Reliability and validity of behaviour tests in Hovawart dogs

The reliability of six judges who assessed the behaviour of dogs simultaneously in the behaviour tests (BTs) of the Swiss Hovawart Club (SHC) was examined on three dates in spring, summer and autumn 2006. The reliabilities were calculated using the Spearman-Brown prediction formula and Spearman rank correlations for quantitative traits and the kappa coefficient for binomial variables. Secondly, the validity of the BT was studied in a retrospective study by comparing the results of BT of dogs which later showed behaviour problems related to aggressiveness with the results of behaviourally inconspicuous dogs. Problematic dogs were reported to the club because they had to be euthanised or re-homed with different owners because of their behaviour problems. The behaviour problems were quantified by using a questionnaire for the owners who had reported the problems. Generally, the reliability between the six judges was high, but there were differences of agreement on the three different dates. The results of the retrospective study indicated that the behaviourally inconspicuous dogs had generally better scores in the BT than problematic dogs. When dogs with a problem of aggressiveness and their matched control dogs were compared, there were significant differences in the trait "playing" in the questionnaire and the BT. Dogs who later displayed aggressive behaviour played less enthusiastically than dogs who remained non-aggressive. However the other traits which were assessed in the BT were not useful to predict which dogs would develop problems of aggressiveness later. (C) 2008 Elsevier B.V. All rights reserved.