10-19-2011 11:04 AM
Hello, I'm analyzing a large database using PROC Surveylogistic. The database has ~300,000 entries (each representing an individual), and only ~3000 belong to my target study group (individuals aged 14-22). I've created an indicator variable AGE1422 that equals 0 if the individual is not aged 14-22 and 1 if it the individual is aged 14-22. I'm not interested in modeling the dependent variable (ER visits in this case) in the non 14-22 year olds, but I know I have to use the DOMAIN statement to calculate correct variances (instead of subsetting my dataset in a datastep using the statement WHERE AGE1422 = 1, then using surveylogistic). Is there a way I can get the ODF RTF system to not output the results where the domain = 0 (i.e. for non 14-22 year olds)? It'd be even better if I could get SAS to ignore the domain = 0 observations altogether to cut donw on processing time, but I doubt that's possible.
Here's my code:
ods rtf file = 'F:\Research\temp.rtf';
PROC SURVEYLOGISTIC data = temp;
model ervisit = sex race datayear insurance;
ods rtf close;