Hi Everyone,

I am analyzing a study that asks participants to use a COVID diagnosis rating scale on a CT scan (outcome - negative, atypical, indeterminate, typical) and the dependent variable is the actual diagnosis (COVID, flu, other pneumonia). We asked 6 readers to rate each study (150 studies, 900 total readings). We would like to see if there are differences in rating categories across actual diagnoses. Chi square tests in this case wouldn't apply because the data is clustered by reader. I was wondering if the following command would work for this study:

mlogit coviddiagnosisrating i.diagnosis, cluster(readervariable)

I'm working with Stata 11 currently.

Thanks,

Anand