The published research literature on any given topic likely represents a highly unrepresentative sample of all that is known. That is because authors and editors are rarely interested in publishing ambiguous or disconfirming results concerning a given hypothesis. Such “publication bias” creates vexing problems when performing formal meta-analyses, or whenever anyone tries to interpret the results of a body of empirical work.Suppose, however, that investigators could agree to collect and post public commitments to their research plans, including their hypotheses and methodologies, in advance of collecting all their data. Not only could simple transparency like this go a long way toward alleviating publication bias, it could also deter other ways researchers have of cherry picking and distorting results.This grant funds a project by the American Economic Association (AEA) to bring just such a thing about. Led by MIT economist Esther Duflo, the AEA will set up a national registry for randomized controlled trials in economics. By linking study designs to related datasets and by making study details more easily searchable, the proposed registry would advance the Foundation’s efforts to promote communication, transparency, and best practices among scholarly researchers.