What Makes Failed Research? “Conventionalist” Reflections on Epistemic Failure in an Interview Study on Educational Classifications

Starting from a neo-pragmatist perspective, this paper discusses the need for developing methodological frameworks that allow to actually theorize epistemic “failure” in social research. I argue that methodological self-reflexivity and a sensitivity for performativity are key ingredients of any such framework. I use concrete examples from a recent interview study with teachers which aimed at reconstructing educational orders of classification to illustrate this argument. This study did not (yet) end up in the drawer, but it raised epistemological issues that beg the question whether it has failed in a more fundamental way: regarding its capacity to answer the research question. Three aspects that are related to this problem are discussed. To begin with, I argue that the first and crucial challenge is to detect and diagnose epistemic failure. This requires ways of theorizing our research activities that are coherent with how we conceptualize our phenomenon. The “sociology of conventions” is presented as one possible foundation for such an undertaking because it offers helpful heuristics for relating our “problem-solving” research activities to the concrete forms of agency involved in producing our data. Second, I believe we need to identify the mechanisms that have led to a particular “epistemic failure”. In my case, this amounts to problematizing conventionalized forms of doing interviews – more generally, it may mean reconsidering taken-for-granted standards and procedures that might have become deeply ingrained in our everyday research activities and hence remain unconsidered. Third, a coherent framework should also enable us to get a grasp on the interplay between epistemic and other sorts of academic failure that are linked to the structures of disciplines and research fields. In their interplay, I believe that these points illustrate that thinking about failure in social research offers a promising point of departure for methodological debates that are long overdue.