| Sign In to gain access to subscriptions and/or personal tools. |
Nonequivalent Group DesignsThe Use of ANCOVA, Multiple Regression, Repeated ANOVA, and Effect SizeNorthern Illinois University In pretest-posttest nonequivalent group designs, naive use of analysis of covariance or multiple regression can lead to Type I errors. In the evaluation of a Title IV-C, ESEA project, thorough analysis of the interaction term of a repeated model analysis of variance was shown to help identify areas of significant change. Meaningful differences were also supported by the standardized difference score, Effect Size. It was suggested that a two-fold increase of Effect Size from pretest to posttest might be an effective guide in the determination of meaningfully significant change.
Evaluation Review, Vol. 7, No. 6,
841-850 (1983) |
||||