Another Perspective on "The Proper Number of Factors" and the Appropriate Number of Steps

Abstract

We examine some of the positions put forth in various arguments related to the proper number of factors and proper number of steps when using structural equation models. Although we agree with Hayduk and Glaser (2000) on a number of points, we do disagree on the nature of determining the number of factors in a model and in the use of the 2-step strategy. We see the issue in estimating structural equation models as fundamentally a problem of appropriately specifying a model based on our theoretical concerns, then appropriately diagnosing ills in our models in a manner that is least likely to lead us astray. We report summary results from sets of simulations and provide a detailed discussion of 1 attempt to respecify a model. Our conclusions point to the inherent difficulty in respecifying models that do not fit, the similarity in results regardless of using 1-step or 2-step strategies when estimating models with only minor misspecifications, and the value of the 2-step procedure to provide slightly better clues in our search for a better fitting model.