Laird-Ware Random Effects Model

Fits a Linear Random Effects Model

Fits the Laird-Ware Linear Random Effects Model. This model assumes that for each subject y=x*b+z*g+e where x and z are known m x p and m x r matricies, b is p x 1 parameter vector and g is a r vector which has a multivariate normal distribution with mean zero, e is a vector of identitcal independent normal mean zero random variables. The purpose is to estimate b, the variance covariance matrix of g and the variance of e.