Earlier this month the Great British Class Survey launched, attracting widespread attention and thrusting sociological arguments about class into the media spotlight. Following on from earlier posts on the LSE Politics Blog, Mike Savage and Fiona Devine clarify the understanding of class on which the project was predicated and address some of the criticisms which have been aimed at it.
Now that the dust from the media blitz surrounding the release of the initial findings of the GBCS has settled, the time for critical scrutiny of the sociological arguments is upon us. Immediate reactions focusing on the utility (or otherwise) of the BBC’s class calculator, are increasingly giving way to engagement with the arguments we put forward in our underpinning article published in Sociology. This is how it should be, and this blog offers initial reactions to important critical blogs by Dave Griffiths and Paul Lambert; Danny Dorling, and Colin Mills.
A fuller response to specific points will be posted as a CRESC Working Paper in the near future. Our aim here is to set the record straight about how we are using class analysis. A critical engagement with the National Statistics Socio-Economic classification (NS-SEC) lay at the heart of our paper. As Griffiths and Lambert recognise, we were offering a different kind of approach to class analysis to the NS-SEC’s deductive model of class, derived from measures of employment relations. Instead, our conceptualisation of class falls squarely within the classic sociological tradition of exploring class formation – the development of distinctive entities which straddle economic, social, cultural and political arenas.
This is no minor point. It follows that we are not claiming that we have the best tool to unravel inequality in all its varied forms. Clearly economists – indeed social scientists of all hues – have more to say about the nature and extent of economic inequality than we can on the basis of the GBCS data. It was not our intention to argue that our measures of class offered the best ‘fit’ to contemporary patterns of inequality, or (for instance) to argue that ‘precarity’ in the broader sense is confined to our ‘precariat’ class. Danny Dorling’s work pursues somewhat different (though by no means contradictory) concerns to ours. His critical remarks do not engage with our specific reflections on what kinds of social classes – in the form of bounded social groups – can be discerned across our measures of economic, social and cultural capital. Indeed Dorling has not deployed the concept of class in much of his work, other than rhetorically.
We make no bones about emphasising that we are reviving classic sociological theorising on stratification. Thus, although we certainly draw on Bourdieu’s influential ‘capital, assets and resources’ approach, we should not be pigeonholed simply as his following dogmatically in his footsteps. In fact, the authors of our paper vary in their degree of commitment to Bourdieu’s thought. There are numerous other theoretical warrants for our concerns. We might rehearse Weber’s distinction between ‘economic’ and ‘social’ class, or Marx’s between ‘class in itself’ and ‘class for itself’, or Giddens’ claims about ‘class structuration’. In all these cases, class formation as a social, cultural and political process cannot be reduced to the economic dimension of class alone. We might go further, and reiterate the Marxist historian E .P. Thompson’s insistence that the ‘making of the English working class’ was not a mechanical response to the economic change associated with industrialisation, but involved the complex mobilisation of political, economic and cultural processes.
This kind of sociological thinking remains important today. We can see it take new forms in complexity theory, with its concern with ‘assemblage’, or critical realism with its interest in ‘emergence’, all of which underscore a way of conceptualising classes which recognises them as relational entities with emergent properties which cannot be reduced to the sum of their parts. We are also informed by feminist arguments about the mutual co-constitution of class and gender relationships, for instance through forms of ‘dis-identification’. Our research seeks to take these sociological perspectives and apply them to contemporary class relationships. Our conceptualisation of class does not reduce it to a purely economic phenomenon, or indeed any kind of uni-dimensional measure of the kind that Mills sees as preferable, but seeks to empirically assess whether classes might be formed as entities which straddle our measures of economic, cultural, and social capital. This exercise precisely brings out the ‘relationality’ of class which Dorling emphasises. Given these concerns, our analytical strategy of inductively examining underlying relationships between discrete economic, social and cultural variables (using latent class analysis) seems fitting.
