Hi all,
I am using R package 'ordinal' to fit a cumulative logit ordinal model with
random effects. Does someone know
1) The optimization method used in estimating the parameters from the
marginal likelihood?
2) In Adaptive Gauss-Hermite Quadrautre, let f() be the function to be
intergrated. The second derivative of function f() with respect to the
random effect needs to be calculated in order to center and scale the
quadrature points. Is the second derivatie of f() derived to an exact form
or approximation form, if the approximation form, what is the method used
to approximate.
Thanks,
-Jiayi
__
Jiayi Hou
Department of Biostatistics
School of Medicine
Virginia Commonwealth University
[[alternative HTML version deleted]]