08-19-2016 02:45 PM
I'm sas new user. I wonder if it's possible to get a 100% concordant percentage and D de Somers value=1 with a logistic regression. It seems so perfect to me!
08-19-2016 03:53 PM
Only if your data has perfect separation. Usually this happens when you accidentally include a variable that's observed with the outcome, sometimes after the fact.
I wouldn't trust the model and would be verifying it thoroughly.
PS I'm going to move this to the stats forums as it relates more to statistics.