Abstract

The debate on empirical tests of social mechanisms suffers from a fragmented view on the relative benefit of the empirical method a researcher considers to be superior, compared to the flaws of all other methods. In this outlook. I argue that disciplinary barriers might be surmounted by a common methodological perspective on the analysis of social mechanisms. First, experimental, quantitative, qualitative, and simulation methods (agent-based modeling) are all required, but also capable to deal with the issue of causal identification, respectively. Second, having established causal identification (among which I subsume strategies to deal with causal heterogeneity). each method disposes of genuine techniques to deal with the most crucial property of mechanism-based explanations: input-mechanism-output (IMO) relations.