Two statistics for use in the Shine-Bower single-subject analysis of variance designs are compared. Advantages and precautions concerning an alternate statistic to the mean square successive difference test for testing the slow change assumptions of the Shine-Bower error term are presented. (JKS)

This paper briefly describes unplanned and planned methods of evaluating differences between means and explains orthogonal versus nonorthogonal contrasts to help the researcher understand a framework of planned comparisons. A small heuristic data set is generated to illustrate the superiority of planned comparisons over omnibus analysis of…

Generalizations of the Scheffe, Tukey, and Bonferroni-t techniques are presented, each of which controls the experimentwise error rate for a particular type of partially or fully planned analysis. All three procedures provide more power than multivariate analysis of variance (MANOVA) tests. (Author/BJG)

The major concern for any research model, whether disproportionate or not, is the research question and how well that question is reflected by the model. Three "exact solutions" for disproportional situations, the hierarchial, unadjusted main effects, and fitting constant methods, are discussed in terms of the research question that each…

See EJ 041 277 for original article. (DG)

Twelve multiple comparison tests, both simultaneous and sequential, are described, analyzed, and evaluated. Overall recommendations are presented in a flow-diagram. (DG)

Social science researchers tend to analyze two dimensional frequency tables using Pearson's Chi-square test, which can be improved upon if the dependent variable represents an ordered qualitative characteristic. An improved modification by the Kruskal-Wallis analysis of variance on ranks is shown for a planned and post hoc analysis. (Author/CM)

The use of analysis of covariance in simple repeated measures designs is considered. Conditions necessary for the analysis of covariance adjusted main effects and interactions to be meaningful are presented. (Author/JKS)

Some of the more simplified methods for contrasts with equal sample sizes in multiple regression analysis are shown to result in erroneous calculations when applied to data sets with unequal sample sizes. An alternative method is provided. (Author/JKS)

The U.S. Army Research Institute for the Behavioral and Social Sciences has developed a wide range of statistical models to test hypotheses generated in relation to an equally wide range of measurement and evaluation situations. The randomized block (RB) design has traditionally been a preferred model for much psychological research. The RB has…

Discusses difficulties encountered in use of the Scheffe procedure to locate the most significant parametric function within a linear statistical model that has been tested and rejected by, for example, analysis of variance. A solution to the problems is presented. (TJH)

This article presents a system for avoiding Type I error increases when increasing the number of contrasts computed. Based on the Bonferroni inequality, the procedure corrects for the number of contrasts tested. Although conservative, the Bonferroni system is recommended for its flexibility, simplicity, and generality. (Author/BS)

While Bonferroni procedures control the risk of Type I errors, their cost is loss of power. Ordered Bonferroni procedures conserve power for more important tests while sacrificing power for less important tests. Both costs and benefits should be considered when choosing weights for individual tests and the overall level of Type I error protection.…