Post

Those of you who know my research know I think randomised trials are a really important part of impact evaluation in education research. But, of course, they have limitations and so they’re not always the right tool for the job.

In this article we set out to think systematically about their strengths and weaknesses in the context of education research. We hope the framework we propose helps to guide thinking about when they are that right tool, and when alternatives better balance those strengths and weaknesses.

We particularly posit that RCTs are most useful for building and testing theory, while robust quasi-experimental designs (where greater statistical generalisability is often feasible) are likely to be better placed to inform policy and practice decision making directly.