The Power of Q?

I recall gatecrashing a conversation between a mate of mine and his PhD supervisor a few years ago. As you do in these sorts of encounters you get asked what you're doing and how you hope to go about it. I explained my research was looking at the life history of Trotskyist activists and my methods were based on a series of in-depth semi-structured interviews. "Have you thought about Q Methodology?" he asked. I hadn't, because at that point I didn't know what it really was. He went on to suggest it was more scientific than my approach as it eliminates researcher bias from the research process.

I came away not entirely convinced. This is partly because not only had years of reading around the qualitative research literature had convinced me of the problematic character of quantitative methods, but also my own (then recent) traumatic experience with SPSS and statistical operations had showed me that a great deal of it is little better than mathematical masturbation. But nonetheless I knew I would at least have to undertake a brief study of Q Methodology, even if only to justify why it wasn't used in my study.

Much to my surprise, I found that Q Methods could have been useful as a supplement to my interview-based data. Despite my aversion to quantitative methods, I would agree with Robbins and Krueger's assessment that "Q method dovetails with the interpretive and cultural turn ... and is a powerful tool for anti-essentialist approaches to subjectivity and for constructivist inquiries into categories of nature, urban form and scientific activity".

So what's it all about?

Q Methodology was developed by the psychologist William Stephenson as a means of revealing someone's subjectivity without the "distortion" caused by the presence of the researcher as facilitator or mediator. This "revealing" takes place by examining an individual's traits and relating one's opinions to all the others. What this suggests is far from being a dark continent the mind, or rather an individual's frame of reference can be observed and modelled.

The application of Q methodology goes through a series of steps. Firstly the researcher must identify what Robbins and Krueger call 'the domain of subjectivity'. This is more than just the topic the researcher's interested in. For example, one question I often asked my respondents was what they thought of the Labour party. If I was interested in using Q to determine the range of opinion among socialists, I would have collated a number of statements from these interviews and pared them down to a series of claims that can be made about Labour. For instance, Labour is '... a party of big business'; '... remains a bourgeois workers' party'; '... committed to social democracy'; and '... is still the workers' friend'.

The respondent is given a number of statements and is invited to rank order them according to preference. The arranging takes place on a scale from (-)5 to (+)5, denoting least to most representative of the subject's opinion. In the case of most Socialist Party members, it is reasonable to expect statements emphasising the erosion of the Labour party as a viable vehicle for working class politics be grouped toward the most representative end, and statements extolling Labour's socialist virtues clustered in the least.

Once completed the sorts are subject to Q factor analysis. What this does is identify correlations and the strength of these correlations between subjects across the sample of statements. If there are clusters this is indicative of a common subjective outlook, which can then be further investigated using whatever tools the researcher desires.

Strangely I haven't been able to find much in the way of critical literature from a qualitative standpoint on Q. One objection that comes to mind are issues where a respondent may give equal weight to contradictory statements, which is something that may be too nuanced for the scale of preferences to pick up. For example, seeing Labour as a party of big business AND as a key arena for socialist political activity is not necessarily mutually exclusive.

That said it could have been useful for the kind of life history research I have been doing. Take for example my interviewees' school days. Their experiences run the full range from school yard tearaway to studious swot. There is no reason why such experiences cannot also be boiled down to a number of statements, and indeed it would be useful to see if there were clusters around certain sets. But it could only ever be a supplement to the core interview data. Q can indicate how one's opinions link together, but that is all it can do. To explore how they actually relate demands an examination of the respondent's narrative, which is only possible through painstaking coding and close analysis of the transcripts, with all its attendant problems and dangers.