topic Re: PROC SURVEYREG, Note on Estimate Table in Statistical Procedures

PROC SURVEYREG, Note on Estimate Table

MillerEL — Wed, 07 Jul 2021 21:10:39 GMT

Hello,

I do not have SUDAAN, so I am using SAS PROC SURVEYREG to age-adjust complex survey data (BRFSS, specifically). The following is my code to age-adjust current smokers (binary, 0=non-smoker, 1=smoker) by a binary rural/urban classification using 5 age groups. (Trying to estimate the number of current smokers in rural vs urban areas).

	PROC SURVEYREG DATA = smokingRU NOMCAR;
	STRATA  _STSTR;
	CLUSTER _PSU;
	WEIGHT  _LLCPWTRU;
	CLASS _LLCPMRU01 agegrp;
	MODEL CurrentSmoker = _LLCPMRU01*agegrp/NOINT SOLUTION VADJUST=NONE CLPARM;
	ESTIMATE 'Rural' _LLCPMRU01*agegrp 0.128810 0.182648 0.219077 0.299194 0.170271   
									   0 		0 	     0 		  0        0		;
			 'Urban' _LLCPMRU01*agegrp 0 		0 	     0 		  0 	   0   
									   0.128810 0.182648 0.219077 0.299194 0.170271 ;
	RUN;

The estimates I get are reasonable; however, I keep getting the following note in my log:

"NOTE: The structure of the Estimates table has changed from an earlier release of SAS."

The most recent code example I could find for this methodology used SAS 9.2. I assume there has been a change in the syntax, but when I look through the documentation, I'm not seeing where the problem is.

Any thoughts? Can anyone else see what I'm missing?

Thanks!

Re: PROC SURVEYREG, Note on Estimate Table

StatDave — Wed, 07 Jul 2021 21:20:06 GMT

I believe that that message refers to the same change as occurred with PROC GENMOD as described in this note.

Re: PROC SURVEYREG, Note on Estimate Table

Reeza — Wed, 07 Jul 2021 21:35:33 GMT

Its a NOTE, not a WARNING/ERROR, so nothing is wrong it's just warning you that something has changed in case you were trying to compare to prior results.

Re: PROC SURVEYREG, Note on Estimate Table

gcjfernandez — Thu, 08 Jul 2021 06:18:15 GMT

Proc Surveymeans in SAS has poststratification weight adjustment and you can use it for age standardization.

Please refer the following: https://support.sas.com/rnd/app/stat/examples/poststrata/poststrata.pdf

Re: PROC SURVEYREG, Note on Estimate Table

MillerEL — Thu, 08 Jul 2021 13:29:48 GMT

Yes, I have a single variable code example for this as well, but it's not working right when I try a crosstab. Unfortunately, I don't have the luxury of learning a new methodology right now. Also, with either method, the estimates still do not match SUDAAN . Proc Surveyreg has, at least in the past, been endorsed by the CDC as a reasonable non-SUDAAN alternative and the results are almost always within .2% of SUDAAN. Also, I already have proc surveyreg code for crosstab adjusting that I understand. But I really do appreciate you taking the time to provide the information, and I'll look into it further when I can!

Re: PROC SURVEYREG, Note on Estimate Table

MillerEL — Thu, 08 Jul 2021 13:33:27 GMT

Thank you!