Date of Award
Doctor of Philosophy
Several multidimensional quantitative approaches have been applied to the analysis of the organization of behavior. These include chi square, lag sequence analysis, and non-symmetric multidimensional scaling techniques. Transition matrices are generally asymmetric and this is a problematic feature for the first two techniques noted above. Hence is is argued that non-symmetric multidimensional scaling (MDS) techniques are the most appropriate for the analysis of such data sets.;To illustrate these three techniques and to evaluate non-symmetric MDS as a valuable new tool which has been underutilized by ethologists, observational data were collected in a focal animal study of two members of a captive family of meerkats (Suricata curicatta). The results of chi square analyses showed that meerkat behavior could be divided into three major groups--Solitary Behavior, Active Interactions and Passive Interactions. The patterns of relationship between these major groups of activities, were ascertained using a non-symmetric MDS program called DEDICOM. Solitary activities could be grouped into subclasses labeled foraging, self maintenance and reconnaissance. Clear patterns of transition between these smaller units were identified using MDS DEDICOM. Also solitary activities are more likely to be followed by Passive Interactions than Active ones. Active Interactions are highly likely to change into Passive Interactions before they are terminated. Lag sequence analyses identified clusters of activities similar to those found using the other two analytical approaches.;Of the three techniques applied, lag sequence analysis was the most difficult to apply and interpret. The chi square analyses produced informative results but still required a degree of subjective analysis. MDS DEDICOM provided clear information both about the clustering of activities into related classes and the patterns of transition between these subgroups. It was concluded that non-symmetric MDS techniques offer the best alternative for the analysis of behavioural organization when the data set to be analyzed is an asymmetric matrix. The simplicity and appropriateness of this technique makes it an extremely valuable tool for the student of behavioural organization.
Sorensen, Linda M., "Analyzing The Organization Of Animal Behavior: The Application Of Nonsymmetric Multidimensional Scaling Techniques" (1989). Digitized Theses. 1800.