Not all causal effects are created equal: selecting pathway cases for in-depth analysis using causal forests

Providing complex answers to causal questions requires both cross-case evidence for the existence of a causal effect and within-case evidence for the mechanisms through which this effect propagates. The former requires a causal identification strategy, while the latter is more amenable to qualitative investigation using methods tailored to mechanistic analysis, such as process tracing. However, once a cross-case causal effect has been estimated, there is little guidance on how to select a case in which to trace the mechanism. After reviewing the limitations of the available algorithms , I present a novel case selection method that uses causal forests to recover granular effects for each case in the sample, and show how this approach can be used to select pathway cases in a manner consistent with their stated goal. I then briefly discuss how this framework can be extended to other types of cases, including typical and deviant cases.