Details

Analyzing Blocked Randomized Experiments

I recently finished a multi-year project with collaborators analyzing an experiment to evaluate the effectiveness of a government job training program for unemployment insurance (UI) recipients in Rhode Island. In presenting the results, I will talk about some methodological considerations that arose organically in this context. Specifically, I will provide an introduction to the potential outcomes framework and show how it enables us to formulate a precise mathematical description of what exactly it is we want to estimate when running a randomized experiment. Then, we will see some principled and less principled methods for performing this estimation. Some directions for future research, including opportunities for students to get involved, will be discussed at the end.