Wednesday, 26 March 2014

The trouble with "Socio Economic Status"

It is dismaying to notice how this term and its abbreviation "SES" is returning to reports and discussions of research in health inequality. I thought we had done away with it at least outside the USA, and adopted "Socio Economic Position" (SEP). The virtue of SEP is that it is a very general term, referring to social class, social status, income and material circumstances. As these measures of life circumstances are often used pretty indiscriminately in research studies, if you want to do some kind of review paper, you often have to use the measure that is present in each paper. But as Nancy Krieger and colleagues pointed out many years ago now, if we are going to get serious about explaining as opposed to describing health inequality, we need to be more precise about what our input variable actually is. Because no one imagines that income, wealth, social status or occupational social class come along and bonk anyone over the head resulting in disease or mortality. If you think about it for one second, you realise that there are always 'proximal' factors like environmental conditions, housing quality, work conditions, psychosocial stresses or health behaviours that do the business for health. But different dimensions of inequality can be shown to have different relationships to these different proximal factors. Social status is most strongly related to the behavioural patterns at the present time (this will not necessarily be the same at all times in all places). Occupational class has the strongest relationship to working conditions. Income is the strongest determinant of housing. And even the income-housing link is different in different countries according to housing policies, for example in Denmark housing is not as segregated as in the UK.

Neither is it useful to talk about "high" and "low" SES (or SEP for that matter). Yes, income can be high or low, but that needs to be explicit before we can make sense of the mediating factors. Social status can also be high or low if we are thinking in terms of caste-like phenomena where members of certain occupations or groups defined in terms of religion, language or skin colour tend to avoid each other's company. Here again, if you think clearly you can imagine how such discrimination might well act according to stress pathways as described by Wilkinson and Pickett. Caste groups in Hindu societies also adopt different behaviours such as diet. A person wishing to pass as a member of a higher caste will need to change these behaviours, and this is what we find in relation to smoking and social mobility. These are real examples of the Black Report's "cultural-behavioural" explanations, that make sense in sociological terms. So one could think in terms of low income or low status if that is what one's explanatory hypothesis for a certain aspect of health inequality entails. But "low" and "high" are of little use as general blanket terms. We know that members of certain ethnic groups such as Catholics in Northern Ireland and The Netherlands, or people or Irish or Caribbean descent  in England for example, tend to have suffered historical discrimination and as a result have lower incomes and also worse health than Protestant or Anglo-European groups. But we do not use the blanket terms "higher" or "lower ethnicity" to refer to this. It is clearer to use  "more" or "less advantaged" as the more general term.

Recently there has also been a return to using education as a measure of SEP. This makes things even more confusing. As important as education is in the life-course factors that determine what kind of conditions and attitudes a person will have in adult life, it is not in itself a dimension of social position. It is a very strong indicator of where someone will end up (and their resultant exposures). In fact, if policy were so organised that people who started off in a less advantaged social position as children were given extra attention at school, population health might be greatly improved. But even this would not eliminate social inequality which arises from the structure of occupations and the income distribution. We will soon be able to see whether the increasing proportion of graduates who end up in low paid (income) and insecure (employment conditions) jobs will change the relationship between education and health.

The use of education as a measure of social position does give us a hint, however, of the thinking that creeps into research on health inequality. As pointed out by Gordon Marshall (even longer ago than Krieger et al.'s seminal work), there is a sneaky eugenic argument that lurks beneath the use of terms like "lower" and "higher" SES. This is that people with superior personal characteristics have better health. So their success at school, their more advantaged occupations,  their higher income and their better health are in fact all consequences of this underlying superiority. The enthusiasm in epidemiology for genetics gives further strength to this kind of thinking. People in "low social positions" tend to smoke? Well, what do you expect, they are not bright enough to figure out the dangers (by the way, what evidence we have points strongly in the opposite direction, that there is no link between social position however measured and the desire to stop smoking).

I do not actually believe that many social epidemiology researchers mean to make this kind of argument. After all, we would put ourselves out of business pretty quickly. There is no point to public health if everything is determined genetically.

1 comment:

  1. Many of us know that genetics does not contributes much to our health. I am glad that you mentioned education which often used as a proxy measure of 'cognitive ability'. I think it is misleading to use education (often socially determined) as innate individual ability.
    If we are looking at social disadvantages to understand the mechanism of health inequality, shouldn't we work to create a combined measure, instead of looking at occupational position, income or education independently?