An Approach to Sample Size Determination for Multiple Comparison Procedures in ANOVA

• In Event: AERA Poster Session 6
  In Poster Session: Best Practices in Applied Statistics

Abstract

Researchers are often concerned with what sample sizes are necessary to achieve sufficient power when designing their research studies. Sample sizes in such studies are often based on the omnibus test and do not necessarily provide sufficient power for post hoc comparisons commonly used to investigate statistically significant main effects. The purpose of this study was to conduct Monte Carlo simulations in order to investigate the appropriate sample sizes needed to achieve sufficient statistical power for numerous post hoc comparison procedures. Results show clear and consistent differences between sample sizes based on the omnibus F test and on the post hoc tests.

Authors

• Michael Senteney, Ohio University

• Gordon P. Brooks, Ohio University