Darwin's Dogs loves to ask people questions about their dogs. We do it for science! The data are useful to try to find correlations between the dogs' behavior and genetic differences. But how reliable are questionnaires filled out by owners on their dog's behavior? Are the answers we get really describing the dog's personality?
Rayment et al. undertook to answer this question in their recent paper, "Investigating canine personality structure using owner questionnaires measuring pet dog behaviour and personality." They recruited over five hundred dog owners on the internet to fill out three different widely-used dog personality questionnaires. Then they asked: do the results of the different questionnaires agree with each other when applied to the same dog? And do the answers given by owners who have more dog experience seem to be different from the answers given by owners with less dog experience?
The three questionnaires that the team looked at were the Canine Behavioral Assessment and Research Questionnaire or C-BARQ, Dog Impulsivity Assessment Scale or DIAS, and Monash Canine Personality Questionnaire, Revised or MCPQ-R. These three questionnaires all use a similar approach. They ask the owner a lot of questions about their dog's behavior, but to create a final description of the dog's personality, they group the responses to related questions together into factors. As a hypothetical example, a dog who barks frequently at the mailman and who has growled when approached by a visitor in the home might score high on a factor called "stranger-related aggression."
The three questionnaires have different factors. C-BARQ has 14, DIAS has 3, and MCPQ-R has 5. (The C-BARQ factors are more numerous because they're more specific.) But if a dog really does have a consistent personality, and if these questionnaires are successfully measuring it, then some of the factors from different questionnaires ought to overlap - for example, a dog who scores high in "Response to Novelty/Aggression" in DIAS would also be expected to score high in "Stranger-Directed Aggression" in C-BARQ. If there were no relationships between factors from different questionnaires, then we'd have to question whether the questionnaires were actually measuring anything real.
In fact, Rayment et al. did find many of the correlations you might expect. For example, the C-BARQ factors "Stranger-Directed Fear," "Dog-Directed Fear," and "Non-Social Fear" all correlated with the MCPQ-R factor "Neuroticism." This does suggest that the questionnaires are measuring something real (a relief to a group like Darwin's Dogs that makes so much use of this kind of tool!).
The authors noticed that a large percentage of their respondents had a lot of experience with dogs. This may not be surprising, given that they recruited their participants by social media in dog-related groups. Given so many dog experts, they had a good sample size for asking the question "are dog experts giving different answers on questionnaires than non-experts?" They looked again at the correlations that they had found between factors from different questionnaires. This time they compared how strong the correlations were considering only questionnaires taken by dog experts, versus questionnaires taken by non-experts. They found a difference - the factors CBARQ "Dog Rivalry," CBARQ "Owner-Directed Aggression," DIAS "Behavioural Regulation" and DIAS "Response To Novelty" were related to other factors less strongly in the non-expert group than in the expert group.
Does this mean that answers from non-expert owners are less trustworthy than answers from expert owners? The authors aren't sure, but they do suggest that anyone using dog personality questionnaires in the future might consider collecting owner information about their level of dog experience.
Questionnaires are an imperfect way of gathering information about dog behavior, of course, but they're the best option we have at Darwin's Dogs, and we're glad they're mostly measuring what they're supposed to be measuring!
Diana J. Rayment et al. "Investigating canine personality structure using owner questionnaires measuring pet dog behaviour and > personality." Applied Animal Behaviour Science, Volume 180, July 2016, Pages 100–106 http://dx.doi.org/10.1016/j.applanim.2016.04.002