Most behavioral scientists know two important concepts -- how to analyze continuous data from randomly assigned treatment groups of subjects and how to assess practice effects for a single group of subjects given a constant treatment at each of several stages of practice. However, except in the case of the repeated measures Latin square design, researchers are not facile in analyzing data from different subjects receiving different treatments at various times in an experiment. This book helps fill the void.

This volume is intended to fill a niche in the statistical toolbox of behavioural scientists. Most know two important concepts - how to analyze continuous data from randomly assigned treatment groups of subjects and how to assess practice effects for a single group of subjects given a constant treatment at each of several stages of practice. However, except in the case of the repeated measures Latin square design, researchers are not facile in analyzing data from different subjects receiving different treatments at various times in an experiment. As the opening chapter states and later chapters elaborate, randomization of treatment sequences for different subjects may lead to unbiased conventional estimates of treatment effects and of time-related effects. Yet, the standard errors of such estimates may be unduly large because the error sums of squares include contributions from nuisance variables such as stage-of-practice effects when treatment effects are of interest.