Relationships between demographic characteristics of shelter dogs and performance on tests of a behavioral evaluation and between performance and adoption success

Abstract

Dog behavior has been studied using diverse methods, including owner-completed questionnaires, records from veterinary clinics that evaluate behavioral problems, and test batteries such as those used by some animal shelters. Measures of prevalence for multiple behavioral issues and analyses of whether dog demographic characteristics are associated with problematic behaviors are commonly reported in studies using questionnaires and clinical records; however, these measures are infrequently reported for all tests and subtests in studies using data from shelter behavioral evaluations. In this study, we determined prevalence of both concerning and dangerous behaviors on 12 tests and subtests of a canine behavioral evaluation by analyzing data collected by the behavioral staff at a NY shelter (n = 1,104 dogs tested between 2014 and 2019). We also examined whether behavior on four tests and subtests with prevalence over 10% (two Sociability subtests, Handling test, and Dog-to-dog meeting subtest) was associated with likelihood of being returned to the shelter post-adoption, and whether dog demographic characteristics (age, sex, reproductive status, and body size) were associated with behaviors on tests and subtests. Of all dogs tested, prevalence measures of dangerous behaviors were always less than 5%. Our measures of prevalence (concerning and dangerous behaviors combined) for food guarding (15.5%), stranger-directed aggression (6.5%), and dog-directed aggression (16.7%) were generally similar to those reported for dogs in other shelters and in homes. For the four tests/subtests examined here, displaying concerning or dangerous behavior was not significantly associated with return to the shelter, and positive predictive values (PPV) were low, which implies that many positive tests will be false positives. We also found few significant associations between demographic characteristics and behavior on tests and subtests and PPV were low. Other factors, such as previous experiences with people and dogs, as well as underlying physiological and neurobiological processes, may be more important influences on a dog’s behavior during a shelter behavioral evaluation than general demographic characteristics. Total number of tests and subtests out of a possible 12 on which dogs were assessed as showing concerning or dangerous behaviors ranged from zero to 7 per dog (median, 0; mean, 0.82). In summary, poor predictive abilities characterized associations between performance on certain tests/subtests of the behavioral evaluation and return to the shelter and between demographic characteristics and behavior on test/subtests. Our data may contribute to the ongoing debate about the usefulness of canine behavioral evaluations in shelters.