How the cases you choose affect the answers you get, revisited

External validity is a major challenge for experimental research. I offer a new perspective on this challenge, drawing on work on case studies and causal inference – the sort of material regularly covered in introductory methods courses in political science – to reflect on the use of experiments in the study of global development and poverty alleviation. I argue that single experiments in this area are often essentially single case studies. They can offer important insights, but generalizing from them suffers from the same (well-established) problems of generalizing from all single case studies – especially in the absence of theoretically-informed attention to the selection of experimental sites. One way experimentalists have sought to improve external validity is through replication. I suggest a more promising approach is to combine experiments with case study and comparative methods to link selection of experimental “cases” to theory.