What was the question again?

A long time ago I did a first degree in the Natural Sciences, and in the 3rd year I did a whole year of History and Philosophy of Science. I'd argue that this would be a useful exercise for everyone at university, as a bridge between humanities and science, as an education in how human civilisation developed over the millennia, and how we conceptualise knowledge and reality. On this latter point, I always feel thoroughly depressed when I see social research methods text reproduce a hackneyed 'positivism' v 'interpretivism' debate, when philosophers of science have left this long behind (in my opinion there is no inevitable great divide, except one we have created ourselves), and the practice of natural science is and was never as simple as hypothesis testing via experiment. 

Anyway, all this came back to me while I was investigating a question in the Citizenship Survey. This survey was short lived, being part of the Labour government's response to 9/11 and the 2001 'northern riots'. As now, it was argued that a lack of 'community cohesion' is a risk factor for community conflict - that seems a bit circular, but never mind - and hence riots and Islamist/ far-right terrorism. The survey aimed to examine community cohesion and conflict through a set of questions on volunteering, community involvement and knowledge of/ attitude to violent extremism. One of these questions, which looks like it directly fits with the 'contact hypothesis' (Allport 1954), asks:

·        how often, if at all, have you mixed socially with people from different ethnic and religious groups to yourself…..at your home or their home? [and variations on this, including 'at work, school or college' and 'at the shops']

This question is a good example for thinking through what the data tells us, what it can be used for and perhaps why the data was created. All this should begin with some thoughts about what the individual's answer to this question means, and so how the question is understood and relates to real lives.

One obvious critique of this question is the usual 'social desirability' problem: given the narrative about not mixing socially as being symptomatic of the stigmatised position of racist/ghettoised, it's likely that some will exaggerate their 'mixing socially'. I suspect that Muslims probably (on average) feel this pressure more than others, but I could be wrong. This is made all the more problematic by the reliance on memory - it's easy to 'remember' the right answer.

More important, though, is the understanding of 'mixed socially' and 'different ethnic and religious groups to yourself'. The former phrase has an explanation in the interview:

·        'By 'mixing socially' we mean mixing with people on a personal level by having informal conversations with them at, for example, the shops, your work or a child's school, as well as meeting up with people to socialise. But don't include situations where you've interacted with people solely for work or business, for example just to buy something.'

This is a bit confusing, in my opinion. Many, or perhaps even most, situations could be interpreted as both. I go into a shop to buy a paper, and have a chat with the guy behind the counter about the news: this is both 'just to buy something', and 'an informal conversation... at... the shops'. I talk to another parent about the school-run traffic: this could be solely for the business of fetching children, with no sociality beyond this.

The latter phrase is, for me, even more problematic. It requires a judgement of one's own 'ethnic and religious groups' and that of others, and this can depend on the level of importance placed on such identity. As an atheist, would I count all religious people for this, and would I even know if they are religious? I could play football with someone who goes to church or the mosque and not know as it's not a conversation we'd have. Would a Catholic have to count a Protestant? Or a Shia Muslim count a Sunni? Do English people count Scottish people as a 'different ethnicity' or Tamil people count those from the rest of India? My theory is that those who take a religious identity more seriously will be more likely to know the answer, whereas someone who has no religious identity may well 'mix' but hasn't registered it. So it's possible that the aggregate data here shows a pattern different to any underlying contact.

Even if we think that the question is actually understood/ interpreted/ answered in the same way across the population, we also have to think about what any answers actually tell us. Yes, a bald figure of who says they socialised with who and when may not be an exact correspondence with reality, but perhaps we can use it for comparisons. However, even here there are problems due to all the other stuff going on... My first question would be how this compares to socialising in general: after all, any 'mixing socially' with those deemed different (by the survey) ought to be seen in the context of all 'mixing socially'. Someone who doesn't answer affirmative here might not do much socialising in the home, full stop. It reminds me of the reasons why I was out on the street every night as a teenager: I didn't have private space at home for socialising, being in a two-up, two-down terrace where my bedroom was where the kitchen would normally have been. Second, what's 'normal' will vary for a multitude of reasons, some economic, some cultural, some geographical and so on. Living in a County Durham pit village will create different opportunities for sociality than Newcastle city centre. Data like this might say something if we can compare like with like, but no two environments are exactly the same.

Thus, the theory laden-ness of data and data creation comes back when the analysis is done at the end. The survey was created for particular circumstances - particularly the fear that white working class and Asian Muslim 'communities' are divided (see Ted Cantle) - and the questions reflect that. Similarly it's possible to analyse the data for this theory. However, perhaps an ethical sociology should approach the data in a different way... before jumping to conclusions and checking whether the white working class and Asian Muslim populations are somehow uniquely different to the rest of society, we should check for other explanations, through thinking about the data more creatively. That's not to say that this quant data is irrevocably flawed, but that like any data it is created in a particular process and context, and that this needs to be considered very carefully.