social mobility ‘research’
— science vs normativity
In Part 1 of this article, we noted that most research on social mobility simply assumes the absence of genetic effects, often without bothering to state the assumption, and that this largely invalidates the policy implications derived from such research.
‘Research’ in this context means making comparisons of intergenerational correlation (the degree to which offspring socioeconomic position is statistically linked with parental position) in the light of researchers’ belief that a lower correlation is necessarily better. This belief may be based on the idea that a lower correlation equates more closely with perfect meritocracy; however, this is rarely spelled out.
Imagine a scientific field called social mobility research starting in an era before a majority of Western intellectuals had acquired their current objections to inequality, or developed their preference for a blank-slate model of intelligence. Folk wisdom has probably always held that there is some degree of meritocracy in where people end up in the pecking order, and that intelligence and other characteristics are at least partly inherited from parents. One can imagine, then, that research would have started with a detailed investigation of heritability of abilities, before moving on to compare the intergenerational correlation expected from the heritability numbers, with the correlation actually observed. Prima facie, a correlation higher than expected would imply a mobility level that did not match perfect meritocracy, while a correlation that was lower than expected would suggest that mobility was too high.
Researchers would have noted the various limitations of this statistical approach, including that:
• whatever the innate (genetic) abilities for which one can calculate heritability figures, one can never be sure how related they are to the skills required for any given social role;
• societies change, hence a meritocratic pattern of correlations expected from heritability figures at one point in time might no longer be relevant by the time children have turned into middle-aged adults.
In particular, researchers would acknowledge that if the numerical level of mobility was found equal to the number expected for perfect meritocracy, this would not prove that the society studied was meritocratic. Consider, for example, a society in which perfect meritocracy would require half of the lower-class population (the more able half) to become higher-class in the next generation, and half the higher-class population (the less able half) to become lower-class. (In using “become” I am referring to their offspring.) But a scenario in which the less able half of the lower class becomes higher class, and the more able half of the higher class becomes lower class (i.e. anti-meritocracy) will generate exactly the same mobility statistics as those for perfect meritocracy.
Researchers would also — hopefully — be hesitant about rushing into policy conclusions, even if the observed level of mobility could be conclusively determined to differ from some supposedly ideal level. A level of mobility that was either ‘too high’ or ‘too low’ would not necessarily imply a society operating at some suboptimal level. Societies are complex mechanisms, and achieving whatever they are doing in the most efficient way may not be a society’s only goal.
Now imagine a different scenario. In the one-party state of Begonia, the government gets tired of the difficulties of determining heritability for purposes of obtaining meaningful measures of meritocracy, and simply decides to prioritise mobility per se — i.e. to regard socioeconomic differences between parent and offspring as intrinsically desirable. The lower the correlation between parent income and offspring income, or between parent education and offspring education, the better. The Begonian government hires a group of researchers and statisticians to form a new department known as the Department of Social Mobility. Their job: to investigate how Begonia is faring with regard to the objective of achieving zero parent-offspring correlation. How ‘good’ (i.e. close to zero) is the correlation level compared to other countries? If it is worse (i.e. higher), what reasons can be adduced for this? Has the correlation level been getting worse (i.e. higher) recently? If there was a period in the past when the trend was clearly ‘good’ (i.e. downward), how do we restore this trend? Overall, the principal question for the new Department is: how can the prevailing parent-offspring correlation in Begonia be reduced?
It is possible, of course, that the government’s objectives match up with those of the population, if the majority is more interested in having as low a linkage as possible between children’s social origin and destination than in bringing about meritocracy. In that case the government could be said to be merely doing its job in trying to bring about minimal linkage, and using researchers to investigate how this might be achieved. However, it is clear either way that the government’s agenda taints the objectivity of the researchers. The researchers’ efforts are focused on bringing about a particular end result. Analogous to the distinction between science (expanding knowledge) and engineering (achieving specific applications), the researchers would be better described as social engineers than as social scientists.
Once upon a time, favouring social mobility meant favouring the upward movement of talented people from the ‘wrong’ social background. The son of a baker who, having the aptitude for being a university lecturer, actually manages to get to that position; the daughter of a plumber who achieves the position of lawyer. There have always been sources of opposition to such movement — hostility to the ‘uppity’ individual who thinks they are better than their background — but, for a time at least, the concept gained favour in the West. Scholarships, grants, and (in Britain) the grammar schools were part of this philosophy of giving weight to meritocracy.
Then came the 1960s, the rise of the ‘social sciences’, and the influx of Marxism-influenced intellectuals into academia and politics. Rather than meritocracy, which was seen as favouring the middle class too much, the emphasis started to be on equality. The blank-slate philosophy — i.e. commitment to belief in zero heritability — was gathering momentum. This was the era when Soviet communism was viewed fondly by mainstream Western intellectuals, and governments such as the British looked to imitate the supposed successes of the Soviet education system.
