Although factorial analysis is widely used in the social sciences, there is some confusion as to how to use the technique's most powerful feature--the evaluation of interaction effects. Written to remedy this situation, author James Jaccard clearly describes the issues underlying the effective analysis of interaction in factorial designs. The book begins by describing different ways of characterizing interactions in ANOVA, elucidating both moderator conceptualizations of interactions as well as that of residualized means. After discussing interaction effects using traditional hypothesis testing approaches, he then covers alternative analytic frameworks that focus on effect size methodology and interval estimation. Jaccard summarizes criticisms of classical null hypothesis testing and offers practical guidelines for pursuing magnitude estimation and interval estimation approaches. In addition, Jaccard shows applications of all three approaches to the analysis of interactions using a complete numerical example; discusses strategies for effectively exploring interactions in higher order designs and designs with more than two levels per factor; highlights the central role of single degree of freedom contrasts and provides numerous illustrations for formulating such contrasts; considers simplified approaches to statistical power analysis; describes approaches to consider when statistical assumptions are not met; explicates the case of unequal sample sizes; considers the impact of measurement error; and demonstrates computer applications.
Factorial analysis of variance (ANOVA) is widely used in the social sciences. It is commonly recognized that one of the advantages of a factorial design is that it permits the researcher to analyze interaction effects between independent variables relative to the dependent variable. Despite the ...