This is very curious. Are you sure the difference isn't caused by
-bootstrap-, rather than -parmby-? Your first example does not seem to
use the -bootstrap- prefix. By default, -stcox- presumably uses the
quasi-likelihood Hessian matrix to calculate standard errors, not the
bootstrap method. I don't know why the output SEs should be headed
"Bootstrap Std. Err".
I hope this helps.
Roger
Roger Newson
Lecturer in Medical Statistics
Respiratory Epidemiology and Public Health Group
National Heart and Lung Institute
Imperial College London
Royal Brompton campus
Room 33, Emmanuel Kaye Building
1B Manresa Road
London SW3 6LR
UNITED KINGDOM
Tel: +44 (0)20 7352 8121 ext 3381
Fax: +44 (0)20 7351 8322
Email: [email protected]
www.imperial.ac.uk/nhli/r.newson/
Opinions expressed are those of the author, not of the institution.
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Michael
McCulloch
Sent: 16 November 2006 23:55
To: Statalist
Subject: st: different results for bootstrap Cox regression with vs.
without -parmby-
Hello,
I have run the same Cox regression analysis with and without -parmby-,
with
different results. The point estimate is the same, but the bootstrap
standard errors differ. Can anybody suggest why this would happen?
without -parmby-
. stcox tcm if stageII==1
No. of subjects = 65.6225431 Number of obs =
33
No. of failures = 65.6225431
Time at risk = 3532.264538
Wald
chi2(1) = 0.66
Log likelihood = -214.28476 Prob > chi2 =
0.4154
------------------------------------------------------------------------
------
| Observed Bootstrap
Normal-based
_t | Haz. Ratio Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------
------
tcm | .7248583 .2863995 -0.81 0.415 .3341442
1.572433
------------------------------------------------------------------------
------
with -parmby-
. parmby "bootstrap, reps(1000): stcox tcm if stageII==1", ///
label command eform saving(tf_unst_II, replace) idnum(411)
idstr(Stage II) for(estimate min95 max95 %8.2f p %8.3f) ///
stars(0.05 0.01 0.001 0.0001) ///
list(idnum idstr parm label estimate min95 max95 p stars if
parmseq==1, clean noobs)
No. of subjects = 65.6225431 Number of obs =
33
No. of failures = 65.6225431
Time at risk = 3532.264538
Wald chi2(1) =
0.79
Log likelihood = -214.28476 Prob > chi2 =
0.3750
------------------------------------------------------------------------
------
| Observed Bootstrap
Normal-based
_t | Haz. Ratio Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------
------
tcm | .7248583 .2629041 -0.89 0.375 .3560612
1.475644
------------------------------------------------------------------------
------
Best wishes,
Michael
____________________________________
Michael McCulloch
Pine Street Clinic
Pine Street Foundation
124 Pine Street, San Anselmo, CA 94960-2674
tel 415.407.1357
fax 415.485.1065
email: [email protected]
web: www.pinest.org
www.pinestreetfoundation.org
www.medepi.net/meta
*
* 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/
*
* 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/
