Testing for mediating variables in management research: Concerns, implications, and alternative strategies

Due to the nature of the data employed and of the concepts tested, standard tests for mediating variables in management research can often violate an assumption on which these tests are built. As a result, estimates from standard tests of mediating variables can often lack desirable statistical properties and lead to incorrect conclusions. This article highlights when data in management research will violate assumptions underlying tests of mediating variables, suggests an alternative strategy that provides better estimates, discusses the theoretical and empirical demands of the alternative strategy, and demonstrates the magnitude to which estimates improve by using the suggested approach.