Relative Performance of Rescaling and Resampling Approaches to Model Chi Square and Parameter Standard Error Estimation in Structural Equation Modeling

Abstract

Though common structural equation modeling (SEM) methods are predicated upon the assumption of multivariate normality, applied researchers often find themselves with data clearly violating this assumption and without sufficient sample size to use distribution-free estimation methods. Fortunately, promising alternatives are being integrated into popular software packages. For estimating model chi square values and parameter standard errors, EQS (P. Bentler, 1996) combats the effects of nonnormality by rescaling these statistics. AMOS (J. Arbuckle, 1997), on the other hand, offers bootstrap resampling approaches to accurate model chi square and standard error estimation. The current study is a Monte Carlo investigation of these two methods under varied conditions of nonnormality, sample size, and model misspecification. Accuracy of the chi square statistic is evaluated in terms of model rejection rates, while accuracy of standard error estimates takes the form of bias and variability of the estimates themselves. An appendix provides data for the paper's figures. (Contains 2 tables, 5 figures, and 31 references.) (SLD)