Detection of aggregative behaviour in binary choice experiments

Aggregative behaviour can be defined as a natural tendency for individuals to cluster in space into groups of varying sizes. In entomology, this behaviour can be detected in binary choice experiments. Analysis of these experiments reveals specific characteristics (unequal counts between repetitions, dependence of individual choices) that make the traditional adjustment tests fail. We demonstrate that the use of generalized linear models can circumvent these pitfalls and deliver a reliable diagnosis on the social behaviour of the studied invertebrates. The strength of this behaviour can then be evaluated through the use of common correlation models.