ABSTRACT
This model was introduced by Rao (1959) who also provided the statis tical analysis. An alternative way of writing the above model is to write it as
where = (1, . . . , 1) is a row vector of ones of length N {. An advantage of writing it in this manner is that if we have another observation matrix on a second treatment group, say X2 = (xi , . . . , Xyv2), which has its expectation given by
where \'Nl = (1 , . . . , 1) is again a row vector of ones but of length N2, then we can write the two models together as
and a comparison of ξ{ with ξ2 can be carried out, i.e., a comparison of two treatment groups. In the most general form, a growth curve model can be written as
where X is a p x N matrix of observations, B : p x q and A : m x N are known matrices, and ξ : q x m is a matrix of unknown parameters. It is assumed that the N columns of E are iid Np(0, Σ).
