s-news
[Top] [All Lists]

[S] survreg question

To: s-news@wubios.wustl.edu
Subject: [S] survreg question
From: Brian Beckage <bb2@duke.edu>
Date: Fri, 31 Dec 1999 00:54:38 -0500
Sender: owner-s-news@wubios.wustl.edu
Hi,
        I'm fitting a Weibull model to interval (and right) censored survival 
data
using survreg in Splus 5.1.  I'm trying to use the "strata" argument in the
model formula as described in Venables and Ripley, 3rd edition, pages 376.
However, including strata(plot) in my model formula does not seem to result
in the fitting of different scale parameters for each level of "plot" and
does not seem to be any different from just including "plot" in the formula
without the strata argument.  Thanks for any help with understanding this.

         Two more related questions.  Are there any Splus functions for fitting 
a
Cox proportional hazard model to interval censored data or for estimating
survival curves for interval censored data?

Thanks,
Brian


options(contrasts=c("contr.treatment", "contr.poly"))
tmp<-Surv(Surv.dat.factor$I1, Surv.dat.factor$I2, type="interval2")

###### fit with strata argument
fit1a<-survreg(tmp~Species+Trt+Gap+strata(Plot), data = Surv.dat.factor,
link = c("log"),
         dist = c("extreme"))
> fit1a
Call:
survreg(formula = tmp
~ Species + Trt + Gap + strata(Plot), data = 
        Surv.dat.factor, link
= c("log"), dist = c("extreme"))

Coefficients:
 (Intercept) Specieslitu
Speciesquru      TrtR   TrtT+R     TrtT      TrtF 
    6.079439    -1.61673
   1.156143 0.3354253 1.013128 1.040709 0.7447958

        Gap
strata(Plot)Plot=gap12 strata(Plot)Plot=gap13 
 -0.6008336
0.5120529              0.4129829

 strata(Plot)Plot=gap11
strata(Plot)Plot=gap1 strata(Plot)Plot=gap3 
               0.376538
     0.4910114             0.2350804

 strata(Plot)Plot=gap5
strata(Plot)Plot=gap6 strata(Plot)Plot=gap15 
             -0.156611
     0.2999407              0.7962731

 strata(Plot)Plot=gap7
strata(Plot)Plot=gap8 strata(Plot)Plot=gap10 
             0.4850185
      0.788601             -0.9218784

Dispersion (scale) =
1.477747
Degrees of Freedom: 1080 Total; 1060 Residual
-2*Log-Likelihood:
3103.197 


###### fit without strata
argument
fit1b<-survreg(tmp~Species+Trt+Gap+Plot, data = Surv.dat.factor,
link = c("log"),
         dist = c("extreme"))
> fit1b

Call:
survreg(formula = tmp ~ Species + Trt + Gap + Plot, data =
Surv.dat.factor,
        link = c("log"), dist =
c("extreme"))

Coefficients:
 (Intercept) Specieslitu Speciesquru      TrtR
  TrtT+R     TrtT      TrtF 
    6.079439    -1.61673    1.156143 0.3354253
1.013128 1.040709 0.7447958

        Gap Plotgap12 Plotgap13 Plotgap11
Plotgap1  Plotgap3  Plotgap5 
 -0.6008336 0.5120529 0.4129829  0.376538
0.4910114 0.2350804 -0.156611

  Plotgap6 Plotgap15  Plotgap7 Plotgap8
Plotgap10 
 0.2999407 0.7962731 0.4850185 0.788601 -0.9218784

Dispersion
(scale) = 1.477747
Degrees of Freedom: 1080 Total; 1060
Residual
-2*Log-Likelihood: 3103.197 

-----------------------------------------------------------------------
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

<Prev in Thread] Current Thread [Next in Thread>