Power and sensitivity in design decisions
This chapter deals with a decision-making for the case of a continuous covariate and discusses the parallel situation of the category-type covariate considered for use as a blocking factor and explores the choice of within-subjects or between-subjects design. A covariate can be incorporated into an analysis to improve the efficiency of the design. The benefit of the adjusted relative to the unadjusted design is more marked in terms of relative sensitivity than in terms of relative efficiency. A researcher was designing an experiment to compare the effectiveness of some instruction techniques. This would require a single-factor between-subjects design with random allocation of subjects to conditions. The within-subjects or repeated measures design is almost always more efficient than the between-subjects or independent groups design. Validity is the extent to which the observed effect on the dependent variable is caused only by the values of the independent variable that define the conditions to be compared.