random effects estimation in Lmer

On 2009-12-15 09:23, David Duffy wrote:

Have you looked at the kinship package at all?  It uses lme, but analyses 
age at onset data from families more nicely than lmer can (see the coxme() 
function and the vignette/documentation).

Actually, the latest version of the coxme() function now lives in Terry 
Therneau's new 'coxme' package.

It no longer depends on 'nlme', but uses its own fitting routines and sparse 
matrix representation, and has a lme4-like syntax for the specification of 
random effects.