlier studies took snapshots of occupations at different times, or they surveyed populations with the usual questions about parental social class and added questions on ethnic identity.  The snapshot approach tends to show general upward mobility, especially among Indians, less so among Caribbeans, who also have higher chances of unemployment.  These results were often associated with patterns of migration.  For example it seemed common to have downward mobility after entry into the UK, especially since qualifications gained overseas did not seem as strongly connected to occupation as was the case for British qualifications.  There were also certain cultural niches affecting occupations, such as the tendency for some Pakistani groups to end up in the textile industry.  Migration history was therefore important.  There was some evidence that Indian and Chinese people manage to regain upward mobility in the second generation, especially if they had parents with high levels of education themselves.  Those from a low skilled background experienced more generational immobility.

Thus migrants would usually have a different occupational history than non migrants.  However, this history tended to be over emphasized, which involves some interesting assumptions.  One is that there is some expected congruence between premigration and post migration class positions and educational qualifications, a kind of international class stability.  Methodologically, snapshots were not very good at identifying intergenerational mobility specifically, since they compared two samples at different times, but the composition of those samples might have changed considerably—some people might have left the country or died, new migrants might have come in, the samples might feature different age profiles.

Approaches which have used questionnaires based on class and ethnicity have been most developed by Heath, including Heath & Ridge (1983) [reference above] these studies compared non migrants with four basic migrant groups [unless I have misunderstood, only one of them was a non white migrant group, however, which implies that the studies were far more interested in migration and its effects rather than ethnic groups as such?]. This one group displayed weaker associations between origins and destinations, to their general disadvantage.  A later study, using GHS data, compared parental class, ethnicity and educational qualifications with destinations, and found differences produced by ethnicity, but the general complexity: for example, a salariat background [defined as in other studies as classes I, II and IVa?] conveyed a general advantage, probably reflecting the growth in that group and the advantage that growth provided for the whole population.

This particular study measures social class following migration, and examines the direct parent – child class transitions.  Two generations were studied and compared, 1971 when they were children, and 1991 when they were in occupations.  People were divided into the three groups mentioned above.  Methodologically, there are great advantages in using this kind of sample.  For one thing, we have actual records of parental social class in 1971 with no need to ask for dubious recall.  Some people will have left the UK between 1971 and 1991, but ‘we also know some of the characteristics of those who are “lost”’ (448).  For example, 15% of the 1971 sample were not recorded in 1991.  We know that 36% of the missing group had West Indian parents, 24% Indian, and 22% African – Asian [what about the other 18%?].  Given the compulsory nature of the census, it is unlikely that there were many errors arising from non response: surveys 15% really have left.  We also know something about the class differences among the leavers: the Indian group, and men generally tended to have a higher social class, but the effects can be considered to be slight.  51,005 people were left in the sample.

As one cohort is being followed through, the effects of having mixed ages and samples, as in snapshots, are avoided.  Data are also gathered from an intermediate point, based on the 1981 census [and we can predict further studies based on later census returns].

The study also modifies class and other classifications in an interesting way.  It adds an unemployed class as a destination category, building on earlier suspicions that ethnic minorities tend to end up in this category disproportionately.  It allocates children to families in ethnic categories if they are living with at least one parent.  A range of ages were used to define ‘children’—the eight year old limit because they wanted to allocate class reasonably by 1991 and the 15 year old limit to get sufficient numbers, for example.  The three ethnic groups were aggregated in this way, in order to preserve sufficient numbers for analysis.  People in Indian and Caribbean ethnic groups in 1991 were assumed to be unaffected by migrant status, even if they were not actually born in England and Wales, since they had all had some schooling here and had spent their entire working life here.

The CASMIN classification scheme was used to allocate people to social classes, and the categories were reduced to three basic class groups—service class, intermediate class, and working-class [in order to maintain numbers again?].  Unemployment was added as a fourth class for the reasons given above.  The sexes were coded differently: routine non- manual female labour was allocated to working-class rather than intermediate class groups [following the controversies that we’ve seen in other studies].  When coding married couples, Platt used the ‘dominance’ approach—the couples were coded according to the higher class of the couple, both in the 1971 and 1991 samples.  Platt decided to include ‘looking after home and family’ in the ‘unemployed category’: ‘like any assumptions, this is not unproblematic [!], but nevertheless seems to be the most appropriate’ (451).