Hopefully these points are helpful as a response particularly to Mills, who is unclear how we are using the concept of class in our study. His own perspective, by contrast, deploys a deductive operationalisation of class, derived from Goldthorpe’s highly influential framework, which defines it in unitary terms, as a measure of employment relations alone. Thus, we are criticised because our measures of class become a mish-mash of different factors. This is of course true, insofar as age, geographical location and so forth are likely to affect our measures of capital, and hence by definition our model of class, but this is only a problem if one accepts Mills’ analytical starting point, which we see as heavily criticised in much recent sociological research.
Recent developments in what is sometimes called ‘cultural class analysis’ (a vitally important current of work which none of our interlocutors has referred to) have shown how classes are articulated with processes of gender, as well as race and ethnicity, age, and such like, and that the cultural dimensions of class inequality are profound. Given these interventions, to try to artefactually differentiate and hold apart the relative importance of class, gender, age, ethnicity, sexuality and so forth is to miss the fundamental weaving together of these different processes into composite entities which we seek to reveal. To repeat, our aim is to reveal classes not as ‘employment aggregates’ (using Rosemary Crompton’s terms), but as distinctive assemblages which crystallise particular combinations of economic, social and cultural capital.
Let us now turn to the technical concerns that have been raised about our methods of analysis. In fact, despite Mills’ typically trenchant remarks, there is a good deal of agreement that, given the skew of the web survey, we had to find an alternative means of elaborating a satisfactory class model. Although we certainly concur with Mills (and Griffiths and Lambert) that the sample size of our nationally representative sample would ideally be larger, given concerns about the skew of the web survey, he surely would not prefer us to have adopted an alternative strategy? Starting from the data we had at our disposal, our approach seems the best we could usefully have taken.
Mills, as well as Griffiths and Lambert, pose the issue of whether our seven latent classes are simply an artefact of the relatively small size of the representative survey. To be sure, it will be interesting to see if comparable analyses on other data sets show different patterns (whilst noting that few if any data sets have the appropriate questions on cultural issues to allow this issue to be properly scrutinised). But a key test of our model has to be answered in terms of whether the seven classes we unravel make ‘sociological’ sense. If the seven classes which are revealed by our latent class analysis appear to be the artefact of the data and modes of measurements themselves, and it is not possible to interpret them as recognisable social entities, then this is a major problem. Much of our work involved seeking to make sociological sense of the seven latent classes, seeing them as the product of forces of economic polarisation and of fragmentation in the middle reaches of the class structure. The delineation of a small ‘elite’ class is consistent with other evidence about the accentuation of social inequality, as is our suggestion that we are seeing fragmentation within the middle reaches.
There is a danger, in reflecting on the arguments of our interlocutors, that the strength (or otherwise) of the GBCS is held to lie in the model we develop on the basis of the small GfK representative survey. To be sure, this is a vital underpinning of our research. However, it is also the remarkable detail which is provided in the web survey, and which we have linked to the national data, which is of fundamental importance in allowing us to develop a fuller understanding of the micro-processes involved in class closure, especially at the higher reaches of the class structure. Although we only had space for a small amount of this analysis in our paper, it is interesting that the maps, occupational profiles, and comments on how university profiles are affiliated to class membership has not attracted comment. Clearly, if the GBCS is judged only by the standards of large scale nationally representative surveys, then it can be found wanting. However, it is actually a much richer source whose potential we will further exploit in future publications. In these days of myriad digital data, surely we need to be open to new sources for social scientific research? Who would be so complacent as to think otherwise?
Note: This article gives the views of the author, and not the position of the British Politics and Policy blog, nor of the London School of Economics. Please read our comments policy before posting.
Mike Savage is Professor of Sociology at the LSE. He was founding Director of CRESC – the ESRC Centre for Research on Socio-Cultural Change.
Fiona Devine is Professor of Sociology and Head of the School of Social Sciences at the University of Manchester.