Fit statistics and Item plots in MIRT - Graded response

[Hesham Gad]

I am trying to fit a confirmatory MIRT model with six dimensions. The model is a graded response IRT mode. The results of model fit were as follows

M2 df p RMSEA RMSEA_5 RMSEA_95 SRMSR TLI CFI stats 40852.32 2678 0 0.1690169 0.1674037 0.170296 0.2572462 0.4269607 0.4461929

indicating a model misfit.

This was not the case when fitting the same model with the same data using CFA.

How can I interpret this contradiction in the results?

In addition, I got this error message:

Error: Can't plot high dimensional solutions.

When asking for plots.

How can I get the ICC and IIC for item in MIRT with high dimensional models?

[Oxana Rosca]

Adding QMC=TRUE  helps resolve many errors and warnings when working with high-dimensional models. I used it frequently when working with a bifactor model on sparse data from a multi-stage math assessment.

[Preston Botter]

While FA and IRT models have fit statistics with similar (or the exact same) names, they are different fit indices with different population values and thus should be interpreted differently. That being said, I would expect model fit information to convey similar information (e.g., your MIRT TLI is 0.43, and I would not expect your CFA TLI to be something like 0.14 or 0.94). 

P

--
You received this message because you are subscribed to the Google Groups "mirt-package" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mirt-package...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mirt-package/dfbe00f2-67e3-4d40-b04a-15d1cd6f3259n%40googlegroups.com.

[Gerry Cheung]

Hi Hesham,

Did get the answer of this question? I have the same question.

Gerry