Anyone know how to get 95% CI for a C-index (Logistic Regression)

[CraigSmith]

Hi all,

Does anyone know if there is a way to code for a 95% CI for a C-index when doing PROC Logistic ???

As my work is on risk models I must report the statistical uncertainty of the C-index (main measure of interest). 

Many thanks,

Craig 

[Ksharp]

That is the ROC index. Check option "rocci".

proc logistic data=sashelp.class;
model sex=weight height/rocci ;
run;

[CraigSmith]

Thank you - so running this will provide a confidence interval now for whatever model I run? 

[StatDave · Accepted solution]

Yes, or you can use one or more ROC statements as illustrated in the example titled "Comparing Receiver Operating Characteristic Curves" in the Examples section of the PROC LOGISTIC documentation.

[Ksharp]

Yes. As StatDave said.