Abstract:The homogeneity of regression coefficients is one of the most important assumptions for appropriate of ANCOVA (Huitema, 1980). However, this assumption is not always met, but usual to reveal the heterogeneity of slopes, and thus leading to biased statistical testing results. Hence, an alternative approach to ANCOVA is required. In the present paper, JohnsonNeyman technique is briefly introduced including a sequence of statistical tests and Statistical Package Syntax. he homogeneity of regression assumption is one of the most important assumptions underlying the appropriate use of ANCOVA (Huitema, 1980). When this assumption is violated, the ANCOVA is inappropriate way to use. In the following numerical example, the ANOVA F is not statistically significant. If the homogeneity of regression assumption is not tested, the conclusion that there are no treatment effects among groups is erroneous. The biased Fcal is the result of violating the ANCOVA assumption. JohnsonNeyman is generally an appropriate alternative analysis. As Johnson and Fay (1950) pointed out, Modern statistical procedures are primarily concerned with two basic problems –determining significance and estimating parameter values. The Johnson-Neyman technique may be used for both problems; however, its primary contribution is the determination of significance of differences in groups, and its unique and additional contribution is in defining the population, in terms of control variables, for which the conclusion of [statistically] significant difference of mean performance may be held. (p. 349) In the present paper, a numerical example was presented to further explain the concept of J-N technique. The practical application of this procedure in the present paper was limited to two groups and one covariate. The system of tests described below was based on the procedure by Pedhazur (1997). The numerical example is based on the design matrix by Huitema (1980) shown in Table 1. The data in Table 1 are based on the experiment introduced by Huitema. The treatments are the two methods of therapy. The covariate is the scores on the sociability scale. The dependent variable is the aggressiveness score. Dummy code is used to categorize the group membership. The overview procedure includes: 1) conduct a full regression model to test whether the main effects and interaction effects are sufficiently meaningful or not, and to obtain the sum of squares for calculating the limits of region of significance; 2) test the homogeneity assumption of slopes by testing the significance of interaction between group and covariate; 3) establish the region of significance.

Parametric Bootstrap Procedures for Three-Factor ANOVA and Multiple Comparison Procedures with Unequal Group Variances

Multiple Comparisons of Treatment Against Control under Unequal Variances Using Parametric Bootstrap

A Parametric Bootstrap Approach for One-Way ANOVA Under Unequal Variances with Unbalanced Data.

A Parametric Bootstrap Test for Two-Way ANOVA Model Without Interaction Under Heteroscedasticity

Parametric Boostrap and Objective Bayesian Testing for Heteroscedastic One-Way ANOVA

A note on extending Scheffé's modified multiple-comparison procedure to other analysis situations

Bootstrap tests for simultaneous monotone ordering of effects in a two‐way ANOVA

Parametric and nonparametric bootstrap methods for general MANOVA

Simultaneous confidence intervals for pairwise multiple comparisons in a two-way unbalanced design with unequal variances

Testing against ordered alternatives in one-way ANOVA model with exponential errors

Bootstrap Resampling Approaches for Repeated Measure Designs: Relative Robustness to Sphericity and Normality Violations

Analysis of small sample size studies using nonparametric bootstrap test with pooled resampling method

A bootstrap semiparametric homogeneity test for the distributions of multigroup proportional data, with applications to analysis of quality of life outcomes in clinical trials

Wild Bootstrapping Rank-Based Procedures: Multiple Testing in Nonparametric Split-Plot Designs

Multivariate nonparametric methods in two-way balanced designs: performances and limitations in small samples

Rank-Based Procedures in Factorial Designs: Hypotheses about Nonparametric Treatment Effects

Heteroscedastic two-way ANOVA under constraints

Comparison of Software Algorithms for Calculating Reml Wald Type Confidence Limits for the Between-Group Variance Component in A Small Sample One-Way Random Effects Model Example

Combination rules for homoscedastic and heteroscedastic MANOVA models from multiply imputed datasets

Testing two variances for superiority/non-inferiority and equivalence: Using the exhaustion algorithm for sample size allocation with cost

A Primer on the Johnson-Neyman Technique : An Alternative Procedure to ANCOVA