Analysing qualitative behavioural data: A Markov chain aid

Abstract

Redundancy of information is liable to be present in successive observations of an animal's behaviour. This redundancy can be tested for, and its effects studied, with a statistical procedure, Markov chain analysis. Efficient sampling frequency and the precision of estimators are both found to be affected by the redundancy present in one typical data set based on video-taped records of a nursing sow. A method of selecting sampling frequencies for observations is described.