Aggressive behaviour by dogs is a considerable social problem, but the ability to predict which individuals may have increased aggressive tendencies is very limited, restricting the development of efficient preventive measures. There is a common perception that certain breeds are more likely to exhibit aggressive behaviour, which has contributed to the introduction of breed-specific legislation. The rationale for such legislation explicitly assumes high heritability of this trait while also implying relatively little variation within breeds; these assumptions are largely untested. We compared behavioural tendencies between 8 breeds that are subject to legislation in at least one country and 17 breeds that are not subject to legislation using two validated psychometric tools: the Dog Impulsivity Assessment Scale (DIAS), which scores elements of impulsivity, including a tendency for aggressive behaviour, and Positive and Negative Activation Scale (PANAS), which scores sensitivity to positive and negative stimuli (which may trigger aggressive responses). We found that the two groups of breeds do not differ significantly in the specific DIAS factor relating to aggressive behaviour, “Aggression Threshold and Response to Novelty”, or any other DIAS and PANAS factors. We found large variation in all behavioural tendencies measured by both psychometric scales within both groups and within each breed studied. Taken together, our findings indicate that breed alone is not a reliable predictor of individual behavioural tendencies, including those related to aggression, and therefore breed-specific legislation is unlikely to be an effective instrument for reducing risk.

Data were gathered via owner report using an online survey combining demographic information with the Dog Impulsivity Assessment Scale and the Positive and Negative Activation Scale. The website link was advertised online (via Facebook and relevant dog/breed specific groups, Twitter, pet fora, via the UK Kennel Club including their Bio-acquisition Research Collaboration page, and the Dog Science Group website). UK breeders of breeds of interest were emailed using the ChampDogs website and encouraged to participate. Participants were informed we were conducting research into breed differences in personality traits, but were not specifically informed about the primary research focus on breed-specific legislation.

To select legislated dog breeds for inclusion, a list of countries with breed-specific legislation was compiled (PETolog, 2017; RSPCA, 2016). Legislated breeds were then selected based on the criterion that the breed is banned or restricted in at least one country that has breed-specific legislation. Using this criterion, we identified 12 dog breeds and two breed “types” (Dogo Argentino “type” and Pit Bull “type”) subject to legislation. We focused on responses from dog owners living in the UK and the Republic of Ireland. A very limited number of dog breeds/types is banned in the UK, and therefore we could obtain data on breeds legally owned in the UK but banned in other countries. We did not attempt to collect data from dog breeds/types banned in the UK, but our online questionnaire was open to the public without restriction, and therefore we obtained some data on banned dog breeds from the British Isles as well. The UK legislation bans Pit Bull “type” dogs, and we obtained too few entries to create a separate group for this breed type. However, we pooled data from three individuals described as “Pit Bull type” (of unspecified breed) and one Dogo Argentino individual together with other legislated breeds in a comparison of legislated versus non-legislated types. The Staffordshire bull terrier is not banned in the UK, and was considered separately from “Pit Bull types”.

In order to define non-legislated breeds, the following criteria were used: (a) the breed was not classified as a legislated breed, (b) the breed did not belong to the same genetic cluster as any of the legislated breeds (since it was assumed closely-related breeds may show behavioural similarities resulting from common ancestry), and (c) data on the breed were available in our questionnaire database. With the exception of the breeds banned in the UK, all breeds had to be registered in the Kennel Club (UK). All dogs listed as crosses were excluded. Pedigree status could not be checked and so breed information is based solely on owner report. 

In addition to comparing the legislated versus non-legislated breed groups, we investigated differences between individual breeds. From the set of breeds used in that comparison, we selected breeds that had a sample size ≥ 20 individuals, with the exception of two legislated breeds - English Bull Terrier and Rhodesian Ridgeback. These two breeds were included in the breed-level analysis despite smaller sample sizes due to their relevance to the primary research question. This resulted in 8 legislated breeds and 17 non-legislated breeds used in the breed-level analysis.

DIAS and PANAS questionnaires

The DIAS consists of 18 items rated on a 5-point Likert scale: “strongly agree”, “agree”, “partly agree, partly disagree”, “disagree”, “strongly disagree”, with “not applicable” as the sixth option; 9 items are reverse scored. Values are summed and the total divided by five times the number of scored items to give an Overall Questionnaire Score (OQS) value between 0.2 and 1. The DIAS is composed of three factors: “Behavioural Regulation”, “Aggression Threshold and Response to Novelty” and “Responsiveness” (Wright et al. 2011). The scores for these factors were calculated from items that load on these factors according to Fadel et al. (2016). The PANAS is comprised of 21 items, and is scored in a similar way to the DIAS (Sheppard & Mills, 2002). This questionnaire is composed of two main factors: “Positive Activation” and “Negative Activation”.

Duplicates, blank entries, entries with four or more missing responses and those for dogs less than one-year-old or owned by the current owner for less than a year were removed from the dataset. We also excluded responses from dog owners not based in the UK or Ireland, to reduce bias resulting from cultural or linguistic differences. We compared the DIAS and PANAS scores between legislated and non-legislated groups (as defined above) and breed level differences for breeds with at least 20 individuals, with the exceptions noted in the previous section.

Link to R code provided in README file. 

.txt files