Abstract
In this book, written towards the end of a long career, John Goldthorpe summarises his views about the nature of sociology. The book builds on a previously published set of papers on this topic, these coming out of reflections on his own empirical research and that of others (Goldthorpe, 2007). He reports that he has become more optimistic recently about the future of sociology, since the kind of research to which he is committed is flourishing, albeit more in continental Europe than in the UK. This involves survey-based studies of sociological problems, such as social mobility, in which statistical analysis is used to identify regularities that require explanation in terms of individual-level causal mechanisms. Meanwhile, some of the competing approaches that he previously felt threatened the future of sociology, notably ‘“grand” historical sociology’ and ‘postmodernist ethnography’, ‘have fallen into decline’ (p. 1). Whether he is right about these trends remains to be seen, but in this book he provides a stimulating account of the character of, and rationale for, the longstanding, and influential, version of sociological work to which he is committed.
The book is relatively short, organised into chapters focusing on a series of propositions that he takes to define sociology as a ‘population science’: that it allows for the high level of individual variability among human beings, that the social actions of individuals must be accorded causal primacy, that the task is to explain probabilistic population regularities, that statistical data and analysis are essential in identifying these, that statistical analysis cannot explain these regularities on its own, and that the task is to identify the causal mechanisms that generate these regularities.
As should be clear from this, Goldthorpe’s conception of what constitutes sociology is very narrow when compared with the range of work done by sociologists today, or even in the past. This is a point he acknowledges, writing that if sociology were to become a population science ‘the scope to which it presently pretends would be significantly reduced’ (p. 126). And in this book he sets out to clarify the character and rationale for this kind of sociology. Moreover, as the label he gives it implies, he sees parallels, or links, with similar work in disciplines like demography and epidemiology.
Goldthorpe does not discuss most of the sociological work that falls outside his definition of it as a population science. Thus, while he recognises that a considerable number of sociologists deny that the discipline can claim scientific status, or even that it should aim to achieve this, he dismisses their arguments as ‘selling sociology short’, and ‘standing in the way of it realising its potential’. Nor does he engage with those who adopt somewhat different conceptions of what a social science entails. His reluctance to address those with whom he shares ‘little common ground’ (p. 3) is understandable, but it could be regarded as a form of parochialism, one that is all too common in social science today: a tendency to operate within one’s own ‘paradigm’ and its immediate vicinity, largely ignoring what goes on elsewhere. In many ways, this further reinforces the fragmentation of the field.
Indeed, Goldthorpe is sharply dismissive even of approaches that actually share much in common with his own, such as Qualitative Comparative Analysis (QCA). He presents detailed arguments in relation to QCA, some of which – relating to the effects of random processes operating in society – clearly require investigation (see Cooper and Glaesser, 2016). However, focusing on what he sees as QCA’s non-probabilistic orientation, he underemphasises its most important feature: that it is a means of exploring the effects of combinations of explanatory factors in a way that is difficult (even if not impossible) to do in regression analysis. Goldthorpe also largely ignores the range of other methods employed by sociologists today. For instance, in one of his earlier volumes he had a chapter on ethnography, albeit a highly critical one (Goldthorpe, 2007: vol. 1, ch. 4), but here there is little mention of this or of any other qualitative method that could be used in identifying causal mechanisms. Indeed, he rejects the arguments of some influential political scientists that case studies, of various kinds, can serve this function: he dismisses Mahoney and Goertz (2006) as ‘proponents of logical analysis’ who adopt an ‘extremely limited – and self-serving – view of quantitative sociology’ (p. 108).
In outlining what he means by ‘population science’, Goldthorpe employs a number of contrasts. One of these is with what he refers to as the ‘holistic paradigm’, characterised by ‘typological thinking’. Here:
variability is in effect treated as occurring essentially among sociocultural entities at whatever more micro or more macro-level they may be distinguished – whether, say, as tribes, local communities, ethnic groupings, social classes or even total, national or state, societies. Such entities are represented as more or less coherent and distinctive ‘wholes’ that in themselves are to be taken as the key units of analysis. (p. 19, emphasis in original)
This holistic paradigm includes a great deal of past work in anthropology and sociology. Goldthorpe specifically mentions the following: Spencer, Tylor, Le Play, Durkheim, Hobhouse, Ginsberg, Radcliffe-Brown, Kroeber, Lowie, Murdock, Parsons, Hoselitz, Kerr and the community studies tradition (whether the work of Warner or that of Young and Willmott). It would no doubt also take in much recent sociological work. Interestingly, Malinowski is excepted from this criticism because of his insistence on the difference between norms and what is done in practice, and his tracing of functions to the needs of individuals (pp. 26–29). At the same time, though, his work clearly does not fall into the category of population science, so this rather crude dichotomy does not cover all relevant variation.
