[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: Testing means of subgroups vs mean of full sample using svymean
Thank you for your insights -- I didn't think about that... The reason why I
wanted to do this in the first place was mainly for presentation-purposes: that
way, I could have the means of the different variables in the rows with the
various sub-samples in the columns of the table and then compare them with the
full sample means of the variables, which would then be in the right-most
column. That was a great idea (or so I thought) but if it doesn't make sense
<email@example.com> To: <Statalist@Hsphsun2.Harvard.Edu>
Sent by: cc:
owner-statalist@hsphsun2. Subject: Re: st: Testing means of subgroups vs mean of full sample using svymean
09/22/2003 10:03 AM
Please respond to
The difficulty here is that the two means are correlated because the mean of
the whole includes the subgroup. So perhaps it doesn't really make sense to
do this. A more common approach is to test each subgroup against the
remainder of the whole. If multiple such tests are used, a correction like
Scheffe or Bonferroni might also be appropriate.
Arnold H. Levinson, Ph.D.
Associate Scientist, Center for Behavioral and Community Studies
AMC Cancer Research Center
Assistant Professor of Preventive Medicine and Biometrics
University of Colorado School of Medicine
> Date: Mon, 22 Sep 2003 00:33:20 -0400
> From: Nblunch@worldbank.org
> Subject: st: Testing means of subgroups vs mean of full sample using
> Dear Stata-listers,
> I am probably missing something here but after spending quite some time
> trying to figure this out, I reluctantly decided to throw the towel in the
> Specifically, I want to test the mean(s) of sub-samples vs the mean of the
> sample using the "svymean"-command. For example, say that I am interested
> the mean income of literates vs the mean income for the sample as a whole.
> can test the means of the literate and illiterate sub-groups against each
> by, say (following the "Survey Data" manual for Stata 8, p. 16), where the
> variable literate is dummy-coded:
> . svymean income, by(literate)
> . lincom [income]1 - [income]0
> But how do I test the mean income for literates, say, against the mean
> for the entire sample?
> Thank you very much in advance!!
> Niels-Hugo Blunch
> World Bank
> HDNSP, MSN G7-703
> 1818 H Street, N.W.
> Washington D.C. 20433, U.S.A.
> Phone: (+1) 202 473 5599, Fax: (+1) 202 522 3252
> Email: firstname.lastname@example.org
> Social Protection on the Web: http://www.worldbank.org/sp
> The Labor Markets Group in SP: http://www.worldbank.org/labormarkets
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
* For searches and help try:
* For searches and help try: