proc surveyfreq question

[agparsons]

Hi,

I am new to survey analysis  and PROC SURVEYFREQ.  Because we are using denominators that are based on survey data I have been assigned to update confidence intervals to take into account survey error. I have been given some code that uses the Standard Error for Weighted Sum (SEWGT) output from SUDAAN, but the programs that generate our denomintors use SURVEYFREQ

Does PROC SURVEYFREQ produce the Standard Error for weighted sum?

here is a piece of the SUDAAN based code I have been given;

StdErr = (Admit / PWD**2) * SEWGT;

Variance = StdErr ** 2;

Here is an example of the SURVEYFREQ code that we are currently using. does this code produce the necessary output,to plug into the above formula (admit and PWD is computed elsewhere in the code)?

proc surveyfreq data = diab_12;

tables (sex age4cat county dist_o ins) * diab_o/ row cl;

strata _ststr;

cluster _psu;

weight _llcpwt;

run;

Thanks,

Alison

[drb4511 · Accepted solution]

Yes. According to the SAS manual "The standard errors computed by PROC SURVEYFREQ are based on the multistage stratified design of the survey. This differs from some of the traditional analysis procedures, which assume the design is simple random sampling from an infinite population." See: http://support.sas.com/documentation/cdl/en/statug/68162/HTML/default/viewer.htm#statug_surveyfreq_g...

[agparsons]

Thank You ! this is really helpful.