Paradigm Wars

I coined the term negativists in this blog post a couple of weeks ago. This is something that I feel I am going to keep returning to over the course of the next several months, perhaps years, as I address methodology and paradigms as a part of my research. I want to explore this negativism and try to understand what drives it. I also want to explore ways that negativism can be addressed so that the idea that numbers and words cannot be friends can be addressed, which as far as I can see can only be beneficial for research.

I chose the word negativists, and its corollary negativism, to make a point about people who use the word positivist as a term of abuse. But in fact it can be applied to any vociferous methodological bloodlust that determines to hoist a paradigmatic standard and defend a corner of what some plainly believe to be an ontological and epistemological battleground.

The debate in the social sciences over ontology and epistemology has often been fraught. Both ‘sides’ of some imagined division have been deemed guilty of attempting to skewer their ‘opponents’ on charges that include gender hegemony, for example “malestream” positivism (Oakley 1998:707), or of misunderstanding the nature of evidence: “the plural of anecdote is not statistic” (Mulgan quoted in Rutter 2012:15). This “dreadful 1980s culture war” (Goldacre 2013) is irrelevant. Oakley (2000) gives an account of how actual practices in both natural and social sciences research defy the assumption that researchers occupy one of either side of a binary world of ‘quality’ or ‘quantity’, ‘interpretivism’ or ‘positivism’, ‘naturalism’ or ‘scientism’.

I reject the expectation that one must declare an adherence to either one of these positions. As both Oakley and Goldacre (and others, like Stephen Gorard) have emphasised, it is the research question which should dictate the approach to addressing it (Oakley 1999, Goldacre 2013), not some perceived membership of a paradigmatic ‘tribe’. By extension, it is the research question, setting, participants and expected outcomes that should inform the ontological assumptions informing the assessment of the external validity of any findings. Knowledge is always provisional and imperfect. If it were not so, so-called ‘positivists’ would not acknowledge the influences of placebo, Hawthorne and research participation effects (McCambridge, Kypri and Elbourne 2014). So-called ‘interpretivists’ would find their work so unique as to not have any applicability beyond the specific circumstance in which it was conducted (after Larsen-Freeman 2000). To debate the minutiae of the philosophical underpinnings informing differing world views is to split hairs. Or, to borrow from Theravada Buddhist teachings, “Conjecture about [the origin, etc., of] the world is an unconjecturable that is not to be conjectured about. That would bring madness and vexation to anyone who conjectured about it” (Acintita Sutta [AN 4.77]).

Social sciences is damaged by this attempt to make the world binary. It is damaged by the assumption that there are mutually exclusive worlds of ‘quality’ and ‘quantity’. I do everything I can to not use those words to describe the work I am interested in. Quite apart from anything, else you will find the work of self declared ‘qualitativists’ to include all sorts of ‘quantitative’ language (more, less, greater, lesser, larger, smaller, a majority, a minority) and the work of self declared ‘quantitavists’ will include all sorts of references to interpretation (we are tempted to conclude, it appears, the general trend implies).

While I have meditated on this issue for a while, this particular post was catalysed by a tweet I saw yesterday which, I think, perfectly sums up the nonsense posed by this attempt to polarise the world of social sciences research.

In response to the suggestion that we should share data, the second tweeter is seeming to suggest that, firstly, the data we collect can be too personal to share – in which case what are we doing collecting it in the first place? We are going to share this information – that’s the point of research. If we have satisfied ourselves and our informants that we will behave ethically with the data (by anonymising it, for example), why not make it available for other people to interpret or combine with other similar data in systematic reviews in the pursuit of fuller understanding? More shocking, however, is the view that replicability is somehow “a quant expectation, not qualitative”. This is demonstrable nonsense. We do research so that we can understand the world better. If this tweet represents a genuine belief that so called ‘qualitative’ research can’t or shouldn’t be replicated, then what value does that research hold, beyond the immediate context (time, place, population, etc.) in which it was conducted? If I see a study that explores Birmingham school teachers’ responses to the implementation of, say, new workload expectations, that uses an ethnographic enquiry, and the results interest me to the degree that I would like to find out how they compare to the same situation in, say, London, why on earth shouldn’t I seek to replicate? But all of that aside, the false binary being perpetrated here is that you are only allowed to do certain things with data based on whether they are numbers or words!

