RMSD_DIF

[Milica Kabic]

Dear Phil,

 

I have a question regarding RMSD_DIF: This index compares the empirical curves of subgroups

to the model-based curves of the entire sample. But theoretically, it should also allow for

comparing empirical and model-based curves for the entire sample as well

(i.e., not assessing DIF but total badness-of-fit), or shouldn't it? If so, wouldn't it make

sense to allow either for a model using multigroup but defining only one group or apply the

RMSD_DIF-Function toa "simple" mirt()-Object? I think, TAM supports the "single-group"-approach.

 

Best regards.

[Milica Kabic]

+ additional question: Could you pls explain the output of the function? $D1/D2 is the group identifier, but how should I read P.0 and P.1 (the more as they are equal)?

(from https://philchalmers.github.io/mirt/html/RMSD_DIF.html):

 

## $D1

##           P.0   P.1

## Item_1  0.056 0.056

## Item_2  0.018 0.018

## Item_3  0.014 0.014

## Item_4  0.019 0.019

## Item_5  0.010 0.010

## Item_6  0.021 0.021

## Item_7  0.016 0.016

## Item_8  0.017 0.017

## Item_9  0.019 0.019

## Item_10 0.018 0.018

## Item_11 0.013 0.013

## Item_12 0.014 0.014

## Item_13 0.018 0.018

## Item_14 0.009 0.009

## Item_15 0.011 0.011

##

## $D2

##           P.0   P.1

## Item_1  0.053 0.053

## Item_2  0.021 0.021

## Item_3  0.013 0.013

## Item_4  0.016 0.016

## Item_5  0.008 0.008

## Item_6  0.022 0.022

## Item_7  0.017 0.017

## Item_8  0.016 0.016

## Item_9  0.017 0.017

## Item_10 0.014 0.014

## Item_11 0.021 0.021

## Item_12 0.013 0.013

## Item_13 0.011 0.011

## Item_14 0.004 0.004

## Item_15 0.014 0.014

Thank you.

Best regards,

Milica

V V sre., 16. apr. 2025 ob 09:44 je oseba Milica Kabic <kabicm...@gmail.com> napisala:

[Phil Chalmers]

Hi Milica,

I'm not sure how this would work with a single group as the RMSD is computed from a 'pooled' model first (all parameters constrained to be equal). If TAM is returning an RMSD for a single-group model this is likely something different.

Regarding the output, the list is reporting the marginalized RMSD values per item category, which for dichotomous items will be the same in both categories (for polytomous items you'll see a difference). 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/CAHWaCoh8tbgmDvHR_TB2uQs5sf9HMRzTDfCW2QjkC3rLAbympw%40mail.gmail.com.

[Phil Chalmers]

Ah, I see what you mean now. Yes this is something that could be supported in RMSD_DIF() as well. I've pushed support for single-group models to the dev version. Thanks for bringing this to my attention. 

Phil