A central task of evaluation is to use systematic inquiry to judge the merit, worth, and significance of programs. As a field in education, evaluation began to mature during the 1960s, when policy makers began funding researchers to study the impact of Great Society programs. Since that time, evaluation and policy-making have been intertwined, such that the history of evaluation research or educational policy cannot be told without understanding how one field of practice influenced the other. The question that organizes this chapter is, How should evaluation inform educational policy-making? Throughout the short history of the field, there have been different answers to this question. Recent educational policies have emphasized the need for evaluation research to provide answers to questions about "what works" so that policy makers and educational leaders can decide what to fund and what to adopt. The epistemology of evaluation that undergirds these policies is what House (1978) might have called a "liberal objectivist" epistemology of evaluation. The liberal objectivist epistemology privileges a particular understanding of what constitutes scientific methods of inquiry as a means to support policy makers' and managers' efforts to hold programs accountable for results. In this chapter, the author contrasts this epistemology of evaluation with a dialogical epistemology for educational evaluation. A dialogical epistemology privileges dialogue as a central component of inquiry, enjoining evaluators to anticipate and respond to the voices of diverse program stakeholders in designing, conducting, and communicating the results of an evaluation. Importantly, a dialogical epistemology is not inconsistent with goals of holding programs accountable, but it differs from the view of accountability in the liberal objectivist epistemology in casting evaluators in the role of participant, rather than arbiter, in judging the worth or merit of programs and in decision making about the fate of programs.