[R] Frailty model -- Conditional or Marginal?

[Thomas Lumley]

On Thu, Mar 10, 2011 at 6:33 AM, hongsheng wu <comp611 at gmail.com> wrote:
> Dear all,
>
> When I ran a frailty model like,
>
> model <- coxph(Surv(Day, Indicator) ~ trt + frailty(group,
> distribution="gaussian"))
>
> The variance of effect estimate of trt was estimated by conditional model or
> marginal model?

The frailty model is a conditional model, that is, the hazards for
different treatment groups are assumed proportional conditional on the
frailty.

     -thomas

-- 
Thomas Lumley
Professor of Biostatistics
University of Auckland