[R] offset in coxph

Prof Brian Ripley ripley at stats.ox.ac.uk
Thu Jul 26 23:34:38 CEST 2007

Removed?  That it was ever there is not my recollection and seems very 
unlikely given that survival is ported from S where glm() does not have 

As far as I know it has only ever been in glm() and lm() in R: the way 
which is described in the White Book is to use the offset() function, and 
this is preferred (it works correctly for prediction, for example).  The 
function form is supported for coxph, and used in the test suite.

Please forget you ever knew 'offset' could be an argument, and use 
offset() in your formulae instead.

On Thu, 26 Jul 2007, Michael Gormley wrote:

> The offset argument used in glm and other functions seems to have been
> removed from the argument list for coxph.  I am wondering if there is a
> reason for this and if there is a possible work-around in order to produce a
> cox-ph object without fitting coefficients?

Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

More information about the R-help mailing list