Comments on Slavin : Synthesizing Causal Inferences

When causal inferences are to be synthesized across multiple studies, efforts to establish the magnitude of a causal effect should be balanced by an effort to evaluate the generalizability of the effect. The evaluation of generalizability depends on two factors that are given little attention in current syntheses: construct validity and external validity. The author concludes with six concrete suggestions for ways that future syntheses could be improved to make them more relevant for educational policy makers.