Coxph

[Tristan Lorino <lorino@vet-alfort.fr>]

Hi,

I do simulations to obtain a grouped survey data set ; I have only two
covariable : GROUPE for the group design, and COV which is the covariable
of interest. Beta (regression parameter for the Cox model, i.e.
coefficient of COV) is set to 0.5.
Then I fit a Cox model. I would like to only test : beta=0.5
How can I do this ? By default, the test is always turned with the
following null hypothesis: frailty = beta = 0.
Do I use a offset ? But

f2 <- coxph(Surv(temps, cens) ~
offset(0.5 * cov) + frailty(groupe), data = "">
seems to be the wrong solution...

Thanks,
Tristan Lorino

------------------------------------------------------------------
Ecole Nationale Vétérinaire d'Alfort : http://www.vet-alfort.fr
Institut National Agronomique Paris-Grignon : http://www.inapg.inra.fr
Probabilités, Statistique, Littérature, etc. : http://daedale.free.fr