Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: stpiece - general hazard estimation question


From   "Bradley, Steve" <Steve_Bradley@baylor.edu>
To   "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu>
Subject   st: stpiece - general hazard estimation question
Date   Thu, 30 Jul 2009 16:54:27 -0500

Statlisters,

I have a question regarding the estimation of hazards when there isn't one for a particular time period.  In my first time period for my firms, I have no exits so the hazard coefficient is very low.  Unfortunately, because there are no exits it also indicates it is non-significant. (p=.969).

						
_t	Coef.		Std. Err.	z	P>z							
TP1	-20.66249	528.1418	-0.04	0.969	
TP2	-5.362681	.8858187	-6.05	0.000	
TP3	-5.62141	.7207315	-7.80	0.000		


As an experiment, if I add one firm exit during time period 1 to confirm this is an estimation problem I get:

          _t |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
  TP1		 |  -7.638006   1.206193    -6.33   0.000     -10.0021   -5.273911
  TP2 	 |  -4.855396    .782516    -6.20   0.000      -6.3891   -3.321693
  TP3		 |   -5.48794    .711463    -7.71   0.000    -6.882382   -4.093498
          

I am wondering if there is an appropriate solution to this problem.  Given there are no standard errors if there is not an exit, can I consider the standard error 0 and use the original coefficient or is it unreliable as well?

Thanks!


Steven W. Bradley, Ph.D.
Department of Management & Entrepreneurship
One Bear Place #98006
Baylor University
Waco, TX  76798
(office) 254.710.3921
steve_bradley@baylor.edu



*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index