[R] censorReg/survreg

[Terry Therneau]

-- begin inclusion --
In addition, is there anyone knowing why censorReg does not exist in R
and what's the difference between censorReg and survreg (besides the
truncation)?
---end inclusion ---

 CensorReg and survreg were both developed in Splus, they fit the same
models, both use maximum likelihood, and they give the same answers.
The big difference is what is done after the fit.  CensorReg comes out
of the industrial reliability perspective, survreg from the
biostatistics one; there are different sets of post-fit plots and
summaries that are usual in the two areas.  One of the programs in my
test suite reprises many of the censorReg plots (a nice test of
residuals and predicted values).
  The author of censorReg, Bill Meeker, is in emeritus status at Iowa
State, and I expect that his energy/interest level in retooling the
package for R is limited.  Life moves on.
   Allowing for left truncation would not be that hard for survreg, and
has been on my "to do" list for 10 years or so.  Given it's priority I
may never get to it.  If someone else (you?) would like to tackle it
drop me a line, I would be delighted to expand the list of contributers.

Terry Therneau