Landmark Paper: ‘Can we at last say goodbye to social class?’, Sarah O’Brien and Rosemary Ford
17 March 2016

You can read the Landmark Paper here: ‘Can we at last say goodbye to social class?’, Sarah O’Brien and Rosemary Ford

The first Landmark Paper selected in the MRS’ 70^th year focussed on the contribution to marketing research, and to MRS, of the leading marketing scientist, Andrew Ehrenberg. The second choice reflects a schema that has been the traditional population classification method used in surveys from before, and throughout, the life of MRS.

‘Social class in the 21^st century’: a challenge to market research?

In the latest issue of IJMR (58/2), you’ll find a review I’ve written of ‘Social Class in the 21^st Century’, by Mike Savage (LSE) and a team of other leading UK social scientists. This book is based primarily on the findings from the Great British Class Survey (GBCS), hosted on the BBC website that attracted 325,000 responses, but due to the bias in this response towards the higher end of the social class pyramid, it was supported by a nationally representative quota sample survey using the same 20 minute questionnaire, plus 50 depth interviews to provide more individual level detail. The findings have received a lot of media attention, and there’s also a paper published in Sociology, ‘A New Model of Social Class?’ (Savage et al, Vol. 47 (2), 2013). Mike delivered an IJMR Lecture based on the book at the MRS on March 1^st, to a full house and there’s coverage of this in my Editorial in 58/3 (May).

In the book, the authors include a call to action: ‘This involves placing the classification process under the political spotlight. It involves the social sciences getting their hands dirty and seeking to wrest intellectual authority from market researchers, consultants, journalists and commentators’, the authors contending that: ‘League tables, market profiling and categorisation lie at the heart of today’s knowing capitalism’.The highlighting of market researchers is by me, but it made me wonder what attention, if any, the market research sector, is paying towards changes in society and how they might, or should, be reflected in the way we classify people from a social grade perspective in surveys. We seem wedded to the traditional method based on head of household occupation, which includes the development of an algorithm to provide a link to the UK Census, first applied to the 2001 Census (a previous IJMR Landmark Paper ‘Social Grading and the Census’, Erhard Meier and Corrine Moy, first published in IJMR Vol.46/2).

Why measure social class?

As Helen Lambert (GfK) described in her presentation in November 2013 at an MRS Census and Geodemographics Group seminar, the four reasons why social grade is important are: create a common currency; differentiation of consumption patterns and attitudes; standard currency for sample and weighting; profiling geographic areas and audiences. These are vitally important factors, but it doesn’t necessarily mean that the tried and tested method based on occupation should not be revisited to see if a different method might be more in keeping with changes in society. As stated in a project conducted by Kantar Media for Channel 4 in 2014: ‘So we think that it’s time for a change! Although there are problems, ABC1 isn’t dead – it’s still a hugely important trading audience and this won’t change in the foreseeable future. However, we can make it better! The arrival of the smartwatch hasn’t made time less important… but it has made it more modern and more relevant to 2014. So we want a Smart ABC1 for the industry’. This was reflecting a concern that the way wealth was distributed today was not really being reflected in the traditional socio-economic groupings. More on this project later.The aim of Savage et al, was to see if they could: ‘identify distinctive social classes who share common lifestyles, identities, social networks and political orientations as well as levels of income and wealth’. Note that there is no mention of occupations, nor is there in their method as we know them. The outcome was new seven groupings.

The Landmark Paper: an alternative to the status quo?

You can read about the evolution of this new social grade schema in Savage’s book, but I have a copy of the 2^nd Edition of ‘Occupations Groupings’, published by the MRS in 1991 (with a Forward by me, when MRS Chairman) based on the familiar A, B, C1, C2, D, E classification. A look through back issues of IJMR/JMRS from 1985 onwards showed that this topic has been largely absent in content, although that of course doesn’t mean that there was no debate elsewhere over this time period, but I would have expected more challenges to the status quo over the past 30 years, especially as social grade, or socio-economic groupings based on the occupation of the head-of-household, sits at the heart of survey participant segmentation. Going back further, the 1979 paper introducing the application of geodemographics in market research (another previous Landmark Paper), suggested that a new era of classification was dawning, but since then it appears to have primarily provided a parallel option, and Savage pays geodems little attention in his book. Two of the five papers I found were on the Census project mentioned earlier; one was on the potential for developing an international classification (a different ongoing challenge); one on the wider field of demographics, and, just one that challenged the established method based on occupation. That is the one I’ve selected as the latest Landmark Paper, however this was published in JMRS back in July 1988 (Vol.30 No.3). It had won that year’s Best Paper award at the annual MRS conference.

The authors, both of whom attended Mike Savage’s recent IJMR Lecture, describe and discuss a research project commissioned by Granada Television in 1987, conducted by BMRB. It’s interesting that TV companies appear to be at the forefront, then and more recently, in looking for alternatives. The Abstract sounds so up to date, describing the ‘so-called North/South divide’, the aim of the project being to ‘explore all aspects of people’s lifestyles and income both in the North West and in the country as a whole’. Very like the objective behind the GBCS described by Savage. The authors also contend that the 1980s was a period ‘marked by an urgent quest for newer, more pertinent market discriminators rather than the conventional ones of age, social class or sex’. However, I would contend this quest then tended to peter out.

