Statalist


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

Re: st: stcox output: p-value and CI don't agree


From   [email protected] (Jeff Pitblado, StataCorp LP)
To   [email protected]
Subject   Re: st: stcox output: p-value and CI don't agree
Date   Thu, 09 Aug 2007 09:49:24 -0500

Michael McCulloch <[email protected]> had a couple of follow-up questions:

> >The p-values in the tables reported by Stata's estimation commands,
> >including -bootstrap-, are always for tests against zero.  -stcox- reports
> >exponentiated coefficients (hazard ratios) by default, but the reported
> >tests are of the null hypotheses that the true underlying coefficients are
> >zero.

> Can either command be specified to test against the value of one?

Michael can use the -test- command to perform tests of his point estimates
against one.  Example:

	. sysuse cancer
	. stset studytime, failure(died)
	. bootstrap drug=exp(_b[drug]), reps(100) seed(12358) : stcox drug
	. test drug = 1

> >	The -force- option of -bootstrap- accomplishes the above, but I must
> >	strongly warn that it is the user's responsibility to verify that
> >	resampling observations and using the weights associated with the
> >	resampled observations is appropriate.

> Are there guidelines or examples for implementing this?

There are no guidelines other than the one stated above, basically you should
know something about the principles behind the bootstrap before trying to use
-bootstrap- in a non-standard way (this is true wether you use the -force-
option or not).

--Jeff
[email protected]
*
*   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