Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: AW: Two weights at the same time?


From   Steve Samuels <sjsamuels@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: AW: Two weights at the same time?
Date   Tue, 17 Aug 2010 10:08:43 -0400

The titles are similar but the problem is not. Daniel, I'm not
familiar with the GSS, but I have looked at:
http://publicdata.norc.org/webview/velocity?study=http%3A%2F%2Fpublicdata.norc.org%3A80%2Fobj%2FfStudy%2F4697&node=30&mode=documentation&submode=ddi&v=2&doprint=true.

The advice below is based on quick reading of that page, and I could
well be wrong. I suggest that you contact someone who has analyzed the
survey data for these years and ask them to send you the code they
used to calculate the weights. In Stata, you should use probability
weights ("pweights") and not analytic weights.

The study does not provide a base household weight, because all
households have the same weight. The FORMWT variable is relevant only
to certain forms. For analyzing variables on those forms, you create a
"final" weight equal to FINAL_WT= FORMWT x ADULTS; for other forms,
FINAL_WT = ADULTS If you are analyzing the ISSP questions on Form 1,
you might want to create a weighting factor to adjust for
non-response; see the section on the ISSP variable. You might also
create post-stratification weights to compensate for the
under-sampling of males.

These weights are for persons. If you are analyzing household
characteristics or HH summaries of individual responses, none of them
applies. You would create a household weight equal to 1 for those
analyses, but see the section on "Weights for 2004-06 GSS".

Good introductions to weighting are in: Groves, Fowler et al., Survey
Methodology, 2nd Ed, 2009 (Wiley); in Levy and Lemeshow, Sampling of
Populations, Wiley, 2009; and in Sharon Lohr, Sampling, Design and
Analysis: Brooks-Cole, 2009.

Steve

-
Steven Samuels
sjsamuels@gmail.com
18 Cantine's Island
Saugerties NY 12477
USA
Voice: 845-246-0774
Fax:    206-202-4783


On Tue, Aug 17, 2010 at 7:13 AM, Martin Weiss <martin.weiss1@gmx.de> wrote:
>
> <>
> Sounds very similar to Jochen`s recent thread at
> http://www.stata.com/statalist/archive/2010-08/msg00121.html
>
> HTH
> Martin
>
> -----Ursprüngliche Nachricht-----
> Von: owner-statalist@hsphsun2.harvard.edu
> [mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von Daniel Goya
> León
> Gesendet: Dienstag, 17. August 2010 13:05
> An: statalist
> Betreff: st: Two weights at the same time?
>
> Hello. I think this question might be more pure statistics than
> stata-related, but hopefully you can help me out please.
> I'm currently working with the GSS (1985 and 2004), and have an issue
> with weights (for simple OLS regressions). The 2004 data includes a
> variable that explicitely takes into account two things: subsampling
> of a group due to two-stage stratification, and the number of adults
> in the household.
> My problem is that with the 1985 data I'm working, I also need to
> correct for two things: people in household, and a form randomization
> mistake from that year. But I have two independent weight variables.
> (formwt and adults in 1985 gss). If my understanding is right, I
> should use analytical weights for both. Is there any way to use two
> weights at the same time? Is it valid at all to simply multiply them?
> Am I talking nonsense?
> I hope that you can help me out please, thanks in advance
>
> Daniel
>
> ps:I'm definitely studying some sampling before going into more
> empirical work...
> *
>
-

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index