ERIC Number: ED406405
Record Type: Non-Journal
Publication Date: 1996-Nov
Reference Count: N/A
Alpha-Max: A Novel New Multiple Comparison Procedure.
Barnette, J. Jackson; McLean, James E.
Alpha-Max is a multiple comparison procedure that is based on the Bonferroni inequality. However, rather than using preset adjusted alpha values for each pairwise contrast, Alpha-Max bases the decisions on the actual p-values for the pairwise contrasts. After a significant omnibus F-test, the pairwise p-values are determined and ordered from lowest to highest. Alpha-Max includes as being significant only those contrasts whose sum of sequential additive p-values is less than the nominal alpha desired. Results for Alpha-Max are compared with results using the original Bonferroni, Tukey's Honestly Significant Difference Procedure (HSD), and Student-Newman-Keuls procedures based on simulated three- and four-group data sets with varying mean patterns and sample sizes. In every case, Alpha-Max had more power to detect pairwise differences as compared with Bonferroni and HSD and in one mean pattern was more powerful than SNK in detecting mean differences at r=K-1 steps. Needed additional research and possible applications of this procedure are discussed. (Contains 15 tables and 24 references.) (Author/SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A