Hybrid pairwise likelihood analysis of animal behavior experiments

The study of the determinants of contests between animals is an important issue in understanding animal behavior. Tournament experiments among a set of animals are used by zoologists for this purpose. From a statistical point of view, the results of these tournament experiments are naturally analyzed by paired comparison models such as the Bradley-Terry and the Thurstone models. A major complication is the presence of dependence between the outcomes of couples of contests with an animal in common. Likelihood analysis of this type of animal behavior experiments in presence of interdependence between contests is computationally demanding. An alternative fitting method that mixes optimal estimation equations and pairwise likelihood inference is then suggested. The performance of the proposed methodology is investigated by simulation studies and then applied to a real data set about adult male Cape Dwarf Chameleons