A primer on multilevel mediation models for egocentric social network data

Abstract

The substantial increase in nested data such as egocentric network data pose unique challenges to researchers, including how to adequately model the mechanisms and processes in which hypothesized effects operates within the complex interdependencies among social actors. While there is a growing interest in identifying the causal mechanisms and their contingencies using such nested structured data, the direct application of principles of (moderated) mediation analysis within traditional multilevel modeling framework is ambiguous and controversial. In this article, I provide a tutorial illustrating an approach to assessing mediation hypotheses based on path-analytic framework popularized by Hayes (2013). Focusing on estimation and inference of indirect effects with regard to nested data structure, I provide examples of the analysis within the context of egocentric network data, and illustrate implementation using Mplus software.