Title: merDeriv: Derivative Computations for Generalized Linear Mixed Effects
Models with Application

Author: Ting Wang

Affiliation: University of Missouri

Abstract:
While likelihood-based derivatives and related facilities are available in R
for many types of statistical models, the facilities have been notably lacking
for models estimated via lme4. This is because the necessary statistical
output, including the Hessian, Fisher information and casewise contributions
to the model gradient, is not immediately available from lme4 and is not
trivial to obtain. In this presentation, we describe merDeriv, an R package
which supplies new functions to obtain analytic output from Gaussian mixed
models as well as quadrature method for generalized linear mixed model with
one cluster. After describing the computation methods, we illustrate a variety
of possible uses of these derivatives that will be familiar to many Psychoco
attendees. These uses include score tests of fixed effect parameters,
heterogeneity in random effects and likelihood ratio tests of non-nested
models.