Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: svy, bootstrapping and pweights


From   "Stas Kolenikov" <[email protected]>
To   [email protected]
Subject   Re: st: svy, bootstrapping and pweights
Date   Thu, 17 Apr 2008 21:43:15 -0500

Setting up the collection of bootstrap weights to be used with Jeff
Pitblado's -bs4rw- IS the best method, if not the only appropriate
one. Check out http://www.citeulike.org/user/ctacmo/article/1036966
which is not easily available though; I can send a pdf file off the
list. It builds upon
http://www.citeulike.org/user/ctacmo/article/582039 which should be
easier to find. Read up; resampling with survey data is full of
strange complications.

I have a basic Mata code that produces the bootstrap weights in a
one-stage (stratified) survey situation. For more complex settings,
more work needs to be done...

There aren't that many occasions where the bootstrap is the last
resort. For a lot of Stata commands, -svy- versions are available
(provided you have PSU and strata information), and for most results
based on them, -lincom- or -nlcom- will give proper answers for linear
and nonlinear combinations of parameters. Sometimes, the data
providers want to hide this information for confidentiality reasons,
but then they would provide those resampling weights with the public
micro data. If you have your own data, you should not have problems
with identifiers.

On 4/17/08, Henry An <[email protected]> wrote:
> Does anyone know how to incorporate probability weights with
>  the bootstrap function in Stata? That is, I don't want the
>  program to randomly select observations from my sample, I
>  want it to select it based on probability
>  weights that I've assigned to each observation. I know there
>  is a "bswreg" ado file that does it with bootstrapped
>  weights but I don't have bootstrapped weights, I have
>  pweights. I essentially want to calculate valid standard
>  errors (or MSEs) using data from a complex survey that was
>  stratified and that comes with pweights in the dataset. Any
>  help is much appreciated.
>
>  Henry
>  *
>  *   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/
>


-- 
Stas Kolenikov, also found at http://stas.kolenikov.name

Small print: Please do not reply to my Gmail address as I don't check
it regularly.
*
*   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 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index