[R] A Harder Score Test Question

[Sean]

Thanks so much.  You saved me a lot of programming time. FYI, here is the
code that produces what I need.


library(survival)

### Data
time<-rexp(100)
grp<-rep(0:1,times=50)
cens<-rbinom(100,size=1,p=.9)

### Score Test From iter.max=0
fit0 <- coxph( Surv(time,cens)~grp , iter.max = 0 )
vcov(fit0)*sum(coxph.detail(fit0)$score)^2

### Score Test From coxph
fit1 <- coxph( Surv(time,cens)~grp )
fit1$score

  


Thanks again,

Sean

-----Original Message-----
From: Charles C. Berry [mailto:cberry at tajo.ucsd.edu] 
Sent: Saturday, July 25, 2009 3:07 PM
To: Sean
Cc: r-help at r-project.org
Subject: Re: [R] A Harder Score Test Question

On Sat, 25 Jul 2009, Sean wrote:

> Does anyone know how get the score and information under the null from
> coxph? I know that I can get the chi-square value of the score from coxph,
> but I need the two components separately.  I have a function that computes
> the two components when I do not have ties but I would like to leverage
the
> options(ties and strata components) already available in the coxph
function.
> The function  coxph.detail has the score under the mle, but not the null.
I
> looked through the coxph code but I don't see where the score test gets
> calculated. If there is no easy way to calculate these values from coxph I
> can always program it myself.

I'd try

  fit0 <- coxph( ... , iter.max = 0 )
  fit1 <- coxph( ... , iter.max = 1 )

and see if I have the pieces I need to work it out from vcov(fit0) and
coefficients(fit1)

HTH,

Chuck

>
>
>
> Thanks in advance.
>
>
>
> Sean
>
>
>
> Sean Brummel M.S.
>
> Doctoral Candidate
>
> Dept. Statistics
>
> University of California, Irvine
>
> Irvine, CA 92797-1250
>
>
>
> Ph: 949-307-1811
>
> Fax: 949-824-9863
>
>
>
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

Charles C. Berry                            (858) 534-2098
                                             Dept of Family/Preventive
Medicine
E mailto:cberry at tajo.ucsd.edu	            UC San Diego
http://famprevmed.ucsd.edu/faculty/cberry/  La Jolla, San Diego 92093-0901