Test-retest reliability and predictive validity of a juvenile guide dog behavior test

Abstract

Abstract The ability to measure stable and consistent behavioral traits in dogs would facilitate selection and assessment of working dogs, such as guide dogs. Ideally, these measures should predict suitability for the working role from a young age. This study assessed test-retest reliability of a juvenile guide dog behavior test and predictive validity using qualification or withdrawal from guide dog training. Ninety-three guide dog puppies (52 female; 41 male) were tested at 5 (mean 4.78; ±0.73 SD) and 8 (mean 7.98; ±0.78 SD) months of age. The dogs were exposed to a sequence of 11 stimuli designed to assess the dogs' reactions to meeting a stranger, obedience commands, body sensitivity, scavenging, and animal and human distractions. The behavior of dogs was digitally recorded and analyzed using an ethogram incorporating both frequency of behavior and specific reactions to stimuli. Test-retest reliability indicated inter-individual consistency in many of the behavioral measures such as jumping, barking, and low greeting posture, as defined in our ethogram. Behavior measures that did not show interindividual consistency between tests included obedience responses, lip licking, body shaking, and scratching. Binary logistic regression models revealed 7 behavioral measures at 5 months and 5 measures at 8 months that were significantly associated with qualification or withdrawal. Uncorrelated measures and principal component scores of correlated measures were combined in a logistic regression model that showed great potential for predicting the probability of a dog qualifying or being withdrawn from guide dog training.