While there has been some increasing interest in external validity, most work to this point has been in assessing the similarity of a randomized trial sample and a population of interest (e.g., Stuart et al., 2010; Tipton, 2011). The goal of this research is to calculate empirical estimates of the external validity bias in educational intervention evaluations that can result from selecting sites purposively. This will help researchers and funding agencies understand the implications of purposive site selection. (Contains 1 figure.)