[R] Categorical variables and Plotting a Cox model with interaction terms

[hossiet]

I'm a new to the R world and I have been following the John Fox Appendix on
Cox regression
(http://cran.r-project.org/doc/contrib/Fox-Companion/appendix-cox-regression.pdf)
to run some analyses. I have been able to generate models and some simple
figures e.g.:

>library(splines) #required for survival package
>library(survival) #req for Coxph 
>Field1<-read.csv("c:/R/Field1.csv")
>attach(Field1)
>mod.Field1 <- coxph(Surv(Survival, Censored) ~ Colour * Size + Site,
data=Field1)
>mod.Field1

Colour and Size are binary (dummy) variables (i.e., 0 and 1) and Site is a
location (i.e., 1, 2, 3). 

Question 1: my site variable is technically a categorical (i.e., not
ranked), random factor but how do I (or can I?)define this?


Fox shows this method for plotting out seperate curves for a single factor:

> mod.allison <- coxph(Surv(week, arrest) ~ fin + age + race + wexp + mar +
> paro + prio,
+ data=Rossi)
> attach(Rossi)
> Rossi.fin <- data.frame(fin=c(0,1), age=rep(mean(age),2),
> race=rep(mean(race),2),
+ wexp=rep(mean(wexp),2), mar=rep(mean(mar),2), paro=rep(mean(paro),2),
+ prio=rep(mean(prio),2))
> detach()
> plot(survfit(mod.allison, newdata=Rossi.fin), conf.int=T,
+ lty=c(1,2), ylim=c(.6, 1))
> legend(locator(1), legend=c(’fin = 0’, ’fin = 1’), lty=c(1,2))

In this example the terms in the model were all main effects. 

Question 2: How would you define the variable Rossi.fin if say there was a
paro*prio interaction term?

Basically, I want to generate a survival plot with 2 lines (e.g. 0=small,
1=big) using my model, but I can't figure out how to define the code becasue
my model has interaction terms. Any help with this would be greatly
apreciated.

Sincerely,

hossiet


-- 
View this message in context: http://r.789695.n4.nabble.com/Categorical-variables-and-Plotting-a-Cox-model-with-interaction-terms-tp2967483p2967483.html
Sent from the R help mailing list archive at Nabble.com.