According to the report "IRT in Mplus" (Nov 14th, 2013), Logit/Probit Link, ML/ULS Estimators can be used for IRT modeling. However, I tried to use ML and ULS for a binary data set, and these two estimators doesn't work. The ML and ULS estimation cannot be applied to binary data, is it correct? Here is a sample of my syntax: DATA: file is dat1.txt; VARIABLE: names are x1-x22; !categorical are x1-x22; ANALYSIS: estimator=ml; !estimator=uls; parameterization=theta logit; !link=logit; Model: Y by x1-x22*; Y@1; SAVEDATA: RESULTS ARE results.dat;
Second question: How can I save the IRT item discrimination and difficulty estimates and their SE to a txt file? The SAVEDATA function only saves the factor loadings and thresholds. Is there a way to save the IRT item parameters?