This is not a mirt question per se (I hope that's okay). I fit an item factor analysis model in lavaan using 3-point likert-type data. The model is a two-parameter (i.e., thresholds and factor loadings) graded response model that uses a WLSMV estimator and theta parameterization.
I'd like to generate item characteristic curves for the items from the model. I was thinking the most straightforward way to do this might be to translate the item parameters from an IFA metric (i.e., threshold and factor loading) to the metric of traditional IRT parameters (i.e., difficulty and discrimination, respectively). However, I'm not sure how to convert them from the IFA metric to the IRT metric. Any suggestions how to do this (or, ultimately, how to create item characteristic curves from this model) would be greatly appreciated.
Thanks in advance!
-Isaac