Date: Tue, 21 Feb 2012 20:26:06 +0000 "Nordlund, Dan (DSHS/RDA)" "SAS(r) Discussion" "Nordlund, Dan (DSHS/RDA)" Re: Sampling question text/plain; charset=utf-8

> -----Original Message----- > From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of > Joe Matise > Sent: Tuesday, February 21, 2012 11:31 AM > To: SAS-L@LISTSERV.UGA.EDU > Subject: Re: Sampling question > > At the simplest, that seems right to me. Except make sure you have .73 > and > not .17 in there :) (Pop target)/(Sample Freq)=post-strat weight. > > If your'e weighting on multiple characteristics, though, it's somewhat > different. > > -Joe > > On Tue, Feb 21, 2012 at 1:25 PM, William Shakespeare < > shakespeare_1040@hotmail.com> wrote: > > > Suppose I have the following population and sample with given > > characteristics: > > > > My population Sample > > N 10000 1000 > > Black 12% 15% > > White 73% 65% > > Hispanic 15% 20% > > > > I can't remember how to apply post-stratification weights but I think > it's > > like so: > > > > .12/.15=.80 > > .17/.65=1.12 > > .15//20=.75 > > > > Am I on the right track?

You need to be careful how you analyze these data. You can't just throw the weights you calculated above into non-survey procs if you are going to do anything more than calculate some point estimates. For example, if you are going to post-weight these data to population totals and do hypothesis testing, you should be using the survey based procedures and your weights should be larger by a factor of 10.

Hope this is helpful,

Dan

Daniel J. Nordlund Washington State Department of Social and Health Services Planning, Performance, and Accountability Research and Data Analysis Division Olympia, WA 98504-5204

Back to: Top of message | Previous page | Main SAS-L page