Experimental Design

Abstract

Abstract This chapter discusses three principles of good experimental design: random assignment, replication, and local control. A theme that runs throughout the chapter is that all complex experimental designs can be constructed from and understood in terms of three simple building block designs. The three designs are the completely randomized design, randomized block design, and Latin square design. Procedures for dealing with unequal cell n 's in a completely randomized factorial design are illustrated using a regression model and a cell means model. The advantages of the cell means model are discussed in detail. To reduce the size of blocks or the number of treatment combinations in an experiment, researchers sometime use confounding. The merits of group‐treatment confounding in split‐plot factorial designs, group‐interaction confounding in confounded factorial designs, and treatment‐interaction confounding in fractional designs are examined. The chapter concludes with a discussion of the use of a concomitant variable (covariate) in experimental designs. The inclusion of a concomitant variable may enable a researcher to (1) remove that portion of the dependent‐variable error variance that is predictable from a knowledge of the concomitant variable thereby increasing power and (2) adjust the dependent variable so that it is free of effects attributable to the concomitant variable thereby reducing bias.