The hierarchical structure of dog personality in a new behavioural assessment: A validation approach

Abstract

Experimental assessments can be useful in the study of individual differences among dogs. One example of such assessment is the Dog Mentality Assessment (DMA), in which stable traits, referred to as personality traits, have been detected. Due to limited access to the DMA for dogs of non-working breeds, a new experimental assessment named Behaviour and Personality Assessment in Dogs (BPH) was developed in 2012 with the DMA as a model. In this study, behavioural ratings from 12,117 dogs assessed with the BPH were analysed in two steps: first, a hierarchical factor analysis procedure was carried out, and second, the construct validity of the extracted factors was studied. Two measures of validity were used: correlations with subjective ratings during the assessment (internal construct validity (ICV)) and correlations with data from a web-based questionnaire regarding everyday behaviour (external construct validity (ECV)). The ECV was also used to investigate on which level of the factor hierarchy everyday behaviour was best predicted. The approach revealed a hierarchy of factors, from one general factor at the top to 28 specific factors at the bottom, with generally high ICV. The first factor, Boldness, is related to six of the eight subtests and is associated with a positive attitude towards unfamiliar persons, interest for object play, low fear, and high degree of exploration. Most of the specific factors stem from the factors Sociability, Playfulness and Non-social fearfulness at the third level in the hierarchy, factors with high or at least moderate ECV. Sociability seems to be the best predictor for attitude towards unknown persons and dogs outside the assessment situation, including positive interest, fear, and aggression. The broader factors at the first levels correspond well to a range of everyday behaviours but for some behavioural tendencies more specific factors appearing at lower levels in the hierarchy were of greater importance. For example, noise-related fear was predicted first by a factor from the 12th factor level. The results from the ECV analysis indicate consistency between contexts and suggest that the BPH can reveal dog personality traits. The information from the assessment may give indications regarding welfare as well as potential problem-causing and preferred behaviour. Given a genetic basis for the traits, the most promising application is in dog breeding, where a combination of broad and narrow factors, relevant for the breed in question, may be used as measures in breeding objectives.