[R] survival:: plotting survfit with two predictors

Marshall Knoderbane mek243 at cornell.edu
Mon Aug 10 20:48:08 CEST 2009

Hi R-Helpers,

I am having difficulty plotting a coxph model with two predictors.  My 
predictors are "morder" (a factor with five levels where the mean of 
each level is plotted as a separate line) and tmean (continuous).   When 
I  run a model with just morder it is fine and the plot is fine.  When I 
add tmean, the coxph model runs fine but this model will not plot and I 
receive an error message.  Someone posting back in April 2009 received a 
similar error in R2.9.0 (which I am using also) when running 
predict(coxphmodel,newdata=expand.grid()).  This code worked for them in 
R2.8.1 and they thought there might be a bug.  I ran my code in R2.6.2 
and I did not have any improvement.  I suspect it is not a bug but 
rather a problem with my code.  I have not been able to find any other 
solutions in the achieve and I would greatly appreciate any suggestions. 

Thanks in advance,
Marshall  Knoderbane

Below is my code for the two models and the error message:

#produces a fine plot with a separate curve for each level of morder
summary(model2) #is fine
#produces the following error:
Error in model.frame.default(delete.response(Terms), newdata, xlev = 
object$xlevels) :
  variable lengths differ (found for 'a$tmean')
In addition: Warning message:
'newdata' had 5 rows but variable(s) found have 355 rows

More information about the R-help mailing list