Our main critical argument is that it is problematic to model social processes on natural processes in this way — whether as universal laws, causal generalizations, or robust empirical correlations — because it leads to rather narrow conceptions of testing and explanation, in which the element of prediction is elevated at the expense of contextual and ontological factors. In short, our target is the law-like conception of explanation and testing that the causal law paradigm elevates to the status of an ideal. (19)
As against inductive and deductive modes of reasoning, we argue more positively that retroductive reasoning provides us with a general form or logic of explanation in the social sciences.
More specifically, we challenge the compartmentalizing tendencies of positivist social science investigation — a logic of scientific discovery followed by exhaustive empirical testing and explanation — and propose instead one overarching logic of investigation comprising three interlocking moments: the problematization of empirical phenomena; the retroductive explanation of these phenomena; and the persuasion of — and intervention into — the relevant community and practices of scholars and lay-actors. (19)