Generally, the data shows that ethnicity does have an effect on mobility, more so than class origins for men, but less so for women.  We can show this by comparing data from the 1971 sample, on parents’ place of birth and social class, with the sample’s own ethnic group and destination class (and the class of their spouse) in 1991.  Tables show the full figures [I have selected particular examples as usual].

Class distribution does vary according to ethnic group in the 1971 sample.  The parents of the children being studied in 1991 had an interesting class distribution.  24% of the white parents had occupations that placed them in the service class, 14% of the Carribean parents were in the service class, and 7% of the Indian parents.  At the other end of the scale, 53% of the white parents were in the working-class, 71% of the Caribbean parents, and 78% of the Indian parents (451).

By the time we reached 1991, those children had attained their own class position and ethnic grouping.  38% of the white adults in the sample were in the service class, 22% of the Caribbean, and 35% of the Indian people.  Looked at the other end, 38% of the white non migrant group were in the working-class, 46% of the Caribbeans, and 35% of the Indian.  Figures for those in the unemployed class were 5%, 13%, and 6% respectively.  Among the unemployed group, 13% of the Caribbean and 6% of the Indian persons were in families where no one was employed, and 5% of the white adults.

Simplified version of Tables 1 and 2

As might be expected, structural changes are partly responsible for this improvement in the position of all the groups [they have all increased their representation in the service class, some more than others]: ‘the expansion of service and intermediate classes and the reduction in the working-class mean that a certain degree of absolute upward mobility was inevitable’ (452).  Over the period between 1971 and 1991, the service class grew from 23% of the parental sample to 38% of the children’s sample, while working-class groups shrank from 54% to 38% [these figures alter slightly between the tables], and, overall, 32% of those born into working class families made it upward to the service class.

This is a measure of absolute mobility, to use the Goldthorpe terms, and, as we should be aware by now, relative class mobility shows a different picture.  If there had been a ‘fair’ share of the service class expansion for children in working class families, we would expect to find 56% of them making it into the service class, but we find only 45% of them did.  At the other end of the scale, working-class groups in total make up 68% of the unemployed [and again, it should only be 56% in a statistically ‘fair’ society]. Similarly, we find service class parents able to ‘protect’ their children more effectively—somehow, proportionately, more children from service class families end up in the service class themselves.

For ethnicity itself, the outcomes for the white non migrant group were comparable to the overall pattern of class mobility described above.  The Indian group, however, have a different picture.  Indian parents seem less able to retain positions in the service class for their children, but they do experience more upward mobility from the intermediate class, and enjoy ‘substantial’ levels of mobility overall.  Those Indian adults ending in working-class or unemployed categories also seem less affected by their parental social class than the whites non migrants who we looked at above [that is, fewer proportionately from working class families].  Citing Platt’s actual data, of those Indian people whose parents were in the service class, 52% ended in the service class themselves, 11% in the intermediate class, and 37% in the working-class or unemployed groups.  Of Indian people who grew up in working class families, 33% ended in the service class, 23% in the intermediate class, and 44% in working-class or unemployed groups [Platt has combined the figures for working class and unemployed Indians here to overcome small numbers in each cell]

The Caribbean sample showed considerable scattering between origins and destinations.  They showed even poorer retention of service class parents.  Of those with service class parents, 30% ended in the service class, 4% in the intermediate class, no less than 50% in the working-class, and 16% in the unemployed category!  Of those with Caribbean working class parents, only 22% managed to get to the service class, 13% to the intermediate class, and 51% stayed in the working class, and 13% ended up unemployed.  The poor retention rates in particular by service class families requires more research says Platt.

Simplified version of Tables 3 and 4

To further explore relative chances, Platt calculated odds ratios, leading to a rather baffling table on page 454.  In general, the odds of staying in the service class if your parents were service class, compared to making it into the service class if your parents were working-class come out at 3:1.  If you come from working class origins rather than service class origins, the odds of you ending in unemployment are 2:1.

