A Simulation Paradigm for Evaluating Approximate Fit
In preparation for the 2010 conference held in honor of Michael Browne, I checked online to ﬁnd out which paper is most often cited from among the many contributions he has made to the quantitative methods literature. I was initially surprised at the result: The chapter Browne and Cudeck (1993) had well over 6,000 citations. This book chapter describes the properties of the root mean square error of approximation (RMSEA) ﬁt index, gives advice on how the index might be used, and illustrates its use. The chapter is the source for the widely used (.05, .08) RMSEA cut-points for evaluating model ﬁt in structural equation modeling. In retrospect, the high citation rate for this publication should not have been surprising. The number of ﬁt indices used in structural equation modeling has multiplied, starting with the chi-square test of exact ﬁt, and expanding to the several dozen ﬁt indices now available in standard software. The growth in the number of ﬁt indices has led to a corresponding growth in the demand for standards for judging model ﬁt. If we are not going to depend solely on the chi-square test of exact ﬁt, which indices should be used, and what values of these indices should deﬁne acceptable ﬁt?