[R] coxme frailty model standard errors?

Corey Sparks corey.sparks at utsa.edu
Wed Dec 5 23:22:04 CET 2007

I am running R 2.6.1 on windows xp
I am trying to fit a cox proportional hazard model with a shared
Gaussian frailty term using coxme My model is specified as:

nofit1<-coxme(Surv(Age,cen1new)~ Sex+bo2+bo3,random=~1|isl,data=mydat)

With x1-x3 being dummy variables, and isl being the community level
variable with 4 levels.

Does anyone know if there is a way to get the standard error for the
random effect, like in nofit1$var?  I would like to know if my random
effect is worth writing home about.

Any help would be most appreciated
Corey Sparks

I can get the following output
nofit1<-coxme(Surv(Age,cen1new)~ Sex+bo2+bo3,random=~1|isl, data=no1901)
nofit1 Cox mixed-effects model fit by maximum likelihood
  Data: no1901
  n=959 (2313 observations deleted due to missingness)
  Iterations= 3 69 
                    NULL Integrated Penalized Log-likelihood -600.0795
-581.1718 -577.9682

  Penalized loglik: chisq= 44.22 on 5.61 degrees of freedom, p= 4.3e-08
Integrated loglik: chisq= 37.82 on 4 degrees of freedom, p= 1.2e-07 

Fixed effects: Surv(Age, cen1new) ~ Sex + bo2 + bo3 
         coef exp(coef)  se(coef)    z      p
Sex 0.2269214  1.254731 0.2151837 1.05 0.2900
bo2 0.5046991  1.656487 0.2510523 2.01 0.0440
bo3 1.0606144  2.888145 0.2726000 3.89 0.0001

Random effects: ~1 | isl 
Variance: 0.3876189

Corey Sparks
Assistant Professor
Department of Demography and Organization Studies
University of Texas-San Antonio
One UTSA Circle
San Antonio TX 78249
Phone: 210 458 6858
corey.sparks at utsa.edu

More information about the R-help mailing list