Multiple-group analysis approach to testing group difference in indirect effects

This article introduces five methods that take a multiple-group analysis approach to testing a group difference in indirect effects. Unlike the general frameworks for testing moderated indirect effects, the five methods provide direct tests for equality of indirect effects between groups. A simulation study was conducted to examine the performance of the methods in terms of the empirical type I error rate, statistical power, and coverage of 95 % confidence intervals. The likelihood ratio test and percentile bootstrap confidence intervals are recommended. The methods are illustrated using an empirical data set.