This week, we take a step back and consider the role that the counterfactual approach to causality plays in modern day social sciences. We discuss various critiques of the potential outcomes framework and of the “causal empiricism” turn in social science research. Many of these arguments are summarised (and contested) in chapter 13 of the Morgan and Winship textbook.
For a particularly strong defense of causal empiricism, see this paper by Cyrus Samii (2016) who argues that the types of methods we have studied on this course provide estimates of causal effects that are no less externally valid than those produced by traditional regression approaches, and are almost certainly more internally valid. As he argues, the causal empiricist approaches we have studied are grounded in a joint realism about what can be achieved from empirical analyses: a realism about whether a given research design is adequate to identify a causal effect; and a realism about the specificity of empirical results.
On the other side of the coin, it is worth reading this paper by Christopher Ruhm (2018), who examines the trade-offs between research methods in terms of answering important questions versus cleanly identifying causal effects. Although he does not dismiss the importance of establishing credible estimates of causal effects, he argues that this can occasionally come at a cost: we may ignore critical issues that are policy relevant because they are not amenable to the types of analysis that we covered on this course. I think this is an important point, and that the paper is very much worth reading.