How reliable are standardised behaviour tests and are they valid in predicting the suitability for use in police horses?

Standardised behaviour tests have the potential to improve equine welfare by enabling the selection of horses suitable for a particular role. A range of tests are found in scientific literature which make such claims. However, with notable exceptions, authors have failed to establish the reliability of their tests or the validity of their claims. This study aimed to determine a set of reliable behaviour tests based on those reported in literature; and to test the validity of these as predictors of horses’ suitability as police horses. Reliability of response to a novel object, social isolation and handling tests was established by tests repeated at three week intervals on 33 horses. Behavioural responses were recorded post hoc to enable the full range of responses to be measured. There was some evidence that the horses habituated to static stimuli used in the tests, and the majority of the tests were at best moderately reliable, with only reactivity tests showing good or better reliability. Tests with moderately reliable results or results which could be explained by habituation, were included in the validity testing. This compared the horses’ responses to the behavioural tests with handlers’ ratings of key aspects of police horse performance. These subjective scales of suitability of police horses were developed from interviews with instructors and yard managers. The items were combined to develop a set of specific indices of police horse roles. Spearman’s rank order correlations were used to explore whether any relationships existed between the horses’ behavioural responses to the tests and subjective ratings of their suitability for their use. The results revealed eight moderate associations which were on the whole between horses’ responses to the reactivity tests (unexpected noise and sudden moving object) and the suitability scores. These findings suggest that standardised behavioural tests are not strongly reliable. The validity analysis indicated that either the tests used in this study did not serve as predictors of suitability for use, or that behavioural responses to standardised tests cannot be generalised to the wider challenges faced by these horses in their roles. However the reactivity tests did reveal stronger reliability and they were the only measures to show any association with the suitability scores, suggesting that reactivity tests may have some validity in measuring a horse’s suitability. Overall, the findings of this study cannot support claims made for these (or similar) tests to be valid predictors of a horse’s suitability for use