Assessing External Validity in Practice

We review, from a practical standpoint, the evolving literature on assessing external validity (EV) of estimated treatment effects. We provide an implementation and real-world assessment of the general EV measures developed in Bo and Galiani (2021). In the context of estimating conditional average treatment effect models for assessing external validity, we provide a novel method utilizing the Group Lasso (Yuan and Lin, 2006) to estimate a tractable regression-based model. This approach can perform better when settings have differing covariate distributions and allows for easily extrapolating the average treatment effect to new settings. We apply these measures to a set of identical field experiments conducted in three different countries (Galiani et al., 2017)