Abstract
This article responds to criticisms of the “honest broker” role of evaluation research done in the positivist tradition. Quantified, statistically representative information frequently has greater utility and perceived credibility than activist research. It also better fits the information needs of policymakers. The roles of evaluations in a dynamic, national policymaking environment are explored. “Mixed results” are shown to be the result of variation in recipients and program implementation as well as assessing multiple outcome variables.