10-08-2013 09:33 AM
I am a new user of sas. I wanna test the interval validation, so I use proc surveyselect for bootstrap resampling.
Below is my codes:
sasfile g load;
proc surveyselect data=g out=gg
seed=123456 rep=1000 method=urs samprate=1 outhits;
sasfile g close;
proc surveylogistic data=gg ;
model y(descending) = exposure;
However, from proc surveylogistic I got a ridiculous confidence interval of exposure: (OR=1.2, 95%CI:1.2-1.2). I mean, after getting 1000 times replications, the sample size increased and the CI seemed too small.. But I used the method of unrestricted sampling without weighting, so I don't how to get the plausible standard error to estimate CI.
How to deal with that? Can anyone help? Thanks very very much~~!!
10-08-2013 09:53 AM
The other thing is, can I get AUC (ROC c-statistics) in proc surveylogistic? And the confidence interval of AUC? Any way to calculate the AUC after the bootstrap using 'proc surveyselect'?