Asp, H. E., Fikse, W. F., Nilsson, K., & Strandberg, E. (2015). Breed differences in everyday behaviour of dogs. Applied Animal Behaviour Science, 169, 69-77. doi: http://dx.doi.org/10.1016/j.applanim.2015.04.010
National Canine Research Council Summary and Analysis:
This 2015 Swedish correlational study used owner surveys to describe relationships between dog breeds and everyday behavior. It is included here because, unlike most studies that investigate breed differences, owner breed identification was confirmed by registration numbers and not based on visual inspection or owner beliefs. In addition, it has a large sample size and serves as a comparison and contrast to the [widely-cited study](Duffy, Hsu, & Serpell, 2008) on breed differences in aggression.
The authors also investigated how common behaviors relate to each other. “Everyday behavior” was measured via the Canine Behavior Assessment and Research Questionnaire (C-BARQ), an instrument designed primarily to identify emotions and behaviors that may be considered problematic in companion animals (e.g., aggression, fear, separation related behavior) in addition to attention seeking behaviors, and how well the animal is trained (referred to as “trainability”). This was combined with parts of another questionnaire (see Svartberg, 2005) about playfulness and sociability; in total, there were 133 behavior questions. Owners of 20 breeds from the Swedish Kennel Club (SKC) were recruited largely through online outreach to breed clubs. The breeds were grouped as either working (specifically herding and guarding) or non-working. The difference between breeds designated as “working” and those considered “non-working” in Sweden is not a matter of function: “working” breed dogs must be scored on a behavior evaluation and earn a set number of show and working trial points in order to be bred. Owners identified their dogs by identification number, which were later matched to SKC records.
This sample avoids the issue of questionable breed identification present in other studies and assures that all the dogs were members of the closed gene pools called breeds, but introduces potential owner biases. Owners who are members of a breed fancy are particularly likely to perceive their dogs’ behavior through the lens of breed-based expectations, and to cherry pick individuals they think will represent the breed well, resulting in the likelihood of a very high self-selection bias. In accordance with previous literature, the questions were condensed to 18 behavioral subscale scores (BSS). For each behavior, the owner reported two factors on a scale of 0-4: the owner’s assessment of the intensity of a behavior or of its frequency. Statistically significant differences between breeds were found for all 18 BSS.
The model accounted for, at most, 34% of variation on the BSS, and that included age, sex, breed, breed group, and several interactions between those variables. This means that the majority (66%) of the behavioral variation cannot be explained by these factors.
Like other studies that attempt to find breed differences in behavior, the authors began with the premise that dogs have been bred for specific jobs (e.g., hunting, herding) for hundreds or even thousands of years and therefore are likely to systematically differ in behavior. They do not, however, discuss or account for the fact that breeding pressures have changed considerably in the last century, with an emphasis now on morphology (Coppinger & Coppinger, 2001). Moreover, within the Fédération Cynologique Internationale (FCI), the most widely recognized association of purebred clubs in Europe, working dog classification results in a very arbitrary division of breeds for the purposes of this study. It is simply those breeds that the FCI has decided must be assessed through the “Dog Mentality Assessment” (DMA) to be qualified for breeding, and who are required to earn working trial and conformation (appearance) points to be declared champions who in turn are the individuals most likely to have opportunities to reproduce. Both groups are extremely diverse with regard to the historical work associated with the breeds, such that it would be surprising if they had any behavioral similarities based on their grouping. The differences that were found here between the working and non-working groups can all be most parsimoniously explained by the different husbandry involved. Dogs in the working group are much more likely to be trained to a defined level (affecting “trainability” and inclination to play with people) and must be able to withstand the very frightening stimuli of the DMA test. Dogs in the non-working group have no such experiential requirements.
These authors’ findings diverge considerably from the U.S. breed survey which used one of the same instruments but without breed verification ([Duffy et al., 2008]). For example, Chihuahuas scored an average of ≥ 1 on the aggression subscales except for owner-directed aggression in Duffy et al. (2008), but scored lower (≤0.8) on the same subscales in this study. That is, scores for the same breed were significantly different between theses two studies both using the same owner survey tests, indicating issues of validity with the survey. Asp, Fikse, Nilsson, & Strandberg (2015) posit that breeding stock differences could account for the discrepancy, or that handling differences and owner perceptions could be responsible. Such possibilities call into question the underlying assumption here that what is being compared are genetic differences; this may simply not be the case.
But even in the unlikely event that the results could be validated as attributable to genetic differences between breeds, given that results for the same breed differ from country to country it is impossible to know which result, if either, is representative of that breed.
Perhaps the most striking limitation in this, along with other studies based on the C-BARQ questionnaire, is the concept of the significance of the findings in the everyday language sense of the term. Statistical significance is a technical term, meaning that the differences found are fairly unlikely to be attributable to chance, that they represent an actual difference, however small. But in the case of behaviors related to aggression here, the average differences between breeds only range within slightly different positions at the bottom of each scale. The range of difference is between “never” and “rarely” or “no aggression” and the low end of “mild aggression” depending upon which scale is used on a particular question. This can be very misleading if the term “significant” is taken to mean “of practical importance,” which certainly has not been demonstrated here.
Abstract and Link to Purchase Full Text of the Original Article:
Coppinger, R., & Coppinger, L. (2001). Dogs: A startling new understanding of canine origin, behavior & evolution. Simon and Schuster.
Duffy, D. L., Hsu, Y., & Serpell, J. A. (2008). Breed differences in canine aggression. Applied Animal Behaviour Science, 114(3), 441-460. doi: https://doi.org/10.1016/j.applanim.2008.04.006
Svartberg, K. (2005). A comparison of behaviour in test and in everyday life: evidence of three consistent boldness-related personality traits in dogs. Applied Animal Behaviour Science, 91 (1-2), 103–128. https://doi.org/10.1016/j.applanim.2004.08.030