s-news
[Top] [All Lists]

time dependent covariates and likelihood ratio test

To: <s-news@lists.biostat.wustl.edu>
Subject: time dependent covariates and likelihood ratio test
From: "Ariel Alonso" <ariel.alonso@fulladsl.be>
Date: Mon, 26 Jun 2006 19:45:23 +0200
Dear All:
 
I am fitting a cox model with a time dependent covariate:
 
model.surro<-coxph(Surv(T1, T2, survind)~(treat+S):strata(trial),data="">
 
where S is the time dependent covariate and surro.time is the data set with the format needed to fit this time of models in Splus. I am also fitting the model without the time dependent covariate and using the original data set (surro):
 
model.original<-coxph(Surv(surv, survind)~(treat):strata(trial),data="">
 
when I try to calculate the likelihood ratio test to compare both models I get:

-2*(model.original$loglik[2]-model.surro$loglik[2])= -25959

which is negative. I also tried to fit the model without S using the surro.time data set
 
model.without.surro<-coxph(Surv(T1, T2, survind)~(treat):strata(trial),data="">
 
and the likelihood ratio test becomes then:
 
-2*(model.without.surro$loglik[2]-model.surro$loglik[2])=837.4
 
which is positive as one would expect. However the estimates from model.without.surro and model.original differ.
 
 
Does anybody could explain what is going on here and how can I calculate the likelihood ratio statistics.
 
Kind Regards,
 
------------------------------------------------------------------------
PhD. Ariel Alonso Abad
Hasselt University
Building D
3590 Diepenbeek, Belgium.
<Prev in Thread] Current Thread [Next in Thread>
  • time dependent covariates and likelihood ratio test, Ariel Alonso <=