Luis Mireles-Flores (TINT, University of Helsinki)
How is it possible to infer reliable token-level policy interventions from scientific causal generalisations? Causal accounts make a number of assumptions to define truth conditions for causal generalisations. Deciding on using one set of assumptions rather than another determines different conceptions of causality. Empirical methods of causal inference have thus distinct causal concepts already built into them a priori. When a causal generalisation is assessed and accepted as genuine, the specific assumptions used strongly determine the inferences that can be made about potential interventions on concrete policy targets. I analyse the potential outcomes framework to illustrate the argument.
