theta_lim & EAPsum

[Dr. Hans Hansen]

Hi Phil,

i use fscores(..., method = "EAPsum") to get theta estimates for each sum score. however, at theta = 4 there is little increase in theta when the sum score increases and i suspect it is impossible for the estimate to go beyond 4. i tried to adjust with the theta_lim argument but the estimation seems uninfluenced by that (even if i narrow the band to theta = c(-1,1) the expected thetas are the same). I guess, EAPsum just doesn't take theta_lim into account, could you look into that?

Thanks a lot, best wishes, Felix

[Phil Chalmers]

Hi Felix,

Thanks, you're right I missed this option when adding in the theta_lim option a while back. It's been added, and also will affect the itemfit() S_X2 statistics when theta_lim is passed. 

You've brought this fscores problem up a couple times now. I think in the next update I'll increase both the number of quadpts and the range of the theta grid to try to avoid these flooring effects in the default cases. Cheers.

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.
For more options, visit https://groups.google.com/d/optout.

[Dr. Hans Hansen]

Great, thanks a lot!

[Dr. Hans Hansen]

When estimating one model, i get a rather strange error now:

fscores(model, method = "EAPsum")

Error in expand.table(cbind(ret$Theta, ret$SE.Theta, ret$observed)) : 
  Frequencies must be on the right of the data matrix.
In addition: Warning message:
In min(got) : no non-missing arguments to min; returning Inf

The other model works perfectly fine. Can you make sense out of it? Find the model attached.

Thanks a lot, felix

[Phil Chalmers]

Thanks for the data. Yes, I was able to reproduce it. It had to do with a necessary argument when missing data were present. Fixed now on the dev, cheers.

Phil

--