Using multivariate statistical analysis to measure ovine temperament; stability of factor construction over time and between groups of animals

The ovine arena test in conjunction with multivariate statistical analysis (factor analysis) may be a means of measuring ovine temperament for practical purposes. Stability of factor construction over time and between groups of animals is considered to demonstrate trait consistency and is, therefore, one of the first steps in validating a temperament/personality test from this perspective. The aim of this study, therefore, was to assess the stability of factor construction, as a measure of trait consistency, using arena test data from three groups of animals with one group (Group 1) tested repeatedly over three rounds (twice at 8 months and once at 22 months of age). Group 1 consisted of 193 mule (Bluefaced Leicester Sire x Scottish Blackface/Welsh Speckled Face dam), ewe lambs (8 months old). Groups 2 and 3 consisted of 189 and 185 mules, respectively (14 months old). All animals were tested for 6 min in a 13 m x 3 m arena. Factor analysis (varimax rotation) was performed twice on the behavioural data (latency to bleat, total number of vocalisations, distance travelled, time spent in different areas of the arena and number of times crossing in and out of pertinent areas), initially using all data recorded on a per minute basis ('Per Minute') for all 6 min of the test (10 factors extracted) and then using total values ('Total'), the summation of the 6 min for each behaviour measured (4 factors extracted). Stability of factor loadings between rounds and between groups was tested using Kendall's coefficient of concordance. For the 'Per Minute' data, 5 out of the 10 factors showed significant (p < 0.05) concordance between rounds whilst 9 out of 10 factors showed significant (p < 0.05) concordance between groups. All four factors generated from the 'Total' data demonstrated significant (p < 0.05) concordance between rounds and between groups. The four factors generated from the 'Total' data were considered to be of potential merit for future studies. These factors were named-'conspecific motivation-fear', 'conspecific motivationdistress', 'activity' and 'low conspecific motivation'. (c) 2006 Elsevier B.V. All rights reserved.