Statistical mediation analysis in social psychology: An overview and application of the single mediator model
Abstract
Mediation has become a prevalent statistical analysis in social psychology. There are multiple ways in which researchers assess a mediated effect. This manuscript provides an overview of four methods: the Baron and Kenny (1986) method, the product of coefficients approach, percentile bootstrapping, and bias-corrected bootstrapping. These methods are then applied to a social psychology example to assess if people’s inferences of a target’s tolerance of historically marginalized groups (e.g. “Racial/Ethnic minorities”) is a mediator between condition (whether that target is described as a Trump supporter or non-political control) and stigma toward that target. In this example, all methods provided the same results. However, this is not always the case and it is the duty of the researcher to make certain they are using the correct method to assess the mediation effect.