How do we study the public?

 ** Originally published in Crick Centre report How Do We Study the Public? **

Researching the public understanding of politics has scarcely been more important. Yet this is a field that is dominated by quantitative studies, whereby ‘the public’ is typically reduced either to public opinion or voting behaviour. I was asked by the Crick Centre to reflect on how we could and should study public understandings of politics. In particular, I was asked whether we can study it in a qualitative way. In this short blog post, I will make the case that we cannot study ‘public opinion’ using qualitative methods, but that there remain many different and legitimate ways to study the public understanding of politics from a qualitative perspective.

In making this argument, I’d like to bring attention to both the character of public opinion and of the qualitative-quantitative methodological divide. ‘Public opinion’ is a quantitative concept designed for quantitative methods; it is also the common sense and orthodox way to study public understandings of politics. While qualitative methods can be used to study public understanding of politics more generally, they cannot be used to study public opinion. A different terminology is required. While this may appear arcane, this reflects fundamental methodological differences between the two methodologies. In doing so, the predominance of quantitative methods as common sense in both academia and beyond needs to be challenged. I will discuss this in relation to focus groups, a method that I have some experience in using.

Focus groups — or group interviews, if you prefer — are unique: they provide us with a method to generate data on how a small group of people discuss, make sense of, and tell stories about certain (political) issues. The resulting data can be very rich and therefore helpful in understanding the worlds in which people inhabit. The methodological trade-off is obvious: you get rich and unique data, but it is difficult to ‘scale up’ that data from the immediate milieu in which it was collected. How do you generalise from a discussion of say, 40 people across 8 discussions in a specific locale to, say, something meaningful, rigorous, and robust about something to do with ‘the public’? And how do you know that people are telling the truth? Or, how do you get a representative sample?

Before jumping into the answers, we need to challenge the terms of these questions. In this context, qualitative research on the public and focus groups in particular is destined to fail when asked to fit these values. ‘Representative sample’ is, for example, a statistical term that is only intelligible in those terms. A representative sample is an essential component in many quantitative methods for studying public understanding of politics. By aggregating individual preferences and beliefs from a systematically random sample and extrapolating that onto a wider population, these methods are able to say some rigorous and meaningful about the public. Although this is technically difficult to achieve, it is philosophically uncontroversial.

Qualitative research does not fit with this. If you are interviewing 40 people in a specific locale, a ‘representative sample’ is impossible. Although quantitative methods for studying the public are and will be always highly valuable, the assumptions and value systems of quantitative methods double-up as the implicit assumptions and value systems for social sciences as a whole — and even, to some extent, society itself. Qualitative research cannot live up to these values, and are set up to fail on these terms. We can therefore benefit from developing an alternative vocabulary and conceptual foundation for qualitative methods into studying the public understanding of politics. We need qualitative concepts for qualitative methods. Here, I’d like to propose three ways of rethinking this foundation — all of which will be familiar to those versed in qualitative methodology.

The first move is to rethink the unit of analysis. In quantitative methods, the unit of analysis is the individual. And in particular, it is their beliefs and attitudes (accessed through survey questions, for example) or their behaviour (recorded through voting, for example). With respect to the former, there is often an assumption that the true beliefs and attitudes of an individual are accessible to the researcher; otherwise, there wouldn’t be so much concern over the potentially biased or misleading wordings of question. In some qualitative research and in focus groups in particular, the unit of the analysis is not the individual. Rather, it is something social: typically something to do with the social constitution of the worlds that we inhabit and make. In concrete terms, this might mean that the unit of analysis might be peoples’ experiences, the kind of shared narratives that people with similar experiences tell, and the process in which identities are made, remade, and contested. These are phenomena that cannot be captured through quantitative methods.

The second move is to rethink sampling. Theories and practices of sampling are less established and accepted in qualitative research when compared to quantitative counterparts. Representative sampling cannot work in qualitative research. If a sample is, say, less than a hundred, then the idea that several people can represent a particular demographic is politically, ethically, and methodologically dangerous. Rather than throw systematic sampling out the window, however, a qualitative alternative is to use what’s known as ‘theoretical sampling’. If the unit of the analysis is social rather than individual, such as a certain type of lived experience, then one can purposively sample say, two or three localised groups with theoretically-meaningful difference in their lived experience, so to generate meaningful comparisons that can be analysed and extrapolated from.

The third move is to rethink scaling up. Scaling up in quantitative research is philosophically straightforward: generalisability means scaling up from a representative sample to a wider population. One qualitative alternative scaling up strategy is analytical generality. If you have, for example, conducted five focus groups in the same area, with people who have similar experiences, and you find that in those five groups that the same narratives and identities are emerging in each one; then one can be reasonably sure that there’s some sort of shared phenomenon going on there; which can be unpicked through transparent and coherent theorising. In other words, one can scale up by making a sort of wager: that those lived experiences have a consistent and structural pattern that we theorise as typical.

These are three ways of challenging the quantitative common sense of researching the public understanding of politics. By using a qualitative unit of analysis, sampling strategy, and scaling up method, we can make rigorous claims about how the public understand politics. Does this mean that we can conduct qualitative studies of public opinion? If the public means an amalgamation of individual opinions and beliefs — which is how public opinion is commonly conceived and discussed — then no. Public opinion is intrinsically bound up with quantitative methods. My suggestion for a qualitative alternative to the quantitative concept of public opinion is everyday narratives. This involves analysing how the public tell stories about politics. This means studying what sort of politics these stories justify or contest and how the identities and narratives that underpin these make and remake the very boundaries of legitimate political action. This is one way that qualitative research in the public understanding of politics can be highly valuable to scholars and non-scholars alike.