topic Re: PROC IRT Person/Item Infit-Outfit Statistics in Statistical Procedures

PROC IRT Person/Item Infit-Outfit Statistics

eewinters — Sat, 25 Feb 2017 04:15:21 GMT

I am currently evaluating using SAS as a tool for work in psychometrics (running University Edition/SAS Studio currently), and I'm really loving the ease of the SAS PROC IRT Code. I'm wondering, is there no way for users to have SAS generate item infit and outfit statistics for IRT models? I can live without some of the overall model fit indicies, but item and person infit and outfit statistics are kind of important.

I've found some work looking at the use of PROC GLIMMIX getting infit and outfit statistics for the Rasch model, but this seems 1. super convoluted and 2. super limited.

Thanks!

Re: PROC IRT Person/Item Infit-Outfit Statistics

Reeza — Sat, 25 Feb 2017 06:36:13 GMT

Did you try ITEMFIT or ITEMSTAT in the PROC IRT statement?

Im not familiar with PROC IRT but I can read the docs 🙂

Usually, checking for options on the model and proc statement are good places to start. The next is to search for papers on lexjansen.com which contains user written papers on SAS topics.

If it turns out you can't do something you can always add it to the ballotware as a suggestion for the next update.

Re: PROC IRT Person/Item Infit-Outfit Statistics

t_n5 — Sat, 05 Jan 2019 06:01:30 GMT

I am also wondering the same thing. I can't figure out how to find infit and outfit with PROC IRT. Did you figure it out ever?

Re: PROC IRT Person/Item Infit-Outfit Statistics

Reeza — Sat, 05 Jan 2019 06:41:07 GMT

Those options don’t work? You may want to start a new thread, this one is almost two years old and it likely won’t be seen.

Re: PROC IRT Person/Item Infit-Outfit Statistics

t_n5 — Sat, 05 Jan 2019 07:17:18 GMT

Thanks for replying! Unfortunately ITEMFIT and ITEMSTAT only give Chi-square fit indices. From what I have found, I might not be able to get Infit and Outfit from this.