Thank you all for the thoughtful notes and helpful suggestions. I really appreciate them very much!
Rosie
----- Original Message ----
From: "sjsamuels@gmail.com" <sjsamuels@gmail.com>
To: statalist@hsphsun2.harvard.edu
Sent: Wed, February 24, 2010 4:33:33 PM
Subject: Re: st: RE: Descriptives using multiple-imputed complex survey design data
Rosie asked: ..."is it a common practice to do descriptive analysis
(mean, proportion, cross-tabs) using all design features, including
weight, strata, and cluster variables? I had assumed that it
is....But then do people use the descriptive statistics to make
inference to the population?"
It depends on the purpose of your study. For most studies, I describe
the *sample* as is, so that people can see the n's. If the purpose
of the study is describe the population, then I also use -tab- -mean-
-prop- survey commands with full features.. But frequently, the
purpose is analytic (regression, testing associations) and much is
known about the population (from a census, for example). In that
case, there is no reason to describe the population.
I suggest that you read a good introductory survey text, for example
Sharon Lohr's Sampling: Design and Analysis.
*
* 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/
*
* 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/
