Detection of time-inhomogeneity in behavioural processes: tests for multiple abrupt changes in boutlengths

Behavioural records often contain inhomogeinities. The number of such changes is usually not known beforehand. In this paper a description is given of parametric as well as non-parametric methods to determine the number of abrupt changes in which have been published in several papers in the biological and statistical literature have been gathered and completed here. The use of the test procedures is illustrated by means of examples.