philchalmers / mirt

Multidimensional item response theory
199 stars 75 forks source link

MultiGroupClass@vcov labeling #211

Closed tg-tommcdonald closed 2 years ago

tg-tommcdonald commented 2 years ago

I'm trying to extract the covariance of the item parameter estimates of every item with vcov. However, I'm not sure about the labeling of the covariance matrix. Let's say we've just ran a 2PL model and we have 100 items and two subgroups ("female" and "male"). We have two estimated parameters per item: a1 and d. Is it true that the first 200 columns and rows belong to the focal group estimators ("male") and the second 200 columns belong to the reference group estimators ("female")?

For me I get the following labeling for column headers (400 columns): a1.1 d.2 a1.5 d.6 a1.9 d.10 a1.13 d.14 a1.17 d.18 a1.21 d.22 a1.25 d.26 ... d.772 a1.775 d.776 a1.779 d.780 a1.783 d.784 a1.787 d.788 a1.791 d.792 a1.795 d.796 a1.799 d.800.

From this I cant figure out which parameters belong to the focal estimates of item 1 and the reference estimates of item 1. Help would be greatly appreciated. Thanks!