Applying Tests of Equivalence for Multiple Group Comparisons: Demonstration of the Confidence Interval Approach.

Abstract

Assessing the comparability of different groups is an issue facing many researchers and evaluators in a variety of settings. Commonly, null hypothesis significance testing (NHST) is incorrectly used to demonstrate comparability when a non-significant result is found. This is problematic because a failure to find a difference between groups is not equivalent to showing that the groups are comparable. This paper provides a comparison of the confidence interval approach to equivalency testing and the more traditional analysis of variance (ANOVA) method using both continuous and rating scale data from three geographically separate medical education teaching sites. Equivalency testing is recommended as a better alternative to demonstrating comparability through its examination of whether mean differences between two groups are small enough that these differences can be considered practically unimportant and thus, the groups can be treated as equivalent.