RESUMEN
We present an approach based on generalized linear models for analysing dominance data. First, dominance is defined as a parameter characterizing the relationship between two individuals, determining the expected number of successes of the first individual in disputes with the other. Second, models known from the literature of two different forms of transitivity are defined in terms of these parameters, and examples of different tests of these models are given. Third, a new model is developed where the traits of individuals involved in the dominance interactions are included as covariates. Finally, we show how two forms of intransitive models of dominance structures can be constructed by including a certain interaction term between the trait variables, and terms taking into account the effect of relatedness between the individuals in the group. We reanalyse several data sets in the literature, and also discuss Appleby's method (1983, Anim. Behav., 31, 600-608), which is frequently used in analysis of dominance data. Copyright 1998 The Association for the Study of Animal Behaviour. Copyright 1998 The Association for the Study of Animal Behaviour.