07-20-2017 09:30 AM
Hi, I am wondering how I can modify the type of confidence intervals that are produced for the RISK statement within the SURVEY FREQ procedure. I have tried using CL(type=X) as an option in the tables statement; however, this does not appear to affect the CI values for the RISK output. Any suggestions on how to do this? Here is the code I am using:
proc surveyfreq data=data_a;
tables catag3*year_a*cocyr / nocellpercent nostd row risk cl(type=logit);
As you can see, here I am requesting the logit option for confidence intervals; however, I have also tried various other options (Wald, clopperpearson, etc.) but the confidence interval ranges for the RISK output do not change.
07-20-2017 10:08 AM
The proportions are not going to change given the same sample design and weights so the Risk statistics won't change. You can request a narrower confidence interval with a larger alpha or wider with a smaller alpha. The default is Alpha=0.05.
Can you describe what is "wrong" with the default Risck limits?