Estimated slope parameter order

28 views
Skip to first unread message

David Perlaza

unread,
May 17, 2018, 9:41:50 AM5/17/18
to mirt-package

Hello Dr. Chalmers,


Using random slope parameters, I generated some data using simdata. I can recover estimates of these parameters using the coef-method on a model obtained from running the mirt function on the data.

I'd just like to know if the order of estimated slope parameters matches that of the "true" parameters used to simulated the data.

For example, given a slope matrix a with two factors:

            [,1]      [,2]
 [1,] 1.08624628 1.3448732
 [2,]     ...       ...

Do a1,a2 in the output of coef-method correspond to column 1,2 of the
a matrix respectively?

$Item_1 a1 a2 d g u par 1.318 1.369 -1.205 0 1

Or is the order not preserved, only the magnitudes?



Thanks kindly in advance for your attention,
David

Phil Chalmers

unread,
May 17, 2018, 3:14:16 PM5/17/18
to super...@gmail.com, mirt-package
Correct. Only asymptotically will the order of these parameters converge to the ordered pattern in the population. It's possible to estimate these parameters under constraints that the ordering is preserved, but that's actually relatively complicated to do and generally not recommended. HTH.

Phil
 



Thanks kindly in advance for your attention,
David

--
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.
For more options, visit https://groups.google.com/d/optout.
Reply all
Reply to author
Forward
0 new messages