Identifying ‘ideal’ companion dogs: development of the canine amicability assessment (CAA)

Dogs in the past were specifically bred to undertake a range of utilitarian roles in which they assisted humans. As a result, an array of dog breeds came into existence. Nowadays, the majority of dogs who have originated from these breeds are kept primarily as pets, whereby they provide companionship to people. Very few individuals continue to be used in their original capacity. In addition, an emphasis on selecting dogs based solely on appearance tends to be prevalent in the modern era. Yet it is likely that some dogs may not be particularly well suited to the role of human companion. Consistent with this possibility, a large number of dogs are relinquished to shelters as a result of exhibiting behaviours people consider problematic, and there is increased public concern regarding dangerous dogs in the community. Currently there are no specific standards regarding dog breeding which take into account the behaviour of the dog relating to its ability to be a successful companion. Nor are there established methods by which these behaviours can be evaluated accurately. While many protocols aimed at measuring dog behaviour are available, few have been developed correctly and most may not be valid or reliable. To improve the welfare of pet dogs and protect the community it is necessary to establish what constitutes an ideal pet dog and then to find ways to assess dogs relative to these requirements. On this basis the aim in this thesis was twofold; first, to determine what characteristics are considered ‘ideal’ in a companion dog and, second, to scientifically develop and then begin to evaluate an objective behavioural assessment which measures desirable dog behaviour. The development of a survey and subsequent collection of data from the Australian public established that most people prefer a dog which exhibits behaviours related to the canine personality dimension called Amicability, which is comprised of the traits friendly, easy going, relaxed, non-aggressive and sociable. To assess these traits the Canine Amicability Assessment (CAA) was developed, consisting of a brief test in an unfamiliar and carefully controlled setting, and during which a dog’s behaviour is evaluated in response to interacting with a stranger, both on and off lead, in the presence and absence of his or her owner. An initial pilot study, followed by the main study in which data were collected from over 200 dogs and their owners, was undertaken. The psychometric properties of the assessment were evaluated using multiple measures of behaviour, including ratings by experts, coding of specific behaviours, and questionnaire data provided by dog owners. Results indicated that the CAA was able to elicit behaviours which could be reliably rated by dog experts and that the canine personality dimension of Amicability could therefore be quantified. Furthermore, the intra- and inter-rater reliability scores obtained were high. A benchmark measure of Amicability was developed on the basis of the ratings provided by experts and this was then used to compare against coded behavioural variables and questionnaire data provided by dog owners. Analyses revealed a number of behavioural variables that were statistically related to the personality construct, although the strength of the correlations obtained were generally relatively weak. These included behaviours relating to interactions with the stranger, body posture, locomotion, vocalisations and tail wagging. Test-retest reliability was adequate for adult dogs, whereas predictive validity, examined by testing dogs as puppies and then again as adults, did not yield promising results. In addition, there were significant discrepancies between expert ratings and information provided by dog owners. It was suggested that owners may be poorer judges of their dog’s personality than were experts, partly because of a generally positive bias towards their own dog, as well as not knowing how their dog behaves in their absence. As a consequence limitations may exist in assessments of amicability that rely on owner surveys. On this basis it was concluded that the most effective approach to gathering accurate information about a dog’s amicability is likely to require using multiple measures. The CAA, a scientifically developed and relatively objective measure of canine amicability, was shown to be a worthwhile tool with potential to provide an appropriate measure of the personality dimension most valued by companion dog owners. Despite some limitations, therefore, and a need for further research and development, this study makes an important contribution to the scientific literature in its field. With further development the CAA will potentially provide a protocol that is accurate, reliable and feasible and which can be utilised in many contexts to gain a better understanding of individual dog behaviour and to quantify a dog’s level of amicability relative to other dogs.