Negativism hurts social science. I will be returning to this theme.



  1. Eli Kean

    Hello, I’m Eli Kean. We haven’t met, but for whatever reason you latched onto my 140 character reply, thought you understood completely what I was intending to say, and had a problem with it. Let me clarify a few things for you.
    1. I have no issue with quantitative methods. I think both quantitative and qualitative are valuable modes of research.
    2. Replicability, as it is understood in the U.S. (where I live), does not mean ‘the same researcher does a similar study’ or ‘a researcher reads a study and decides to try it out themselves somewhere else’. Replicability means one researcher handing over all of their methods, methodology and participants to another researcher, and that second researcher replicates the same study in situ, with the intention of getting the exact same results. Replicability is an element of research reliability (the extent to which research is consistently repeatable), which is something that comes out of quantitative research. That is not my opinion, that is fact.
    3. There are several methodologies in qualitative research which would be difficult if not impossible to adequately replicate (using my definition not yours). There are some methods of qualitative research for which generalizability is not even an intended outcome. For example, how do you propose that someone replicate and/or generalize a phenomenological study? Or one using narrative theory? In the qualitative research that I’m engaged in, it is understood and accepted that the researcher themselves and their background will influence the data, and two researchers may not find or interpret the data in the same way. It is quite easy to interpret numerical data based on the statistical rules that have been set out. It is much more difficult for two separate researchers to take a mountain of narrative data and interpret it the exact same way every time.
    4. I think you should be careful about assuming that you know what someone, whom you’ve never met and lives in an entirely different country, means with one 140 character tweet. Doing so shows you to be impulsive rather than patient, preferring to critique without communicating. Patience and the ability to communicate respectfully are important skills for researchers to have. If you had questions about what I meant it would have been nice to hear from you directly rather than see my tweet up here being criticized based on your misunderstandings and assumptions.

    • Hamish Chalmers

      Hi Eli

      Thanks for commenting. I appreciate your thoughts. If I may, I will address them in the same order in which you have presented them.

      1) Great! But can we not move away from this binary distinction between qualitative and quantitative methods? As I have touched on in this post, this particular distinction is unhelpful – not least because it attempts to lump a whole load of different ways of addressing a question as either being all about numbers or all about words. This is a false dichotomy.

      2) Replicability (and I have looked to American sources as well as British and Australian sources to help me here) is the extent to which a re-study is made feasible by the detail in which its procedure is described in the original. When we use exactly the same data and methods to re-analyse the findings of a study and find them to deliver the same results, then we have a reproducible study. Here is a definition of replicability from the American Educational Research Association (page 2). I get that we are two nations divided by a common language, but it seems, at the very least, that some of your fellow Americans may find your definition insufficient.

      3) This point is almost entirely contingent on a shared understanding of the term replicability. If we accept AERA’s definition as correct then I don’t see why a study using a phenomonolgical approach shouldn’t be replicable. And I return to the false binary of qual and quant, which are the words you used in your tweet. A badly described randomised trial is unreplicable, a well described phenomenology is. It has nothing to do with whether it uses numbers of words.

      4) Well this a little ad hominem, isn’t it? I was unaware that one needed to be personally acquainted with someone and come from the same continent as them to critique their scientific ideas. If you don’t feel 140 characters is enough to make your point, write a blog post. If you don’t want your ideas critiqued, don’t make them available to the world. As a part of a general and developing meditation on what I see as unhelpful tribalism in social science I am critiquing ideas, one example of which has been publicly shared by you. I am not critiquing you.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s