Interestingly, the authors built a level of replicability and stability into their search for new discriminatory variables by re-contacting participants. Not surprisingly, like Savage, they queried whether social class, as then defined, was relevant in a more egalitarian society. This is the 1980s, a period of great economic change in Britain under Margaret Thatcher that decimated manufacturing and began the real demise of coal mining, the last deep mine closing just before Christmas of last year. The rise of an economy based on services began back then. They cite a BARB finding from 1987 that nearly a third of the BARB panel had changed their social class over the previous 12 months. Some of this huge difference was eventually explained by miss-coding, but there was also issues to do with classifying the new types of jobs then emerging (one of Savage’s new groups is personified by the call centre agent). The authors quite rightly state that core criteria for any classification schema are that, firstly, it provides real discrimination matching that provided by social class; secondly, it must be replicable and stable – hence the follow-up phase of research. Their new method focussed on lifestage (6 groupings); lifestyle (5 clusters); disposable income; acquisition (26 listed items covering durables, financial products, holidays, home ownership, home improvements etc). They then selected four measures of discrimination for the analysis: maximum penetration; range of penetration; Lorenz based curve; ‘mail-shot’ (similar to the Lorenz curve). Social class and lifestage emerge as powerful discriminators, but not lifestyle.

Feet of clay?

However, and very importantly in the quest for a replacement for social class, they found difficulties in replicating social class in the follow-up stage. A large section of the paper discusses the reasons for this. Whilst they saw interviewer training as part of the problem, there was also the degree to which interviewers could be expected to probe answers, especially in sensitive situations (but today, a large proportion of commercial research is conducted online without interviewers being present). The authors’ state in their Conclusions: ‘In many ways our findings match those reported by BARB, and they raise serious question-marks about the industry’s unquestioning use of Social Class. It seems that a major industry-funded investigation would not be out of place (perhaps via the Market Research Development Fund) and that the pursuit of alternatives must continue….We feel we cannot afford to be complacent about the use of Social Class and whilst it is not possible to say goodbye to it as yet, the industry must face up to the very real concerns surrounding it’. They argue that lifestage provided the best course to pursue.

On the face of it, this was quite a devastating finding – the core measure of social class used by the industry was shown to be both unstable, and with replication issues. I’m not aware that the MRDF ever tackled this issue, and of course that body is long gone, or that this challenge was ever subsequently addressed by others, which seems worrying. The work by Savage et al does provide a new look at this topic, but unlike the work by O’Brien and Ford, the book does not address the key criteria of replicability and stability that was set for the Granada research project. Also, the full GBCS survey took 20 minutes to complete and therefore impractical for general survey classification purposes, but this also applied to the Granada project, and there is no evidence as to whether the shortened version of the GBSC, still there if you want to complete it, is sufficiently discriminatory, stable or replicable. See for yourself.

Where we are now

Of course there was the development of the National Statistics Socio-Economic Classification (NS-SEC), first applied to the 2001Census, which was empirically validated, but is still based primarily on occupations. The paper mentioned earlier by Meier and Moy described how the algorithm was developed to link to this. I did some research to find out whether Savage’s view that market research was ahead of the game was justified (well, I asked a few research methodologists what was going on today!). Through this I discovered the Channel 4/Kantar Media project, which interestingly cites Mike Savage as an advisor – great to see collaboration between the social and commercial sectors of research in this way. I also discovered the TGI WHY Code that creates nine segments across the whole sample using a mass of data collected in this survey, whereas Channel 4’s aim was to re-structure what are the traditional ABC1 groupings. A common feature to both is the use of Pierre Bourdieu’s concepts of social and economic capital, as the C4 project was based on the TGI WHY Code research. Whilst Bourdieu’s work has also heavily influenced Savage’s research, Savage also introduces the concept of cultural capital, missing from the two ‘commercial’ methods. However, what is clear from all three methods is that the data used to drive these segmentations are infinitely more detailed than that collected to derive traditional S-E groupings.

This raises three issues. Firstly, the difficulties of administering a survey where the classification questions are almost a survey on their own; secondly, is there an assumption that the survey data itself provides the basis for a classification, and thirdly, could these form the basis for a universal method in terms of replicability and stability? Perhaps the age of a generic method are ending, to be replaced by a host of customised methods to meet differing needs and provide competitive advantage, as happened with geodems. I have a feeling, however, that despite the reservations about a socio-economic based schema it will survive for the reasons that make the others rather problematic. However, I still return to the problems of replicability and stability raised by O’Brien and Ford, especially in an era where interviewers are becoming an endangered species. Has this been re-visited, or swept under the carpet?

We are where we are, and maybe the traditional socio-economic based schema still remains the best bet if it remains easy to administer in providing an accurate picture, whatever is meant by ‘accurate’, and provides a valid method for discriminating the attitudes and consumption patterns of the population, but in a fast changing world, I would have thought that this important tool needs checking-out from time to time as recommended by O’Brien & Ford nearly 30 years ago. Joy Reynolds, then Chair of the MRS Field Committee, in her Introduction to the 2^nd Edition of the MRS ‘Occupational Groupings’, mentioned before concludes with an interesting comment: ‘The concept of classifying people according to the head of the household’s occupation seems to be an anachronism in the 1990s and is being widely discussed in the market research world, but until something better takes its place, we hope that the inclusion of this section will be useful to all who dip into this book’. The current 7th edition does take a slightly different three step approach: determine the household the participant lives in; determine who the chief income earner is in that household; grade the job that person does. However, I think we are still waiting for that more radical solution as argued for for by O’Brien and Ford. Will the groupings described by Savage prove the basis for a practical, reliable alternative, the ‘something better’ anticipated by Reynolds back in 1991? As researchers, whether ‘market’ or ‘social’, our business is to be at the forefront in understanding society, identifying changes and tracking trends, but are the traditional methods we use to derive class out of touch with societal change? 

Read the Landmark Paper: ‘Can we at last say goodbye to social class?’, Sarah O’Brien and Rosemary Ford