Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Date of Award
Master of Arts (MA)
Department of Graduate Psychology
Allison Ames Boykin
S. Jeanne Horst
Researchers utilizing either experimental or quasi-experimental research often want to compare group means. However, with more than two groups, comparing group means may result in an inflated Type I error rate, the probability of wrongly rejecting a null hypothesis. Researchers often employ analysis of variance (ANOVA) methodology to compare more than two group means. Post-hoc comparison procedures (PCPs) are utilized to indicate which group means differ following a significant ANOVA. SPSS provides 18 options for PCPs. The purpose of this study was to determine which PCP provides the best power while maintaining Type I error control when assumptions of ANOVA are met and when they are not met. Data were simulated in a variety of conditions to address this issue. Only those tests designed for assumption violations, Dunnett’s C, Dunnett’s T3, Games-Howell, and Tamhane, adequately controlled Type I error in all conditions. Power results were similar for all four tests, with the Games-Howell being slightly higher than the other four tests. I recommend using the Games-Howell procedure unless extenuating circumstances exist.
Sauder, Derek, "Examining the type I error and power of 18 common post-hoc comparison tests" (2017). Masters Theses. 524.