A first fit index on estimation accuracy in structural equation models

Abstract

Structural equation modeling (SEM) is commonly used in marketing and the social sciences such as education, psychology, and economics. A prominent issue thereby is the assessment of model fit. Although there is a huge variety of fit indices, researchers do not know how well they correspond to the estimation accuracy in terms of discrepancy between the estimated model values and true values of the respondents. By means of a comprehensive simulation study, we show in this paper that our newly developed estimation accuracy fit index (EAFI) evaluates this kind of estimation accuracy significantly better than existing indices and thereby comes closest to the actual relations between the model’s variables. Furthermore, we derive context-specific EAFI orientation points. Our findings provide guidance for researchers in management, psychology, marketing, education, and the medical sector to evaluate the model fit.