Modeling Misspecification as a Parameter in Bayesian Structural Equation Models

• In Event: Structural Equation Modeling SIG Virtual Paper Session: Applications and Methodological Investigations

Abstract

Accounting for model misspecification in Bayesian structural equation models is an active area of research. We present a uniquely Bayesian approach to misspecification that models the degree of misspecification as a parameter. The misspecification parameter can be interpreted on its own terms as a measure of absolute model fit, and allows for comparing different models fit to the same data. Additionally, the approach allows for diagnosing local misspecification. These features are confirmed using simulation studies; the simulation studies also verify that the approach permits adequate estimation of and inference about structural estimates. Finally, a data analysis example shows the potential of the approach applied to a fairly typical problem.

Author

• James O. Uanhoro, University of North Texas
  Presenting Author