|
Abstract:
|
Educational research often studies subjects in naturally clustered groups of classrooms or schools. When designing a randomized experiment to evaluate an intervention directed at teachers, but with effects on teachers and their students, the power for the treatment effect needs to be examined at both levels. If the treatment is applied to clusters, power is usually reduced, however at the same time, this type of design decreases the probability of contamination, which can also reduce the power to detect a treatment effect. Designs that are considered preferable under ideal conditions may not have the highest power when contamination is an issue or when the treatment needs to be evaluated at more than one level. We study the efficiency of several designs for detecting effects at multiple levels under a unified framework, and explore the effects of contamination in the different designs.
|
