Symmetrically dependent models arising in visual assessment data

Given data hom bilateral visual assessments on N subjects at k occasions, w e consider inference for contralateral correlations (C) between fellow eyes and lateral correlations (L) among p different assessments of the same eye . Under permutation symmetric dependence structure between observations fro m fellow eyes and among observations from the same eye, we obtain maximum l ikelihood estimates of L, C, and L - C. Based on the large-sample estimates of the corresponding covariance structures, we test the hypothesis that th e association between fellow eyes is constant across time and the hypothesi s that lateral and contralateral associations between any two occasions are the same.