Statalist


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

Re: st: -bs- with the xi: prefix and -gologit2-


From   Sara Mottram <s.mottram@cphc.keele.ac.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: -bs- with the xi: prefix and -gologit2-
Date   Fri, 29 Feb 2008 15:43:32 +0000

Thank you to Maarten and Richard for their helpful comments. I now have the estimates of bias.
Best wishes,
Sara

Maarten buis wrote:

--- Sara Mottram <s.mottram@cphc.keele.ac.uk> wrote:

I am trying to get an estimate of bias in the regression coefficients
from a partial proportional odds model that I have fitted using -gologit2-. A colleague tells me that she has previously obtained an estimate of bias from the -logit- function using -bs- in Stata 7. I
can replicate this (with -logit-) as long as I don't have any
indicator variables (i.e. I don't use the xi: prefix) in Stata 9.2.
However, the output I get when I use -bs- with -gologit2- doesn't
make much sense, I get too many sets of coefficients (5, when I only
3 levels to my dependent variable - see output below).

. bs "gologit2 cohortsev agegrp1 agegrp2 agegrp3 gender ass_4 rad_k ffdef knflex, npl(gender)" _b _se
First, off all you are both bootstrapping the regression coefficients
and their parametrically estimated standard error (you are computing
the standard error of the standard error). This is legal (the standard
error is an estimate so there will be sampling error around that
estimate as well), but I don't think that that is what you want, which
may explain why you get "too many sets of coefficients". To avoid
estimating the standard error of the standard error you will have to
remove _se from the command.

Second, you do not need to use -bs- if you want to see the bias. You
can see the bias after the -bootstrap- command by typing -estat bootstap- after you have used -bootstrap-. So this way you can
use the more up to date implementation of the bootstrap.

Third, just as the standard error is an estimate and has itself a
standard error, so does the bias, and Efron and Tibshirani (1993, pp.
138) warn that the uncertainty around the bias can be large. So, be
careful.

Hope this helps,
Maarten

Bradley Efron and Robert J. Tibshirani (1993) An Introduction to the
Bootstrap. Chapman & Hall/CRC.


-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434

+31 20 5986715

http://home.fsw.vu.nl/m.buis/
-----------------------------------------


__________________________________________________________
Sent from Yahoo! Mail.
A Smarter Inbox. http://uk.docs.yahoo.com/nowyoucan.html
*
* 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/
--
Sara Mottram	
Research Assistant: Biostatistics
Primary Care Musculoskeletal Research Centre
Primary Care Sciences
Keele University
Staffordshire, ST5 5BG
Tel:  	+44 (0) 1782 584711
Fax:  	+44 (0) 1782 583911
Email:	s.mottram@cphc.keele.ac.uk

*
*   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