Sceptical or clerical

There is considerable and growing interest among social scientists and funders in the potential of research that mixes the approaches usually termed ‘qualitative’ and ‘quantitative’. One of the obstacles to further exploration of that potential is the frequently cited notion that these approaches represent more than methods – they are incommensurable ‘paradigms’. This paper discusses some of the reasons why these purported paradigms are not as much of an obstacle to mixed methods work as is sometimes portrayed. In doing so it touches on the nature of social scientific progress, and the role of different kinds of theory in that progress. The paper concludes that, as currently conceived, much theory writing is worse than useless. It is, like the notion of qualitative and quantitative paradigms, an obstacle to the development of theoretically appropriate mixed methods work. Why not mix methods? Commentators such as Teddlie and Tashakkori (2003) and Cresswell (2003) have recently advocated greater use of a mixture of methods in social science research. This has become something of a general phenomenon in the last decade (Brannen 1995), in economics (Lawson 2003), psychology (Debats et al. 1995), and sociology (Rogers and Nicolaas 1998). In the UK ESRC Research Methods Programme there are, at the time of writing, two funded projects solely and explicitly about developing models for combining methods. In education, this trend has been linked to concerns about improving both the skills base of educational researchers and the quality of educational research (NERF 2001), as well as a recognition that the methods chosen to conduct research should fit the research question being posed (National Research Council 2002). In general, the debate has, for the most part, moved away from whether data from different sources should be combined, to a much greater focus on how this is done (National Institute of Health 1999). However, attempts to build the capacity of research communities to conduct such work immediately hit a considerable problem (among others, see Gorard