Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: Suest and Cox regression
From
Steve Samuels <[email protected]>
To
[email protected]
Subject
Re: st: Suest and Cox regression
Date
Wed, 28 Jul 2010 12:56:47 -0400
Richard, it is not true of -streg-, because that command is based on
standard likelihood theory. To give a familiar example, in multiple
linear regression, if the error term is assumed to have a Gaussian
distribution, the likelihood score for an observation is its
residual: a function of its y, x's and the estimated betas. The
equations for forming the covariance matrices in -suest- (see the
manual) don't work for -stcox- because of the non-standard scores.
Steve
On Wed, Jul 28, 2010 at 12:23 PM, Richard T. Campbell <[email protected]> wrote:
> Thanks Steve, you are, of course, course correct.
> But is this not true also for -streg-? And -suest- works nicely
> after streg. I guess I don't understand why -stcox- should be
> any different but there is obviously something I am missing here.
>
> At 11:00 AM 7/28/2010, Steve Samuals wrote:
>>
>> I'm not expert in the theory of -suest-, but after looking at the
>> equations in the manual entry, here's my guess: -suest- works when the
>> program computes a score for each observation that is a function of
>> the regression coefficients and of the observation's data. In Cox
>> regression, the score is _also_ a function of a random parameter that
>> depends on who is in the observation's final risk set.
>>
>> Steve
>>
>
>
>
>
>
> *
> * 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/
>
--
Steven Samuels
[email protected]
18 Cantine's Island
Saugerties NY 12477
USA
Voice: 845-246-0774
Fax: 206-202-4783
*
* 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/