Goldthorpe’s objection to the holistic paradigm is not just that it ignores the degree of variability to be found at the individual level, and therefore within categories as well as between them, but also that it fails to assign causal primacy to the actions of individuals. The focus, instead, is typically on normative or material constraints which are assumed to standardise the behaviour of individuals. He sees it as thereby avoiding any conception of actors making decisions. His argument is that while the goals and decisions of individuals are highly variable, at the level of the population they will generate relatively stable regularities, and it is these that sociology must explain, albeit through the identification of causal mechanisms consisting of individual decision making.
While Goldthorpe goes into considerable detail about the difference between a population science and a typological approach, his discussion begs some key questions. He claims that the holistic approach has failed to produce convincing explanations, yet he acknowledges that his version of sociology has also been much less successful in this respect than in documenting regularities (p. 16). And while he is right to highlight the issue of heterogeneity within categories, he is brusquely dismissive of Ragin’s (2000) development of fuzzy set analysis as a means of dealing with this issue. There are also questions about why all explanations must be at the level of individual decision making. He quotes Hollis (1977: 21) to the effect that ‘rational action is its own explanation’, but this is to treat rationality as if it were an intrinsic feature of actions, whereas it is necessarily a product of evaluation, on the basis of some conception of what would be rational in the relevant circumstances. It seems to me that the sociologist’s unavoidable reliance upon simplifying models of behaviour is underplayed in Goldthorpe’s rejection of typological approaches.
Another key distinction that he employs to specify the character of population science, and to justify it, is between sociological and historical work. Where the former is concerned with explaining regularities within a population, the latter focuses on the development of particular institutions or practices, and the explanation of particular events. Of course he recognises the legitimacy and value of historical analysis, but he argues that much work done under the auspices of sociology and some other social science disciplines is actually historical in character (see pp. 49–50).
This is perhaps the most significant distinction that Goldthorpe makes. It corresponds to that between sociology as a generalising science and historical inquiry as idiographic. But there is a fundamental problem here because his concept of population itself seems to be idiographic in character. Goldthorpe draws a parallel with Maxwell’s ‘statistical physics’ (p. 10), but Maxwell was studying populations of gas molecules that he could be fairly confident were representative of such molecules generally. By contrast, the main focus of social mobility research, to use Goldthorpe’s main field of enquiry as an example, is on what happened during some particular time period in one or another national population, or what trends or differences can be found across some set of national populations. This is quite clearly an idiographic focus: the aim is not to discover or explain probabilistic patterns that are universal across human populations. Furthermore, while social mobility research has an interest in the causal mechanisms that generate the local patterns it studies, its main product is not theories about those mechanisms, analogous to Maxwell’s laws, but accounts of particular trends, and the reasons why they have occurred.
At one point, Goldthorpe mentions Mackie’s (1980) account of causation, which makes the point that in producing any explanation there must be selection from a wide range of causal factors. The key issue is the criteria to be employed, and there seems to be a significant difference in this respect between generalizing and idiographic sciences. In the former, the criteria are internal to a discipline, being designed to reflect the distinctive character of the natural kinds it investigates, whereas in idiographic sciences the criteria are, as Weber pointed out, necessarily related to prevailing social values. While Goldthorpe appeals to Weber – in a lively critique of Burawoy’s ‘public sociology’ – he does not take in this key aspect of Weber’s methodological views explicitly. Yet research on social mobility clearly relies on such a framework, even if there is a tendency for this to be forgotten (Hammersley, 2014: ch. 5).
Goldthorpe places great emphasis on scientific rigour, in particular the employment of probabilistic sampling techniques, rigorous measurement supported by checks on reliability and validity, and the use of multivariate statistical analysis – which he sees as concerned, above all, with ensuring that the relationships among variables have been correctly identified. However, in my view he does not acknowledge sufficiently the severity of the problems involved here: the problem of determining what are relevant variables that need to be included in the analysis if spurious relationships are to be avoided and genuine ones not to be overlooked; that of obtaining data about all of these variables, plus having sufficient data in all of the cells if interaction effects are to be explored; and the problem of operationalising concepts in ways that do not distort their meaning.
While there are plenty of questions to be asked about the idea that sociology should be a population science, this is an interesting and thought-provoking book. It addresses genuine and difficult problems – not least about the nature of sociology as a discipline, at a time when it is more fragmented than ever. Unfortunately, Goldthorpe’s approach is not entirely convincing even in its own terms, for the reasons indicated. Nor is it ever likely to take over the discipline, despite the thriving and fruitful research tradition from which it springs.