Individual Variability as a Moderator of Structural Equation Model Parameters

• In Event: AERA Roundtable Session Sunday 11:45 am - Gold 4
  In Roundtable Session: Advanced SEM Approaches for Modeling Interaction, Heterogeneity, and Variability

Abstract

Latent random variability models are a useful framework for research questions that involve intraindividual or intragroup variability as the focal interest, rather than individual or group averages. These models use multilevel structural equation modeling to parameterize within-level variability as a between-level latent variable to be estimated as a predictor, outcome, or mediator simultaneously with other latent variables in a model. The current paper extends this technique by utilizing principles from Moderated Nonlinear Factor Analysis to allow latent variability to act as a moderator for parameters in a structural equation model. An empirical application is used to demonstrate estimation procedures in the Blimp software package and interpret relevant model parameters. The significance of moderation utilizing individual variability is then discussed.

Authors

• Joshua Shulkin, University of Maryland

• Gregory R. Hancock, University of Maryland