For researchers, there was also the practical problem that heritability is difficult to incorporate into statistical analyses of mobility. Far easier just to focus on the raw numbers for movement between classes, and not to worry whether genetics might mean that zero correlation between positions of parent and offspring is not just unrealistic but undesirable.
The result: social mobility came to mean, simply, absence of correlation between parents and offspring. This is still where we are today, half a century on. Now, instead of asking “how easy is it for a talented baker’s son to become a lawyer”, the average article about mobility simply asks, “what proportion of working-class children end up middle-class?”, with the implication “the more, the better”. Since the answer “all of them” would presumably seem too ludicrous even for left-leaning readers, the basis for comparison generally used is other countries, or previous years. If the proportion is lower than in Denmark (a popular touchstone), we should be concerned. If the proportion has failed to rise in recent years, we should be concerned. If the proportion has actually fallen, we should be extremely concerned.
The original point — that the relatively talented might be given a helping hand — has been lost. The redefinition of mobility has been achieved partly by demonising anyone who raised the issue of heritability as a “social Darwinist”. The vehemently negative reaction to The Bell Curve by the intellectual establishment provided an obvious demonstration of this phenomenon.
Politics may use results from science and statistics to make its points, but it rarely hinges on scientific truth. Politics is about what people want, or what they would prefer. This may include not only such goals as “more money for people like me”, but also “a society that fits better with my ethical ideals”. Another popular goal may be “more power for people like me”, where “me” is a member of an elite responsible for determining what the ideal society should be like.
All this means we should not expect political debate about topics like social mobility to be objective or scientific, notwithstanding superficial appearances. Such debate will depend largely on normative assumptions — what is the right thing to do? — and these are beyond scientific or logical demonstration.
When it comes to an academic discipline, on the other hand, it is reasonable to expect that scientific standards will be adhered to, and in particular, that normative assumptions will be kept out. If it proves too difficult to keep normative assumptions out — because the science by itself, stripped of ethical biases, is too dry and dull for consumption — then it is reasonable to expect that any such assumptions will be clearly spelled out, and that space will be devoted to alternative viewpoints. Such efforts towards neutrality are particularly important in the social sciences, because the results are likely to be seized on for political ends.
Sadly, the social sciences seem to have gradually lost whatever initial impetus towards neutrality they may have had, and instead become promoters of one particular political world view. The results-will-be-seized-on point, rather than encouraging neutrality, is now typically applied as follows:
The other side, who are morally wrong, will try to exploit anything we produce that might support their point of view. Therefore we must massage what we do — as far as possible without actually being untruthful — in such a way as to avoid giving them any ammunition.
On a statistical level, it would be interesting to be able to put a number on the degree to which innate ability is correlated with eventual position, and then to look at differences in that number over time or between different countries. Unfortunately, determining such a correlation is not feasible at present, since we cannot directly measure innate ability. It could be argued that we barely understand the concept of “innate ability”. As with other research in the human sciences, we are working with ill-defined ideas, and trying to generate quantitative results that will somehow paper over the cracks in our understanding.
This inability to get at the fundamental variables generates a demand for a quick and dirty workaround. Such a workaround may ‘work’ adequately for purposes of informal political discussion; the question is whether there is a place for such workarounds in disciplines that are aiming to be scientific — or indeed in any field of academic research. If a quick and dirty workaround is employed in such a context, it is reasonable to expect that researchers would spend a significant amount of time drawing attention to the limitations of their approach, both at the start of their academic papers, and when they come to the drawing of putative conclusions. Such expectations are not, unfortunately, fulfilled in practice. Gaps in understanding, and limitations in theoretical approaches, are largely ignored. Researchers occupy themselves with poring over statistics, and inventing narratives to explain why things are ‘worse’ (i.e. mobility is lower), meanwhile leaving a big question mark over the significance of their version of mobility and why it deserves to be treated as a worthy goal.
The humanities have been increasingly influenced by Marxist perspectives. One of the tenets of Marxism is that cultural products must always be viewed in their historical context. This tenet may in fact be helpful for understanding the background of contemporary sociology. Sociology as a discipline was started by individuals who believed that society could — and should — be changed, and who believed it would be the role of sociologists to provide the blueprints for the better society. The discipline has thus always had an instrumentalist rather than a theoretical slant: the emphasis has been on changing things rather than understanding them. Such an agenda is bound to create certain biases. There is likely to be a bias in favour of seeing society as something that is intrinsically flawed, and against seeing it as something that is organically or functionally determined. There is likely to be a bias among sociologists in favour of promoting political change, and against simply observing and describing in the manner of anthropologists or naturalists.
