AIC-type Theory-Based Model Selection for Structural Equation Models

Structural equation modeling (SEM) software commonly report information criteria, like the AIC, for model under investigation and unconstrained/saturated model. With these (non-)nested models can be compared. This comes down to evaluating equalities (e.g., setting some paths equal or 0). These criteria cannot evaluate inequality restrictions on parameters, while AIC-type criterion called GORICA can. For example, hypothesis stating that one predictor has more (standardized) strength than other predictors. paper illustrates inequality-constrained hypothesis-evaluation in SEM using (in R). Examples will presented confirmatory factor analysis, latent regression, multigroup regression.