topic Re: compare Gwet's AC1 in Statistical Procedures

compare Gwet's AC1

AudStat — Mon, 28 Sep 2020 13:15:46 GMT

Dear all,
I need to compare 2 agreement coeficients (Gwet's AC1) to know if diagnostic value is better than another.
Does someone know how to compute it with sas ?
Thank you in advance for your help

Re: compare Gwet's AC1

Watts — Mon, 28 Sep 2020 14:15:40 GMT

The AGREE(AC1) option in PROC FREQ produces Gwet's AC1 agreement coefficient.

proc freq data=have;
    tables rater1 * rater2 / agree(ac1);
run;

Re: compare Gwet's AC1

AudStat — Mon, 28 Sep 2020 14:37:38 GMT

Thank you very much for your answer but I 'm looking for a method to test the difference between two AC1 : Indeed, I have an AC1 (with 95%CI) (Rater1 *Rater 2) that I would like to compare with another AC1 (with 95%CI) (Rater1 * Rater 3) . I already calculate each AC1 with the proc freq procedure...

Thank you very much for your help

Re: compare Gwet's AC1

Watts — Mon, 28 Sep 2020 15:38:42 GMT

Depending on the framework (e.g., are the same subjects being rated, etc.), here are a couple of resources.

SAS Note and the MAGREE macro for more than 2 raters:

Compute estimates and tests of agreement among multiple raters

Gwet (2016). "Testing the Difference of Correlated Agreement Coefficients for Statistical Significance." Educational and Psychological Measurement 76(4): 609-637.