Causality as Conundrum: The Case of Qualitative Inquiry

Abstract

My argument here is a tentative one, it comes out of work in progress. In summary, it is that social scientists cannot avoid assuming some notion of causality, but that this concept involves deep problems that are as yet unresolved, and which may have significant implications for how we do research. There has been substantial criticism of quantitative research for failing to deal with the issue of causality adequately, notably by realists who insist on the need to identify underlying causal mechanisms rather than simply documenting recurrent patterns in the data (for example, Sayer 1992; but see Kemp and Holmwood 2003). Here I will focus primarily on the case of qualitative research. The core meaning of the term 'causality', it seems to me, is that one type of thing (X) tends to be followed by another (Y), and that this occurs as a result of some force exerted by the occurrence of an X, rather than by happenstance. Note that this core sense does not assume that an X is always followed by a Y, or that what precedes a Y is always an X. In other words, using this core sense of the term, in designating X as the cause of Y there is no assumption that X must be a necessary or a sufficient condition for the occurrence of Y; even less that it should be both a necessary and a sufficient condition. The latter requirement defines a much stronger sense of 'causality', one whose applicability is uncertain; though some models of the process of explanation assume this strong sense, for example both analytic induction and Ragin's comparative method (on analytic induction, see Hammersley 1989; Ragin 1987), and I wouldn't want simply to rule them out.