I’ve (literally) just finished* my evidence review for the JRF on poverty and social networks that I’ve blogged about here, here and here and sent it off for their feedback (which I’ve just received some of, all generally quite positive *phew*). Because it’s still drafty, I’m not going to release it to the world for a while yet, but I did want to blog about the methodological thoughts it’s led me to have, and this links to the title of the post. The phrase “it’s not what you know but who you know” really demonstrates the “folk” interest in the links between social networks and socio-economic mobility. However, now I’ve read a load of this evidence, methodologically, it actually seems to be more about how you know about these social networks. This is for two specific reasons: methodology, methods and disciplinary contexts; and bigger epistemological experiences about experience and how this is “collected” by the research process.
On the first point, the review was the first time in a while to read a good chunk of North American social science. I’ve read about the “methods wars” between qualitative and quantitative social science in America, and the feeling that the latter had won; I’d also listened to American delegates at the International Interpretive Policy Analysis conference bemoan the dominance of the rational actor model and quantitative methods. But I’d really not realised how dominant quants where. Successive were based on big datasets, with endless methodology sections explaining how survey instruments were used, with big scary equations that, as a qualitative researcher, I didn’t have a clue about. What struck me about them was they usually ended with, firstly, a lot of uncertainty about their findings and secondly a lot of unanswered questions. So, for example, some regression models would support theories, others would contradict them; when you controlled for different factors the overall conclusion was “meh”. And the unanswered questions didn’t help – invariably there would be the conclusion “but we cannot understand causation here”.
Which brings me to one of my favourite methodology books I’ve read for a long time, Mike Savage’s Identities and Social Change in Britain since 1940. It’s essentially a history of social science methods in the UK from 1940 and it’s the historical approach that I love. To go back to the title of this whole blog – urbanity and history – that I chose three years ago, it’s always struck me how social scientists don’t get history, or even have much of a sensitivity to time. Savage sort of makes this point – the technocratic sociology that emerged in the 1950s was a study of the now. I often find it funny – and I’m guilty of doing this myself – how the strictures of social science writing mean that people will present evidence that is decades old in such a way as to suggest its contemporary, without any reflection on how the social world might have changed since then, and reflexively, how that knowledge has since changed the world that is now the object of study. The presumption is the knowledge is scientific and presented in the linear, “standing on the shoulder of giants” way. What I love about Savage’s book is it highlights the social context where the methods of modern British sociology/social science, particularly the survey and semi-structured interview, emerged from.
So, why does this all matter, well for me it brings us onto the second bigger epistemological question. The best way to talk about this is how it emerged in the evidence review itself. In terms of that folk idea of “it’s not what you know, but who you know” is the idea that if poorer people know wealthier people then they will do better in life. It also leads to the more negative idea, which I’m deeply uncomfortable with for its use in “underclass” discourses, of “cultures” of poverty – that experiencing poverty around others in the same situation lessens your chances of leaving poverty because you behave like them and this excludes you from wider society and opportunities. A lot of this sort of theorisation lies behind the “neighbourhood effects” literature I read for the evidence review; I discussed the problems with this in this blog post. However, I want to go back to a striking theme running through the neighbourhood effects literature: basically, most quantitative studies that try and find a neighbourhood effect either find a very small impact, or none at all, or the data is too messy to form consistent conclusions. Qualitative studies tend to find there is an impact, people recognise it when they work or live in the neighbourhoods. This is a contrast discussed in a very well-title article by Atkinson and Kintrea – Opportunity and despair it’s all in there (£) – and in the critique of neighbourhood effects by Tom Slater (who critiques the distance of the quants researchers from their research subjects).
In this quants study of neighbourhood effects in Scotland, van Ham and Manley suggest that it might be down to the geography that we measure neighbourhood effects on. Usually we have to do statistical analysis at the level of census super outputs areas or datazones, or similar administrative boundaries that have quite large populations (1000+). Like any “contagion”, if neighbourhood effects are diluted among a large population then they will have less of an impact. Van Ham and Manley suggest that we might need geographies of just a few hundred people to find a neighbourhood effect. However, this leads me to question (prompted by a point someone made at a Poverty Alliance panel in Glasgow on Friday 28 February) whether this then is a neighbourhood effect – surely if the geography becomes really small then we’re talking about a sociological or socio-psychological impact; residence is just by-the-by as the social network could be anywhere? All points raised in the quants neighbourhood effects literature.
However, I think we are talking about a bigger epistemological point here that we need to recognise fully and debate and play with. For example, this study for the Joseph Rowntree Foundation found no quantitative neighbourhood effect on young people’s employment opportunities from postcode discrimination, a neighbourhood effect, whereas this qualitative study also for the JRF, found that young people who live in deprived neighbourhoods that are distant from local labour markets and geographically isolated have reduced geographical horizons for their job search strategies, a neighbourhood effect. These different findings, I think, can only be explained by differences is how we know, no what we know. These are different experiences of young people in deprived neighbourhoods understood in different ways.
As a qualitative researcher I use my research to illuminate how socio-structural factors – particularly socio-economic inequality and status – have an impact on people’s lives. However, a problem for me is that a broader focus on “lived experience” of conditions such as poverty can easily lapse into the personal and focus on individual deficiencies or strengths, effectively blaming the poor, rather than emphasising the socio-structural forces that limit people’s opportunities and psychological reactions in any given context. This is a criticism that is levelled quite often at neighbourhood effects researchers, but most I know put the socio-structural first and want to understand neighbourhood effects as something that makes the situation that much worse. At its most problematic though, this individualising of experience and “witnessing” of negative experiences does lead into the sort of nonsense the Centre for Social Justice come up with – the production of ignorance – and deflects us from an emphasis on socio-structural causes. This is very apparent with the recent debate in the UK on changing the definition of child poverty and the conflation of cause and impact in the proposed measure (as well as initially coming up with something that is unmeasurable). I'm also always left with a nagging feeling that without the devolution of taxation and welfare matters to the Scottish Parliament and Government, policy-makers up here falls way to easily onto person-blaming, behavioural policy answers in areas like public health and social work, rather than the more obvious fact that if a few people had a lot less income and wealth and a lot of people had a little more, then a lot of problems we have would be far less severe. We get away with this victim-blaming because we can frame it in our lovely, fluffly, Scottish social democratic discourse.
And I’ve just realised I’ve written myself into a corner. What am I saying? It seems that a lot of methodological choices in the social sciences are actually deeply circumscribed by the disciplinary and cultural context you find yourself in. If I was in the US my career would’ve probably have done nowhere because of my aversion and inability to wrangle big datasets and produce equations that are penis extensions. However, in the UK, with a methodological pluralism that welcomes qualitative research at the ethnographic end, I need to be careful that my research does highlight the structural, and doesn’t presume that cultural changes are all that are needed in society to produce more positive outcomes.
* I finished the review on Thursday 27th, started writing this on Friday 28th and then got way-laid by a gin and tonic. I have now finished writing it.