The Great British Class Survey was launched on 3 April 2013 and quickly became one of the most popular stories on the BBC website. Mike Savage giveshis account of how the project was organised and reflects on whether this model has wider potential take up for social science research, or whether it is likely to be an idiosyncratic exception to normal patterns.
The Great British Class Survey is an astonishing example of how social science research can have dramatic impact in the digital age. The headlines are remarkable. The simultaneous online publication of an academic paper in Sociology, the presentation at a section plenary at the British Sociological Association Annual Conference, and a BBC news campaign, all on 3 April 2013, saw the story take off to astonishing public interest. There were over 6.9 million unique visits to the story on the BBC web pages, which became its most popular in 2013. There was a huge volume of exchanges on social networking sites, notably Twitter and Facebook. There was a dramatic rise in people buying tickets to theatre, presumably to improve their cultural capital. The story was taken up in all the print media, and indeed reported all over the world, where it has also enjoyed huge popularity. The story in the New York Times was their most emailed ‘world news’ story of the year.
The project originated as an experiment developed by the BBC’s Lab UK which was opened in 2009. Lab UK’s function is to create mass participation online experiments with two express outcomes – peer-reviewed science and popular content for online and broadcast. It has collaborated with numerous academics, mostly natural and behavioural scientists, to develop interactive web based experiments which provide large scale data. The highest profile of these has probably been The Great Personality Test, fronted by Robert Winston, which involved collaboration with social psychologists at Cambridge. The model is that the academics devise the experiment, in the form of a web questionnaire, which Lab UK then hosts on their website and delivers the data to the academics to analyse. The academics are expected to get ethical approval for the project through their university committee structure, and the BBC provides no financial resources for the academic teams (since this is deemed to be potential conflict of interest). Lab UK specialises in developing interactives to encourage respondents to take part in the experiment by giving them immediate ‘feedback’ on their responses. The academic results are expected to be peer reviewed, and then publicised (in appropriately ‘popular’ format) on their web pages, and possibly through other media and documentary programmes (as with The Big Personality Test).
The GBCS differed from this model in one crucial respect: the involvement of BBC Current Affairs, through Philip Trippenbach who was seconded to Lab UK to work up interests in class. It was clearly important at this outline stage that there was a lively sociological debate on class which persuaded him that the topic was of ‘scientific’ interest. There was genuine interest from BBC colleagues in Bourdieu’s thinking and the cross fertilisation of cultural analysis with the study of class which had taken place in British sociology. It was through his networks that he contacted me and Fiona Devine, in 2010. His own interesting account of the GBCS is attached here.
Philip Trippenbach successfully made the pitch for internal resources to the BBC, and once these were agreed, along with Richard Cable at Lab UK, was the main point of contact as Fiona and I developed the questionnaire. We academics were largely left to ourselves, mainly using questions from previous research projects we had been involved with, notably Cultural Capital and Social Exclusion. However, in testing the questionnaire, input from Lab UK was important in suggesting formatting of questions. Phillip worked with BBC Lab UK to develop the web platform the questionnaire.
Following previous BBC Lab UK models, the experiment was launched in a high profile campaign devoted to raising public interest. A slot on the BBC’s One Show and numerous local and national radio programmes took place on 26th January 2011. These were also linked to two BBC documentaries on class, fronted by Andrew Neill, which also encouraged viewers to do the test. Initial take up was good with over 100,000 respondents within three days of the launch, though as expected it tailed off afterwards.
The initial model which we agreed with the BBC was based on the science model, in which our research would figure in a documentary to be launched in summer 2011. The BBC also commissioned an American web design company to produce attractive visualisations and interactive for the findings. However, this initiative failed. In retrospect, in accordance with documentary time frames we were expected to work to an impossibly tight timetable of about 14 weeks to deliver the findings. This deadline looked even more implausible as it took several weeks to download the data from Lab UK into an accessible format. Because the BBC did not put any resources into the academic research, we also had no dedicated research staff who could be deployed, though Niall Cunningham, recently appointed as a Research Fellow at the ESRC Centre for Research on Socio Cultural Change (CRESC) public spiritedly put a great deal of effort into making the data usable through SPSS and Excel software.
It soon became clear that major problems would prevent any rapid analysis of the data. The GBCS sample proved to be highly skewed towards the educated middle classes. We argued that without a nationally representative survey as well, the GBCS would be very hard to use to derive a nationally representative picture of the British class system. Lab UK therefore agreed to commission a nationally representative survey to buttress the web survey. Furthermore, in two critical areas we asked respondents to write in their occupations and the name of the university they attended. This produced very messy data with numerous different categories being used.
