Reference analysis was first introduced by
Bernardo (1979) and further developed by Berger and Bernardo (1992a).
Berger et al.(2001) proposed a new method to obtain exact reference
priors, which is proved to be one of the most successful methods to
derive noninformative prior distributions. In this paper, the algorithm
proposed in Berger et al.(2001) is used to obtain reference priors
for the growth curve model with general covariance structures.
Simultaneously, some applications of the corresponding results are
presented.