In some experimental evaluations of classroom- or school-level interventions, random assignment is conducted at the student level and the program is delivered at the higher level. This paper clarifies the correct causal interpretation of “program impacts” when this study design is used and discusses the implications and limitations of this research design. A real example is used to demonstrate the paper’s key points.
This paper illustrates how to design an experimental sample for measuring the effects of educational programs when whole schools are randomized to a program and control group. It addresses such issues as what number of schools should be randomized, how many students per school are needed, and what is the best mix of program and control schools.