Following the theme of the first two presentations, this presentation will focus on the choices available for research synthesis when summarizing research evidence. The presenters will describe the current research synthesis practice of the What Works Clearinghouse (WWC) as well as several alternative models, including inverse-variance weighted fixed effect and random effects meta-analysis and Bayesian meta-analysis. The presenters will begin with an overview of the context of the WWC's intervention reports, with specific attention to the method used to synthesize the results of multiple studies and the context in which synthesis occurs. The data used for the illustration will be derived from published WWC intervention reports in which at least one synthesis across two or more studies was reported. Given that most of the syntheses that the public-policy clearinghouses do involve only a very small number of studies, they could consider not doing any syntheses at all unless the number of studies reaches some minimum threshold (e.g., five studies). However, consumers really would like to know something about the typical effect, even if the number of studies is small, and are likely to adopt idiosyncratic rules to help them do this. Alternatively, the clearinghouses could take a different approach to synthesis either in place of or as an adjunct to current practice.