Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Svy commands for analysis of two populations with different survey designs


From   "Hillgrove, Tessa (SA)" <THillgrove@arcbs.redcross.org.au>
To   "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu>
Subject   st: Svy commands for analysis of two populations with different survey designs
Date   Fri, 26 Mar 2004 10:35:30 +1100

I have a question relating to the svy commands in Stata. Is there a way to
perform analysis between two populations while taking into account the
different survey design used select the samples; or do I need to ignore my
survey design when performing analysis between the populations? 

The survey questionnaire for both groups was identical. My problem is that
my two different populations have different survey sampling designs.
Population 1 sample was a proportionally stratified random sample, selected
from the electoral roll to reflect the current estimates of the age and sex
distribution in the population, with a total of 12 strata. Population 2
sample was random sample from the entire lists of registered physicians in
specified categories (eg GPs, surgeons), with a total of 6 strata. 

The problem is that I want to perform analysis comparing variables in both
populations, and I'm not sure how to do this while taking into account
different survey designs used between the two populations (ie using the
sample weightings). The weightings and the finite population correction
variable and have been calculated for the two samples independently, but the
data from the two populations have been combined into one dataset for
analysis between populations (each strata group has a unique number).
However, some unexpected numbers have made me think that I am not
calculating these correctly. For example, when using the command svytab, the
"Population size" is equal to the combined total of the two populations,
even though they are not really independent (as the Population 2 is also
part of the Population 1). This makes me suspect that calculations using the
svy commands are assuming the same survey design was used for the whole
dataset, when there were two different designs used.
I would appreciate any suggestions about how I can get around this
difficulty.
Thanks!



Tessa Hillgrove
Project Officer
Quality and Business Systems
Australian Red Cross Blood Service (SA)
301 Pirie Street Adelaide South Australia 5000
Phone: (08) 8422 1364
Fax: (08) 8422 1300 
thillgrove@arcbs.redcross.org.au





E-mail Disclaimer:
Any personal information in this e-mail must be handled in accordance with
the provisions of the Privacy Act 1988 [Cth].  The information contained in
the e-mail may be confidential and if the person receiving the e-mail is not
the intended recipient they should immediately advise ARCBS by e-mail
response to the ARCBS sender and then deal with the e-mail as directed by
ARCBS. The views expressed in the e-mail are those of the individual sender
unless otherwise stated to be the views of ARCBS. No warranties are provided
that the e-mail is computer virus or other defect free.
*
*   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/



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