Estimating behavioral transition rates: Problems and solutions

Continuous time Markov chain (CTMC) models offer ethologists a powerful too l. The methods are based on well-established procedures for estimating the rates at which one state (e.g. resting) changes to some other set of states (e.g. feeding, fighting, etc.). Unfortunately, ethological data typically differ in a very critical manner from the type of data to which these metho ds are usually applied: ethological data are usually heavily censored in th e sense that each behavioral state shows frequent transitions to several ot her possible states. This occurs when several competing processes can each end a bout. We used computer simulation of various behavioral models with known transit ion rates to investigate the unknown performance of four of the most popula r statistical tests for screening data prior to application of CTMC models; this included a modification of one of these tests derived under the assum ption of random censoring. Two of the four tests failed completely and woul d result in rejection of nearly all data even if the model did fit the assu mptions of the CTMC methods. Only Barlow's total-time-on test performed wit h an acceptable a error rate under all conditions. None of the tests were p articularly effective at detecting certain types of departures from the CTM C assumptions. Guidelines are given as to how much confidence should be attached to appare nt changes in transition rates.