How to display information added by items using GRM IRT

[Kate Toft]

I'm using IRT GRM analyses for looking at item level data for a patient-reported questionnaire measuring swallowing-related quality of life. I'm using Stata v16. I've calculated item information function graphs for individual items, but would like to see/calculate numeric values for the information added by each item to the overall questionnaire (see table 3 in https://www-sciencedirect-com.ezprox...14?via%3Dihub#! for an example of what I mean).
I can only seem to generate graphs on item information rather than tables of numbers (unlike difficulty and discrimination values). How can I generate numeric data on item information rather than just graphs?
Thank you
Kate

[Weiwen Ng]

Late response, but here goes.

The link provided only functions for the user or for staff/students at Stirling University. So, I can't see what the OP meant.

I don't believe that there is a one-number value for the information added by each question. As far as I know, the information provided by each item depends on theta. This is one of the advantages of IRT over classical test theory: the standard error of your estimate of the latent trait is allowed to vary by the value of that trait, reflecting the fact that the instrument could be more accurate at measuring, say, high levels of a trait than low levels, or vice versa.

What the graph commands actually do behind the scenes is that they generate 300 values of theta ranging from -4 to +4. (NB: you can reset those defaults; for many disease symptoms the low end of the scale might not be that relevant, and I tend to do plots from -1 to +4). They will then calculate the information provided by each item at each value of theta (and thus the test's overall information function, which is the sum of all the information functions at each theta), or the probability of each response category at each value of theta (and thus the expected score for each item, or the expected total score). You can access that by the , saving(filename) option. For that matter, if you like to edit the graphs (e.g. colors, axis titles, etc), I usually find it better to use that option and then just twoway line ... , (options).