Structures of Causality and Design-Based Inference

• In Event: Causal Inference: Accounting For Heterogeneity

Abstract

While the commonplace Holland-Rubin counterfactual model of causation has proven to be both very useful and amenable to powerful design-based inferential tools, it is by no means the only causal structure in widespread use in contemporary Political Science. For instance, some models of state building, or the expansion of market-based economies, have utilized arguments analogous to evolutionary natural selection. Others have emphasized path-dependencies that imply temporal instability in causal effects and render standard estimates of ‘average treatment effects’ rather uninformative relative to underlying theoretical claims. Still others have pointed out that causal effects may be asymmetrical, embody hysteresis, or be conditioned by relations of necessity or sufficiency.

This paper considers the challenges posed by theories that are founded in, at least in part, alternative causal dynamics. The paper then proceeds to discuss how a serious treatment of alternative causal models can lead to the better application of standard methodological tools, and open the possibility of assessing causal theories of a counter-factual sort against alternatives not rooted in the same structure. Examples are taken from International Relations and Comparative Politics to illustrate the challenges and potential avenues for progress.

Author

• Marcus J. Kurtz, Ohio State University