How Many Dimensions are Needed to Describe Temperament in Animals: A Factor Reanalysis of Two Data Sets
Abstract
Factor analysis has commonly been used to infer the dimensions of animal temperament. However, the results were often complicated by large number of broad and situation-specific factors caused by low psychometric adequacy of the correlation matrices,undermining the assumptions of factor analysis. In this study I reanalyzed the datasets obtained by Royce, Foley & Yeudall (1973) and Gervai & Csanyi (1985) including,however, only the variables with high correlations (multiple R^ > 0.3) and psychometric adequacies (Kaiser-Meyer-Olkinmeasure > 0.5). This yielded more stable andsimpler factor solutions than in the original studies. Specifically, even though the present reanalysiscannot rule out the existence of other temperament factors, it indicates thattwo general dimensions, Activity-Exploration and Fear-Avoidance, are present in suchdiverse species as mice and the paradise fish.