Hello, I am running a CFA using M-Plus with categorical data and based on results from EFA. The fit indexes are within the acceptable range, but my RMSEA is too high. The modification indices do not seem to indicate anything troublesome.The scale is a 10-point Likert scale. I'm wondering how to interpret the high RMSEA value against the other cut-offs. Any recommendations? Would you please let me know possible explanations for problematic RMSEA ?
ESTIMATOR IS WLSMV Chi square = 729.669 Degrees of Freedom 14 P-Value 0.000 RMSEA = 0.23 CFI=0.97 TLI= 0.94 WRMR=1.05
I assume you have a rather large sample size to make the chi-square fit so much worse than CFI. Generally speaking, RMSEA is more in agreement with chi-square than CFI because it is also based on chi-square. CFI is a more lenient fit measure that relates not only to the model's chi-square but also to the chi-square of a model where the variables are uncorrelated.
I am surprised that the modification indices don't help you improve the chi-square fit.