Perhaps there was a time when academic sociology was still characterised by a certain degree of pluralism. By now it is reasonable to surmise that the vast majority of currently active academic sociologists have, as a background philosophical stance, a belief that sociological investigation must always be informed by awareness of social injustice. Every social phenomenon, according to this perspective, needs to be seen in relation to the unfairness that supposedly characterises society in general, and Western society in particular. It seems unlikely that someone who did not signal their commitment to this philosophy would nowadays be able to acquire a full-time post in academic sociology. The philosophy of social justice has become as much a part of the apparatus of contemporary sociology as the theory of evolution has in biology.
In the 1970s, there came a slight backlash from individuals such as H.J. Eysenck who pointed out that if ability had a heritable component, we would expect a degree of intergenerational correlation. Rather than taking the point about heritability on board, the social mobility establishment reacted in the form of explicit rejection. For example, Arthur Goldberger (1979) asserted that “heritability estimates serve no worthwhile purpose” in relation to social policy. His argument for why researchers could ignore heritability went as follows: if we knew that certain genes were predictors of short-sightedness it wouldn’t stop us wanting to correct short-sightedness in adult life; ditto for any genetic predictors of low socioeconomic status. In effect, the argument invokes a disease model for below-average IQ: relatively unintelligent individuals are to be regarded as handicapped in the same way as people lacking a limb, making them automatic candidates for intervention.
I argued above that, if the priority had been understanding rather than action, the field of social mobility research would have started with a detailed investigation of heritability, before moving on to compare the intergenerational correlation expected with the correlation actually observed. The fact that the order of events was inverted, and that ‘research’ initially proceeded without any consideration of heredity — then later proved stubbornly resistant to suggestions that such consideration was relevant — suggests that political objectives, and not knowledge or understanding, have been the primary drivers all along.
The practice of applying an undisclosed zero-heritability assumption is common in academic research, and not confined to sociology. For example, economics professor Thomas Piketty spends several pages of his book on inequality, Capital in the Twenty-First Century, discussing the alleged failure of education systems to deliver social mobility, without once referring to the issue of inherited ability. While the data he presents may be meaningful, his accompanying commentary is misleading and tendentious.
Professor Piketty was surely aware of the issue of heritability when he wrote his book. It is therefore hard to avoid the impression that the issue is being concealed, by use of statements that are intentionally vague, with key terms being left undefined. For example, Piketty writes:
In all countries, on all continents, one of the main objectives of public spending for education is to promote social mobility. The stated goal is to provide access to education for everyone, regardless of social origin. To what extent do existing institutions fulfill this objective?
Professor Piketty here mixes up two possible objectives for state education: (A) to promote social mobility, (B) to provide access to education for everyone regardless of social origin. Piketty nowhere explains what he means by “social mobility”, so one must assume he is adopting the usual meaning of “absence of intergenerational correlation”. It is not at all clear whether (A) actually is one of the main objectives of state education “in all countries and on all continents”. In any case, the two objectives may be related but they are not the same, and the differences between them are highly pertinent to the question of whether mobility can be used as a proxy for meritocracy. If “the stated goal” is really, as Piketty asserts, to provide access to education for everyone regardless of social origin (in all countries and on all continents) then it does not necessarily follow that there also has to be a goal to promote greater movement between social strata. It depends on the background assumptions about heritability.
Having conflated two distinct issues, Piketty proceeds to look at numerous comparisons of intergenerational correlation between different countries and different times. In each case, his implication is that when apparent mobility is lower, either compared to an earlier time or to another country, we have a suboptimal situation. No explanatory background is given to this normative assumption. Readers are simply expected to take it as a given that mobility (including, in effect, random mobility) is an intrinsic good. In this, Piketty is simply imitating the standard approach of ignoring genes and adopting the simplistic equation:
mobility = meritocracy = good.
Tacit denial of heredity has become the default practice in academic papers, and in newspaper columns and internet articles. However, in a 600-page book devoted to the topic of inequality, one might have expected at least some exploration of the philosophical underpinnings.
© Fabian Tassano
published 9 September 2022
1. This follow-on article to ‘Social mobility research — on another planet!’ was originally due to be published in early 2020. That it has taken far longer than planned is partly due to living in a post-COVID world but primarily reflects lack of support for genuinely critical perspectives such as those of myself and my colleagues at Oxford Forum. I am hoping in due course to publish a final instalment, in which I look in greater detail at the language, and hidden assumptions, of the academic literature on social mobility.
2. Quote by Arthur Goldberger is from: Goldberger, A.S. (1979), ‘Heritability’, Economica 46, 327-347. Goldberger was an economics professor at the University of Wisconsin.
3. Quote by Thomas Piketty is from: Capital in the Twenty-First Century, Harvard 2014, chapter 13.