Reading Guide to: Bowles S and Gintis H (2002 ) 'Schooling In Capitalist America Revisited', in Sociology of Education, Volume 75, 2: 1 - 18 ,  and [online]  (thanks to Bruce Guenard)

This is a densely argued piece of work, devoted mostly to statistical explanations and arguments, and I have summarised these very ruthlessly here, partly because I don't fully understand them. I was also amazed to see that so much work had been done by these two authors to follow up the pieces that I have collected on my website. A full bibliography is available at the end of the published piece .


Overall, this study suggests that a wide range of new empirical results indicate that there is an intergenerational persistence of 'economic status', and that there are high economic returns in terms of income to years of schooling. There is a greater effect than is found with hereditary IQ. However, the contribution that schooling makes to cognitive development is not the major way in which education produces economic benefits: it is more the personality traits that schools teach rather than any skills. This is exactly what was predicted in Schooling In Capitalist America (SICA).

SICA argued that the cognitive development enhanced by schooling was weakly linked to economic success. Instead, socialisation, via the correspondence principle, explained more of the variance: that was because educational environments were very similar to work environments. Further, parental economic status is passed on to offspring, and only partly through educational opportunity. Success is not related to cognitive skill, nor is there any sign of growing equality of opportunity of access to high-status jobs. Finally, the evolution of the schooling system could not be explained through the development of a democratic ideal, but through the existence of class conflict.

The statistical claims made then are still valid. Research is continuing and will lead to more work, but so far the arguments are:

(a) The connections between parental economic status and the status of their children used to be thought irrelevant. But the studies that showed this contained a serious measurement error which under emphasised permanent income. More modern studies show the effects are substantial, with correlations varying between 0.3 and 0.7 (page 2). The income of parents is as good a predictor of children's income as years of schooling or cognitive score [IQ]. Marvellous diagrams like the one on page 4 [a beautiful three-dimensional version of the density diagram found in Lockwood] shows that there are twin peaks in terms of density [that is, the extent to which membership of the class is passed on from parents to children]. The poorest sector have a 19 per cent density and the richest 22 per cent, for example. However, it seems likely that specific mechanisms of transmission are at work in different sections of the distribution, since density varies quite a lot. The average figure overall is 0.36, but there are considerable differences concealed within this average. For example, it seems likely that wealth transmission is an important mechanism for the richest groups, and vulnerabilities to illness or violence for the poorest.

(b) The role of schooling especially can be isolated. If we start with IQ, there are substantial correlations in scores between parents and offspring (0.4 - 0.7). Effects can be estimated by suggesting that a change of one standard deviation in cognitive score will produce a change in earnings of 1/7th of a standard deviation of earnings. Measures of years of schooling seem to have a greater independent effect. There are no signs of cognitive development increasing as a factor in determining income, however. Measurement errors relating to variations in tests and timing also appear to be insignificant. How important is IQ in connecting the cognitive performance of parents and kids? How important is cognitive performance in determining future earnings? Answers to the first question seemed to show a fairly low correlation -- cognitive performance explains 0.035 - 0.018 of the variance in incomes. Illustrating this in another way , if IQ is the sole determinant, then the richest kids are 12 per cent more likely to get to the top income decile, while in reality they seem to be between 16 and 44 times as likely to get to that top decile.

(c) What are the effects of schooling on labour market success? Here, the noncognitive elements seem very important, at least as revealed following a survey of employers in 1998. For the employers, 'attitude' seemed the most important factor, then 'communication skills', far above academic performance or test scores. A British survey in 1998 is also quoted -- one-third of employers reported a skills shortage which they defined as a 'lack of technical skills', but 62 per cent reported a shortage of candidates with a suitable 'attitude, motivation, and personality'. In the USA, there is a specific diploma which may be taken by high-school drop-outs [evidently intended to provide them with qualifications to assist their future employment]: diplomates possessed higher cognitive skills but found themselves with very similar incomes to those who had not taken the diploma. The importance of 'socialisation' seems to be clear, especially in people's abilities to retain a job, and so become personally effective in it [a 'fatalistic' attitude seems to be important here, and we shall return to it below]. A clever model on page 8 shows that year spent at school rather than the possession of cognitive skills seems to have a contribution to earnings. This is based on a collection of studies between 1950 and the 1990s in the USA, so some estimate of changes over time can be made. In general, the effect of cognitive skills is relatively small, explaining a mere 18 per cent of the total variance of schooling against income -- and there has been little change over time.

