Sensitivity of fit indices
Xitao Fan and Stephen A. Sivo provide a very good resource for any researcher whishing to carry out a Monte Carlo Simulation within the SEM framework. The simulation design and the data conditions are not only documented in an ideal fashion, but also very accessible to read. Furthemore, they show how to cope with and, furthermore, to respond and incorporate the reviewers’ comments. The only downer is the some lack of documentation. If the authors would have published their paper in the Teacher’s Corner of “Structural Equation Modeling”, readers would have benefited from the SAS commands being used in this study. Anyway, read this paper!
One Response to “Sensitivity of fit indices”
Leave a Reply
Michael Scharkow on March 27th, 2008
At the risk of spoiling the fun: The key finding in this paper is that out of the 5234126 fit indices available in the notorious software, only RMSEA and Gamma seem to be indifferent to all kinds of undesired condititions, like sample size and model complexity. Since Gamma is not easily available in AMOS or MPlus, the winner is … RMSEA which has been my favorite all along