G_MEZ VILLASCUERNA JAIME (liyo5gj1@lin3.smf.nottingham.ac.uk)
Fri, 11 Sep 1998 16:52:01 GMT0BST

Hi everyone!

I am estimating the Proportional Hazards Cox Model. I have checked
the Proportional Hazards Hypothesis through the introduction of
different functions of time and the covariates in the
model which I look at to see if they are statistically significant.
For example to check the presence of non proportional effects in the
covariate Z(k+1) of the type LN(t)* Z(k+1) I have estimated the
following model:

R(t,Z)=exp( function of the first k covariates + beta (0)*Z(k+1) +
beta (1)*Z(k+1)*LN(t))

I know how to interpret the result if both beta (0) and beta (1) are
statistically significant.

But what happens if the
covariate (that is beta (0)) is not significant and the time
dependent function of the covariate (that is beta(1)) is
significant?. Could be this interpreted?. How?.

Thanks a lot for your help.

Jaime Gomez.
Doctoral student.
Department of Business Economics.
University of Zaragoza. SPAIN.

Jaime Gomez.
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news