THE ANALYSIS OF MULTISTRATUM AND SPATIALLY CORRELATED REPEATED MEASURES DATA

In repeated measures experiments on animals, the animals are often housed in pens and as well as the correlation between measurements over time, the induced correlation due to the pen effect must be taken into account. Current research in the analysis of field experiments is focused on the use of spatial methods. If in addition we have repeated measurements over time, we need to take into account both the temporal and spatial dependence. An approach is presented for the analysis of repeated measures data when an additional level of dependence exists. We consider both the complete and incomplete data situations and discuss estimation based on residual maximum likelihood. The procedures are illustrated using data from two experiments.