Disentangling the SES-health correlation


In the field of social health research, it has been shown difficult to disentangle the correlation between poor health and low social economic status (SES) into causal effects in either direction. Both causal interpretations seem plausible: when in poor health, improving your career becomes more difficult, and people of a low SES generally live in a more unhealthy manner, for instance due to poor housing and eating habits.

Despite all the difficulties of disentangling this correlation,Case et al. (2005) have made a very serious attempt in doing so. Using British `National Child Development Study’ data, which followed all children born in the week of March 3, 1958 in Great Britain, many measures of health and social economic status were available to the researchers for respondents of (very) early age, adolescence, and adulthood (age 42). This allowed for the theoretical model depicted in figure 1.


Since the researchers were able to disentangle all those effects, the level of detail in their analyses is very high. The theoretical support of their claims is however somewhat shallow, whereby I had to extract this model from the analyses, rather than from the theoretical description. Nevertheless, several relevant distinctions are made, such as for instance the distinction between the effect from SES from the father and from the mother.

The most important finding is summarized by the authors by the following: “Controlling for parental income, education and social class, children who experience poor health have significantly lower educational attainment, poorer health, and lower social status as adults.” In the many analyses performed, it becomes clear that this impact is robust to all other controls. Additionally, it was shown, that the impact from parents’ educational level and SES on their children’s SES is largely mediated by the children’s health and SES during early adulthood. In general, the authors conclude that their findings support three theoretical mechanisms: The life-course models (childhood health has a lasting impact on SES), the fetal origins hypothesis (uterine environment has a lasting impact on SES), and the pathways models (childhood health influences early SES, which subsequently influences health at a later age).

Although I think this research has been worked out really well and the research problem is of great interest, I have two issues with it. The first is probably due to a difference in `taste’ between the econometric literature it was found in and the sociological literature I’m used to. It comes down to the way some analyses were reported: in stead of showing estimated parameters, the authors show the level of significancy of the relative impact of groups of predictors. This allows to show a great amount of information in a single table. It also allows to see whether or not a group of predictors is capable of explaining the effects of other predictors. But it does not inform the reader on the magnitude of the estimated parameters. Apparently, the authors were more interested in the question what does have an effect and what not, instead of the question how much effect some factors have. To me, this seems as a missed chance, for more understanding of the problem at hand would have been gained when both measures (parameter estimates and level of significance) would have been reported.

The second issue is of a theoretical nature. The authors investigate almost all relationships between parent SES, child health, adult health, and adult SES. Almost, for one interesting relationship is missed: when poor parent SES leads to poor health of their children and when poor health as a child (directly and indirectly) leads to low SES as an adult, a “sociobiological vicious circle ” (Conley & Benett, 2000) may be present. That would mean that social inequality would remain in existence (partly) mediated by poor health.

An article by Conley & Bennet (2000) investigated this to some extent: they focused on the relationship between birth weight (as an indicator of poor health without influences of the person himself) and life chances. They were able to show that low birth weight indeed leads to lower educational attainment, thereby generally corresponding to the findings of Case et al. (2005). However, they showed that the income of the parents or mother (as an indicator of SES) does not influence the odds of having a child with low birth weight.

While many differences between these studies exist, it is interesting to see how they complement each other. It seems that poor health as a child has an enormous and continuing impact on life chances in later age, but that this risk of low SES is not transferred to ones’ children by means of poor health.

CASE, A., FERTIG, A., PAXSON, C. (2005). The lasting impact of childhood health and circumstance. Journal of Health Economics, 24(2), 365-389. DOI: 10.1016/j.jhealeco.2004.09.008

Conley, D., Bennett, N.G. (2000). Is Biology Destiny? Birth Weight and Life Chances. American Sociological Review, 65(June), 458-467.

Leave a Reply