So, the contribution of schools to cognitive development is low, and the contribution of cognitive development to earnings is also independent of school -- the years of schooling seem much more important. How might variations in earnings be explained, especially among individuals with similar demographic characteristics? We need to go on to investigate the role of 'personality', especially the importance of 'integrity' and 'conscientiousness' as predictors of success (page 10). A number of studies, including Jencks' [who famously offered a rival interpretation of the data originally displayed in SICA, which showed, among other things that 'luck' played a major role in determining future incomes, instead of social class] have pointed to factors such as 'industriousness, perseverance, and leadership' as well as 'study habits and other behavioural patterns' (page 10). These seem to be very important in predicting hourly earnings, four times more important than test scores, twice as important as family background, and 50 per cent more important than years of schooling. Studies of these factors include research on motivations, and here items such as a fear of failure, or a preference for a challenge over affiliation, seem important. Other studies have pointed to church attendance, social participation, and even 'the cleanliness of the respondent's home' [this reminds me of lots of old British research, such as Fraser's]. Some recently reworked correlations have produced even more evidence that motivational and behavioural variables explained most of the effects of years of schooling on subsequent income.

There is evidence that these factors impact on the earnings of women especially, in both the US and UK. Behavioural traits seem more significant for them. The studies that showed this have attempted to address the usual measurement errors, for example by trying to devise an independent test of personality which was not itself related to level of wages. The most significant personality variable for these studies appeared to be fatalism, which had a negative influence on earnings of about seven per cent (12). Aggression had an influence of eight per cent, and withdrawal one per cent. There were gender differences here again -- women seem to face greater penalties for being 'aggressive', and men for being 'withdrawn'. Men can actually be rewarded for 'aggression' and women for 'withdrawal'! These factors vary in their importance according to the status of the job as well, so that 'withdrawal' can be rewarded for higher-status jobs, and penalised for lower-status jobs. There are also some national differences between the UK and the US [little explored here]. However, Bowles and Gintis end their summary of these studies by agreeing that no one knows whether personality traits 'are simply proxies for (or perhaps contributors to the acquisition of) unmeasured skills or are valued as such by employers' (page 12).

On a more general note, the correspondence principle in SICA stressed the schools' role in socialisation, especially in getting pupils to accept authority. But this was insufficiently detailed in terms of analysing how new attitudes were learned, which ones, and by whom. People read the argument as suggesting that individuals were passive, the oppositional movements could not be explained, and of that the authors were guilty of seeing pupils as 'oversocialised' [with a reference to a famous critique of Parsons and Freud by Wrong].

In response, Bowles and Gintist now think in terms of a model of socialisation [which looks a lot like rational choice theory to me -- a pretty odd development for a couple of marxists, and one that seems to contradict their 'gramscian turn' in 1980]. Basically, schools are able to influence the cultural models to which children are exposed, and they offer an 'oblique' transmission of values, compared to the 'direct' transmission offered by parents. Schools are able to marshal considerable rewards and punishments to affect the cultural patterns of pupils: pupils find themselves having to instrumentally adopt some of these values. Schools have a special role in the development of modern technical knowledge which can stand in opposition to parental values. If these different values to clash, pupils often have to make a pragmatic choice between them, and this is affected by the rewards on offer. It is worth reminding ourselves that there is a close association between the personality traits desired by schools and the grades that pupils receive. Thus there is a high pay-off for adopting some school values at least, and this is how schools help to reproduce society, not as simple reproduction, but more as a form of evolutionary reproduction. School values seem to pay-off in cultural and social terms as well as economic terms (14).

The model is to be spelled out in much more detail in subsequent work. In the meantime, it is apparent that actual patterns can produce heterogeneous mixtures of cultural traits or stable clusters of either oblique and direct values. Research on this model is to replace the standard sociological work on socialisation in the overall theory.

In conclusion, the analysis in SICA seems to have been upheld. Functionalist models of the adjustment between education and the economy have been refuted, and the more specific and concrete historical analysis justified. It is the case, however that overt class conflict has diminished in the USA, and that it has never been that important in US history anyway [another difference here with the UK?]. Bowles and Gintis agree that they need a much fuller account of conflicting and contradictory pressures in schooling, and point to their 1981 [1980?] article as making a contribution here. They agree that SICA is affected by the times in which it was written, where there was a long economic boom and a much more lively counter-culture. This partly led them to assume that some democratic alternative was possible in the foreseeable future, and they are not so sure now. They agree that schooling needs much more limited and specific reforms, and that they should be offering advice on these. They end by arguing that schools are uneven and contradictory 'testing grounds and battle grounds' (15).