Sensitivity Analysis to Omitted Confounders in Structural Equation Modeling Using Tabu Search

• In Event: Sensitivity Analysis in Structural Equation Modeling

Abstract

Sensitivity analysis is important for education research because it can show how vulnerable a model is to omitted confounders. The purpose of this study is to show how the Tabu search (TS) algorithm can be used to conduct a sensitivity analysis for structural equation models (SEM).
Currently available methods for conducting sensitivity analysis for SEM are limited because they require a researcher to pre-determine possible patterns of path coefficients between a phantom variable and other variables in the model (e.g. Harring, McNeish, & Hancock, 2017). Tabu search can overcome this requirement of manual specification for sensitivity analysis because they can algorithmically search for the smallest set of paths between an omitted confounder and the model’s variables that would change authors’ conclusions, without needing to search through all possible combinations of paths.

Authors

• King-Yiu Suen

• Katerina M. Marcoulides, University of Minnesota