Optimal Allocation of Assessment Time in Randomized Pretest–Posttest Designs

    loading  Checking for direct PDF access through Ovid

Abstract

The widely used pretest–posttest design traditionally consists of a pretest and a posttest of equal lengths. However, greater statistical power can usually be achieved by using a posttest that is somewhat longer than the pretest. This article provides a method for calculating the optimal allocation of pretest and posttest assessment units so as to maximize power and presents the accompanying reduction in required sample size for a range of parameter values. When insufficient information is available to estimate the optimal allocation precisely, the effects of allocating 75% of the assessment units to the posttest are evaluated.

Related Topics

    loading  Loading Related Articles