Statalist


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

Re: st: Correcting for SEs in complex survey design


From   Erin Hye-Won Kim <hk31@duke.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Correcting for SEs in complex survey design
Date   Sun, 22 Feb 2009 12:58:58 -0500

Hi Steve,

Thanks a lot for answering the questions. I will check the paper you referred
to.

Best,
Erin

Quoting Steven Samuels <sjhsamuels@earthlink.net>:

--


Erin:

With survey data, standard errors are based on between-PSU variability. You do not need a family level cluster variable.

If you wish to model the clustering of responses within households and to estimate the household, parent, and child contributions to variation, try -gllamm-, available from ssc. It will accommodate the survey weights and allows a cluster (PSU) variable that is not in the multilevel model.

Take a look at: Anders Skrondal1 and Sophia Rabe-Hesketh. Some applications of generalized linear latent and mixed models in epidemiology: Repeated measures, measurement error and multilevel modeling. Norsk Epidemiologi 2003; 13 (2):265-278 265

Regards,

Steve

On Feb 20, 2009, at 1:50 PM, Erin Hye-Won Kim wrote:

Dear statalisters,

I am analyzing data from a survey with complex design. The dataset  has both
clusters and strata, so I used

svyset psuid [pweight=weight], strata(stratid)

The dataset is representative of individual parents and many  parents have
multiple children. To analyze the relationship between a parent and  each of
his/her children,

I reshape the dataset from a wide to a long format as in the following

parent1 child1 child2.... ==> parent1-child1
                              parent1-child2

In regressions after reshaping, I have to correct for SEs since siblings are
correlated, so I tried the following

svy: regress Y Xs, cluster(parent_id)

STATA says "svy" cannot be used with "cluster" option. Is there any  way to
correct for SEs while circumventing this problem?
---------------------------------------------------------

In addition, I tried family fixed-effects models by controlling for parent_ids.

svy: aregress Y Xs, absorb(parent id)

Again STATA says "svy" cannot go with "areg". There are too many parent_ids to
control manually. Is there any solution to this problem?
---------------------------------------------------------

*
*   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/



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