Although we attempted to provide headline findings from the data in the months after the release, and although we were able to provide some telling information about inequality in Britain, we did not at this point have a clear story about class, other than some telling data on a very wealthy elite. Filming for the documentary did begin, but it was called off in short order, given the recognition both from ourselves and the BBC that the analysis was not yet developed enough to justify a feature.
In summer 2011, the GBCS might not have been seen as a failure, exactly, but it certainly had not produced the rapid, popular, findings which were hoped for. Lab UK intervened at this point to suggest a more scaled down timetable, and the interests of the documentary makers was diverted elsewhere. Lab UK supported one vital development, an agreement to archive the data at the UK Data Archive at Essex, in return for the UKDA agreeing to clean the occupational and university level data to allow systematic analysis. This proved a vital precondition to further analysis.
During the rest of 2011 and 2012 academic research on the project proceeded in fits and starts, in between other commitments. Team members such as Niall Cunningham, Yaojun Li, Andrew Miles, Sam Friedman and Mark Taylor explored aspects of the data. But fundamental problems remained, especially in allowing the skewed web based GBCS to effectively interface with the nationally representative sample. However, during this period, the building blocks for the published analysis were put in place: the use of multiple correspondence analysis (by the leading international authorities Johs Hjellbrekke and Brigitte Le Roux), our use of latent class analysis (led by Yaojun Li and Mark Taylor) to examine the underlying patterns across our measures of economic, social and cultural capital. Most importantly of all, Mark Taylor’s idea of weighting the GBCS infinitesimally to allow it to be linked to the GfK nationally representative data set finally allowed us a means of linking the data sets. [For more on the theory and methods of the GBCS see here.]
Late in the summer of 2012, we showed the BBC Lab UK our notes from our latent class analysis. Their interest immediately perked up, presumably because they saw these as crisp findings which could be readily disseminated. At this point they decided on a new strategy. Rather than using a documentary format to publicise the findings as they had originally planned, they decided to promote it as a news story to coincide with the publication of the findings in a major journal. We rapidly wrote up the paper, and submitted it first to Nature who indicated it was not suitable for their pages, before submitting it to Sociology. It was made clear that rigorous peer review was vital for the strategy to succeed, though we were also hopeful that it might be published at the same time that we were slated to give a section plenary address at the British Sociological Association. Once the paper was accepted for publication (subject to revisions), LabUK developed its own campaign to promote the findings on publications, which famously – notoriously – involved the ‘class calculator’. This was their own creation, developed in collaboration with web designers Applied Works. We gave advice on its construction, but this was a BBC product.
This has been a long narrative, but I wish to end with one major point. The launch of the GBCS results may appear to have been a slick, well organised, exercise. In some respects it was. But it was also an improvised operation after a series of problems, glitches, and indeed, failures. Its ultimate effectiveness depended on three major qualities. Firstly, the willingness of both the academics and different groups within the BBC to change tack at short notice, to improvise, and for other parties (notably the UK Data Archive) to come on board and take a strategic interest and devote resources at short notice to the project. Secondly, although ultimately successful because of its capacity to communicate through digital modes, the academic team fundamentally operated in a very old fashioned way. Only one (very partially) of the academic team’s nine members was resourced, or ‘line managed’, to work on the BBC project. Instead, their intellectual curiosity and interest was the spur, and the work was often done in their spare time after the day jobs. Thirdly, and finally, time is needed to do good academic research, and the time schedules of social scientists and media production are difficult to synchronise. Both sides learnt this lesson the hard way, but finally found a way of reconciling their time horizons.
Is this a replicable model for the social sciences to learn from? Shortly after the launch of the GBCS results, we found out that BBC Lab UK is going to be closed – hopefully temporarily – while the BBC is working on a new way to present its science content. There are now 2.5 million responses across the different Lab UK experiments, and because many people do several experiments, there is also remarkable potential for data linkage, relating questions of social class to personality, stress, money, and such like. It is therefore possible that the GBCS, far from being a standard bearer of a new kind of digital social science research, might end up as a celebrated, but isolated, example of a ‘path not taken’. Hopefully the BBC’s new science content will include BBC Lab UK. Time will tell.
Note: This article gives the views of the author, and not the position of the Impact of Social Science blog, nor of the London School of Economics.
Mike Savage is Professor of Sociology at the LSE. He was founding Director of CRESC – the ESRC Centre for Research on Socio-Cultural Change.