ERIC Number: ED219413
Record Type: RIE
Publication Date: 1982-Mar
Reference Count: 0
Selecting an Appropriate Multiple Comparison Technique: An Integration of Monte Carlo Studies.
Myette, Beverly M.; White, Karl R.
Twenty Monte Carlo studies on multiple comparison (MC) techniques were conducted to examine which MC technique was the "method of choice." The results from these studies had several apparent contradictions when different techniques were investigated under varying sample size and variance conditions. Box's coefficient of variance variation and bias coefficient were used to quantify the degree of variance heterogeneity and estimated bias of the Type I error rate for the sample and variance condition with each technique studied. Twenty-five multiple comparison techniques were covaried with study characteristics. Wide disparity in data limited recommendation of a "method of choice." Results indicated that the Kramer modification for unequal sample sizes was the preferred method when the Tukey MC technique was elected by the researcher. Neither of Box's coefficients were found to predict bias in Type I error rate. A simple t test administered after a significant omnibus F test (the "Two-stage t-test") has been obtained, appeared to be the most simple and accurate procedure for multiple comparisons. Further replications of this technique are needed to confirm its accuracy. (Author/DWH)
Publication Type: Speeches/Meeting Papers; Reports - Research
Education Level: N/A
Authoring Institution: N/A
Identifiers: Monte Carlo Studies; Multiple Comparisons; T Test; Tukey Statistic; Type I Errors
Note: Paper presented at the Annual Meeting of the American Educational Research Association (66th, New York, NY, March 19-23, 1982).