exploratory within item multidimensional IRT models

[Asghar Minaei]

Hello Phil,

I am currently working on an exploratory within item multidimensional IRT analysis. When I ran my analysis, I noticed that the slope parameters (*a*) of all items across the three dimensions were negative. This issue often occurs due to the default identification constraints used by the mirt software, and  one way to address it is to change the default settings.

Could you kindly explain how I can modify the default identification constraints in the *mirt* package for an exploratory within item multidimensional analysis? Your guidance would be extremely helpful to me. My codes are:

fit_compensatory <- mirt(data = data, model = 3,

                 itemtype = "2PL", method = "MHRM", 

                 SE = TRUE, rotate="oblimin")

                 

extract.mirt(fit_compensatory, what="converged")

params <- coef(fit_compensatory, simplify = TRUE)

M2(fit_compensatory)

With best regards,

Asghar 

[Phil Chalmers]

Hi Ashghar,

The item EFA model you have here is rotationally indeterminate, so you can always multiply the slopes by -1 and achieve the same model-data fit. Note that coef() also accepts a rotate argument, which might explain the nature of the problem as mirt() no longer accepts rotation (this is a post-hoc operation, so has since been separated into functions like coef() and most commonly summary()). HTH.

Phil

--
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 visit https://groups.google.com/d/msgid/mirt-package/c73386f6-43e7-48ca-bd6d-814925dbef33n%40googlegroups.com.