ABSTRACT
Educational research and development has reemerged as a central national policy issue. This chapter presents the logic of randomized experiments, discusses reasons for their infrequent use in education, and presents several ways that evaluators may apply experiments to the special circumstances surrounding education. Scholars have observed that schooling occurs within a complex and multilevel system. Federal policies affect how states operate their school systems, states mandate district-level requirements, districts influence school-level policies, school-level leadership affects classrooms, and classrooms and subgroups within them affect the content, pace, duration, and quality of instruction. New approaches to advancing educational change, most notably through the movement toward school-based adoption of externally developed whole-school reform models, also have strong implications for the perceptions of schools as highly distinctive and complex institutions that are resistant to research-based models of change. The complexities of school systems also may result in less-than-ideal program implementations or imperfect compliance with participants' treatment assignments.