These class odds ratios are moderated by ethnicity as well.  The advantages of service class origins are smaller for Indians and Caribbeans than for white non migrants, so small for Carribean groups that the advantage approaches zero (454).  The chances of ending in working-class groups actually increases for Caribbeans and Indians of service class origin, showing that service class parents are unable to protect their offspring so well.  Intermediate class origins improve the chances of upward mobility for white non migrants, giving them twice the chance than is the case for Caribbeans.  The only disadvantage that the white migrant groups seem to suffer is that their parental social class seems to have a greater effect in their ending up in unemployment, compared to the Indians, where there is no such effect of working class origin.  [This isn’t really much of an advantage though, since other factors seem to propel Indians into unemployment at a greater rate overall.  What the figures are pointing to though is a very interesting picture where class and ethnicity interact differently to limit upward mobility and affect downward.  In policy terms, anyone wanting to prevent work white working class of people from ending up in unemployment would address class issues, while preventing Indian working-class people from ending up in unemployment would require addressing cultural and ethnic ones?].

Generally, higher class origins protect non whites less.  Ethnic minorities experience much more fluidity, both upward and downward.  This is almost certainly not to a greater openness of British (E & W) society to ethnicity rather than class, however, since Indians and Caribbeans generally do worse than white groups, and show less protection against downward mobility.

When we include gender, more complications arise.  There is the usual problem posed by women occupying a different distribution of class positions, and also having their position affected by their marriage.  For this reason, the best group to study the effects of class are unmarried women: ‘it is therefore the single respondents who are largely driving the differences between the sexes’ (455).  It seems, for example, that if we look at those ethnic groups in a working-class occupation, many of them are women—52% of Caribbean women are in working class occupations, and 30% of Indian women.  Caribbean men and women both figure prominently in the unemployed category (19% and 9% respectively), but we need to remember that Platt considers homemakers to lie in this category as well.  What this means is that Caribbean homemakers are not compensated by their partner’s class, often because they are not married—Caribbean unemployed homemakers are ‘predominately lone parents (or in unmarried partnerships)’ (455).  This is not the case with Indian homemakers who are usually married and therefore allocated a class according to their husbands’ occupations.  [One of the effects of gender and ethnicity, then, is on the kind of household that is constructed, with consequences for class position. There is also a measurement problem, of course, arising from Platt's coding decisions as above].

Platt applies various statistical models to try and measure the associations between origins, destinations and ethnicity, and she does this separately for men and women to reveal gender differences.  I do not claim to understand fully the models being used, but I gather that one possible outcome is to reveal no association between these variables, so that only the overall distribution of each variable affects the value of the other variables [so lots of Caribbeans would produce equal amounts of variance in social class position but no more?].  At the other end, another possibility is ‘saturation’, where there are strong associations between pairs of variables, and these associations then dominate connections with third variables [so a very strong association between ethnicity and gender would produce an equally strong connection with social class?].  After discussing what would count as a good enough fit for this model on these data [pass], Platt displays the results.

The first possibility, assuming no association does not fit the data, but the best model is not the saturated one either.  Instead, we seem to be able to show a set of three paired relationships:  between ethnicity and origin, origin and destination, ethnicity and destination. Destination chances are constant across class groups with ethnicity, and there is a clear link between origins and destinations too.

These associations are not found so strongly with females though.  ‘This indicates that while ethnicity interacts with both origins and destinations, we can still observe common class processes affecting women’s outcomes, regardless of their ethnicity’ (457).  Gender is more important than ethnicity for women, in other words.  Across the ethnic groups, marital status provides a common link for women.  Perhaps, aspirations and the effects of class position might be transmitted differently for women?  For men, class processes appear to operate differently according to their ethnicity: parental social class positions were influenced by ethnicity, and their own ethnicity continues to have an influence.

In conclusion, we can see differences between groups according to their ethnicity, not only differences between whites and non whites.  Ethnic minorities have different distributions of classes of origin.  Their destinations reflects these class origins, and they have responded to the familiar structural changes of class distribution.  Yet there are particular patterns according to ethnicity as well (458).  It seems that migration experiences have still had an impact on the people being studied, but so have their own experiences growing up in Britain.

 There doesn’t seem to have been a familiar pattern of downward mobility followed by upward mobility.  Instead, there has been some retention of class position, more so for Indians than for Caribbeans [and some decidedly continued downward mobility].  Platt entertains the curious idea that this failure to recover might tell us something about the second generation not being as dynamic as the migrant parents, but concludes that we have evidence of ‘an environment that is particularly antipathetic to Caribbean success’ (459).  Gender does seem to offer some common ground among groups.  Class is still an important factor for groups even though it has different specific effects: generally, class seems especially important for women, and ethnicity for men.