Deaton v. Banerjee on RCTs

Short video summaries of a debate between economists Abhijit Banerjee and Angus Deaton on the merits and demerits of randomized control trials in development economics.

I think they're both right. When done well, experiments can provide compelling evidence that completely shuts down policy debates so that everyone can move on and work on other things, as with the US Department of Education's experimental evaluation of abstinence-only sex education curricula.

At the same time, Deaton is quite right that experiments can be executed well or poorly, and a poorly designed and/or executed randomized experiment will often provide evidence of lower quality than a well-designed and executed non-expermiental study. More broadly, I think Deaton with agree with my good friend Burt Barnow, who said "experiments are not a substitute for thinking". Indeed.

Hat tip: Rebecca Thornton