[R] loglogistic survreg

[Thomas Lumley]

On Tue, Sep 18, 2012 at 7:06 AM, Jeremy Brown <jebrown_141 at yahoo.com> wrote:
> Hi,
>
> I'm trying to graph the hazard function using the survreg function with the distributional assumptions to be loglogistic. If h(t)=[lambda*alpha*(lambda*t)^(alpha-1)]/[1+(lambda*t)^alpha], what output from R is alpha?

  survreg() fits location-scale models, so the model being fitted is
that log(survival time) has a logistic distribution with location
parameter given by the regression model and scale parameter estimated
in each stratum.  I think that means exp(fitted values) is alpha in
your parametrisation.  In a model with only an intercept, that would
be exp(intercept).

   -thomas

-- 
Thomas Lumley
Professor of Biostatistics
University of Auckland