Simulating Omitted Variables in Structural Equation Modeling: Sensitivity Analysis With Brute-Force Search Algorithm

• In Event: Sensitivity Analysis in Structural Equation Modeling

Abstract

When missing certain variables, a structural equation modeling (SEM) analysis could lead to a completely different interpretation and inference (Stavola et al., 2014; Harring et al., 2017). Omitting related variables from SEM is one of the major misspecifications but an overlooked problem in misspecification literature (Kaplan, 1990). The purpose of this sensitivity analysis is to reveal to what extent the SEM model is robust to unmeasured variables. The algorithm we proposed in this study was based on the Brute Force search. We systematically enumerated all possible omitted variables and checking whether each enumerated variable satisfies the condition that is markedly altering the conclusions of SEM analyses.

Authors

• Huan Kuang, Florida State University

• Zeyuan Jing, University of Florida