Hello Phil,
Are the absolute fit indices (M2, RMSEA, CFA, TLI) really useful for IRT models? Why don’t researchers report them? Do they work for international large scale assessment data where there are matrix sampling and many missing values?
I estimated a unidimensional 2PL model with the TIMSS data for a country. The model converged in a few seconds but the M2 function took more than 2 hours. I stopped it and didn’t wait to see what happens. Can you please comment?
Regards
Afshin