This is a guest post by Simon Frankel Pratt. He is a lecturer in the School of Sociology, Politics, and International Studies at the University of Bristol.

In the social sciences, research and data are often divided into the categories ‘quantitative’ and ‘qualitative’. This is incoherent and should stop. There’s nothing informative in this distinction in terms of the logic of enquiry, the mode of inference, or the way data are used to support claims about the world. There is nothing methodological about it. But it won’t stop because if it did, our discipline would further marginalise non-positivist research.

I complained about this on Twitter, and I will expand on these complaints here. I’ll start with the philosophy of social science problems. But then, I’ll talk about power and hegemony.

The philosophy of social science part

Quantitative research does exist, in that the term seems to refer to something fairly stable and consistent: the use of statistics and mathematical models to find patterns in empirical data.[1] Arguably the word ‘quantitative’ is still misleading, but it at least has a stable referent.

But qualitative research is a residual category containing a heterogeneous and largely incommensurate array of other approaches, differing on fundamental matters of ontology and epistemology but grouped together on the sole, arbitrary basis that they use ordinary language and not numbers. Identifying work as ‘qualitative’ sheds no light on how it works, and identifying work as both quantitative and qualitative does not mean ‘doing mixed methods’. It just means you talk about stuff with words as well.

All data are qualitative, and all research is qualitative, in the sense that they are about the qualities of something. ‘Qualities’ are the traits, features, aspects, states of being, etc of the world around us. Whether or not we describe the world around us through numbers or through ordinary language, we are representing its qualities. 

Quantitative data are therefore also qualitative data. Quantification is ultimately linguistic: it is a form of translation. Most of our descriptions start as ‘ordinary language’, and in some cases, we ‘code’ those descriptions using numbers rather than words. Quantitative methods are the ways we do this, and the ways we analyse the resulting translations.

But this raises a question: why would we divide research in such a fundamental way based solely on the language of description and analysis?

There is more or less only one methodological situation where social scientists use numbers, statistical modeling, and other mathematics: there are too many data to talk about them with words, and it is too hard to discover associations in them until we standardise them in ways that allow for en masse comparison. But what about when scholars use ordinary language to compare cases and look for interesting correlations or patterns using ordinary language? Isn’t this an example of using the same logic of inference but without quantification? There are good examples of (positivist) ‘small n’ research that does this. Indeed, this is the infamous KKV approach. See, qualitative positivism!

Actually, this is also quantitative, in that it still involves talking in terms of degrees of variation, similarity, association–the quantity of a quality, basically. You don’t need to use numbers to quantify something. But once you have more than a few cases, comparison becomes computationally difficult, and the use of mathematical expressions becomes key in arranging data in ways that let us see the signal in the noise.

The problem arises when you recognise that not all social scientists are looking to compare cases in order to find interesting correlations. Positivists are doing this, and perhaps if restricted to positivism alone, you could have some meaningful division between ‘mathematical’ and ‘non-mathematical’ methods. Insofar as the distinction between ‘quantitative’ and ‘qualitative’ functionally means this most of the time, that is already the case.

Since we aren’t all positivists, though, this just further enshrines the hegemony of positivism (and the fetishation of maths as the only way ‘real science’ is done) while making it harder for the rest of us to explain what we do and why we do it. Interpretivist who studies culture? Get in the qualitative sack. Post-structuralist who studies how ‘culture’ is a fractured discursive illusion? Get in the sack. Marxist who studies the material effects of capitalism? Sack. Does it matter that you all have nothing in common when it comes to epistemology, ontology, and methodology? No, hush, the sack is sound-proof and the real scientists need to concentrate on their work. With numbers!

The thing is, this distinction also harms positivists too, at least if they care about philosophical coherence. Positivism is ultimately a form of instrumentalism, or at least of radical empiricism, and this is usually easier to understand when we see a mathematical model. We know that this is an abstraction based on data that has gone through a coding process, during which it has been simplified and standardised. The risk, though, is assuming that ordinary language descriptions are somehow more ‘realistic’ because they are ‘richer’ in detail, rather than themselves being a product of human interpretation. As Patrick Thaddeus Jackson has pointed out, this contributes to the deterioration of positivist research, as it lapses into naive realism. [2]

So what are some better ways to draw methodological divisions? I’ve already pointed out that ‘mathematical and non-mathematical’ is not so great. Cesi Cruz suggests ‘desk’ and ‘field’, which may speak to some major differences in methods of data collection but not necessarily those of methodology, as an interpretivist might, for example, analyse a recorded TV interview they found online in the same way as they would one they conducted themselves in the field. This is still miles better than quant and qual, though, and shows some interest in the actual conduct of enquiry. The old ‘ideographic/nomothetic’ divide is out, because all descriptive language is to some extent theory-laden and contains some implications of generality.

But the challenge in finding a better distinction raises a worrying question for graduate training: what if research approaches cannot ever be divided into only two types?

The power and hegemony part

This brings me to power, the discipline, and the hegemony of positivism. The real reason for the persistent categorisation of research into quantitative and qualitative, despite the latter being a residual category, is not because nobody has ever thought about it before. It is because the discipline prioritises, empowers, and fetishes the former. It doesn’t matter that this binary division, and probably any binary division, makes no sense. The only way to compete with Those Who Count, for journal space, for funding, for jobs, and for control over graduate curriculums, is to band together.

No, I do not share philosophical foundations with Marxists and poststructuralists,[3] but I share their disciplinary position: my work is undervalued in comparison to people who measure degrees of variance and then run some kind of regression analysis to explain them. ‘Qualitative’ research exists as a term because it’s an identity that sounds better than ‘anti-/non-positivist’, and without a common identity, we would just be shoved out entirely. It isn’t clear to me that we won’t be anyway, at least in North American political science departments and journals.

The need to think about such things as epistemology and ontology in the first place is an outcome of power, and the lack thereof. If positivists do not care about the philosophical consistency of their approach, it is because nobody forces them to, and in a practical sense it’s entirely possible to conduct and produce research without seriously interrogating the assumptions underlying the logic of enquiry. If positivists do not have to invest time and effort in understanding positivism, they certainly do not need to understand things that aren’t positivism, unless forced to do so. The proud coalition of Camp Qual is just our best attempt at doing that.

So the even more worrying question is: how do we force positivists to care?

 [1] Yes there is non-empirical work that also has maths, but game theorists tend not to call their research ‘quantitative’.

[2] At a certain point, social science positivists will likely associate ‘Carnap’ with a short sleep in an automobile.

[3] I actually sort of do, sometimes, but it’s complicated.