Summary & Analysis: Attacks by packs of dogs involving predation on human beings
October 21, 2021
No Comments
This paper covers a small series of unique case studies.
The National Canine Research Council Research Library houses, in one searchable database, scholarly materials in our areas of interest and expertise. Our goal is to make available descriptions of studies from the peer-reviewed literature in order to inform discourse and enable ongoing research through accurate representations of sources.1 We provide links to the abstracts and where to purchase full texts (some of which are open access). We hope that the Research Library will also be useful to journalists, persons engaged in canine-related occupations, grant makers, and any interested researchers or readers. We invite all those interested to make use of the Research Library, which is searchable by Author, Content Type, and Topic.
To meet the standards for inclusion in the research library, research papers must generally be:
We do not attempt to include every study that meets these criteria. This is neither practical nor desirable in our effort to streamline the literature review process for scholars. Instead, we have included the most comprehensive works, those that can be considered seminal in each area when such exist. We have also included those that are the most frequently cited in the literature whether or not the project’s methodological rigor merits this recognition.
The three content types in the Research Library are:
We strongly encourage you go back to the original sources to confirm that you agree with our analysis. When making attributions to material found after using this Research Library, the original source material should be cited. Material quoted directly from the Research Library should be credited to the National Canine Research Council. If you have questions or comments please contact us.
1. For a sample analysis of how findings can be distorted by poor choice and use of cited material, see the 2016 open access paper in the Journal of Veterinary Behavior, “Who is minding the bibliography? Daisy chaining, dropped leads, and other bad behavior using examples from the dog bite literature.” All three authors are affiliated with National Canine Research Council.
in selecting a companion dog
No better than flipping a coin
A literature review
This paper covers a small series of unique case studies.
DBRFs are extremely rare, and because research indicates that they are largely preventable and may disproportionately affect children, there has been a push to better understand the circumstances and variables that contribute to such incidents.
This 2019 study by Willen et al. provides strong evidence that “fear-induced aggression” in dogs housed in shelters is highly modifiable, raising critical concerns about the validity and ethical implications of behavior evaluations used to determine whether the dogs are “adoptable.” The researchers found that a simple human intervention—30 minutes of positive human interaction per day over five days—more than doubled the number of fearful dogs who passed the Safety Assessment For Evaluating Rehoming (the SAFER assessment). In addition to calling into question the reliability of commonly used instruments like SAFER, these findings suggest that many dogs labeled as aggressive and sometimes euthanized may, in fact, be experiencing fear that can be alleviated through brief, human engagement. The study also demonstrated shifts in dogs’ emotional states, as measured through provocative battery testing, further reinforces that behavior observed in stressful shelter environments may not reflect how a dog will behave in a home. Although the study’s design introduced some confounding variables that may distort or mask the effects of others (e.g., music, scent, and room environment), its takeaway is clear: behavior evaluations conducted without first addressing stress can produce misleading results that carry life-or-death consequences. This research underscores the need to rethink the role of behavior assessments in shelters and prioritize practices that support dogs’ emotional wellbeing before making outcome decisions.
This 2019 review by Patronek et al. of studies claiming to validate or support the reliability of canine behavior evaluations used in animal shelters is the most comprehensive review to date. Their findings are unequivocal: no current evaluation demonstrates the scientific rigor necessary to justify its use for predicting individual dog behavior or guiding high-stakes outcomes. Through detailed analysis of 17 peer-reviewed studies, the authors show that meaningful inter-rater, test-retest, and inter-shelter reliability is largely absent, and that claims of validity—whether construct, convergent, or predictive—are weak, methodologically flawed, or based on misunderstood statistical concepts. They clarify key distinctions between predictive validity (correlations at the group level) and predictive ability (accuracy for individuals), exposing how inflated expectations for behavior tests persist despite high false-positive rates and poor real-world applicability. They argue that without strong, replicable evidence of both reliability and validity across shelter contexts, behavior evaluations lack the scientific foundation required for determining dogs’ fates and should not be used as such.
This 2016 study by Fadel et al. explores how owner-reported impulsivity varied between working and show dogs of two popular dog breeds, Border Collies and Labrador Retrievers. They conclude that working lines showed small differences in impulsivity traits. While the study is frequently cited in discussions of breed-typical personality, its findings illustrate the critical difference between statistical and practical significance. The large sample size (1,161 dogs) likely drove the statistical differences, but effect sizes were minimal, and no behavioral relevance was demonstrated at the individual level. The study also faced key methodological limitations, including unverified owner reports of breed lineage, potential owner bias, and reliance on the Dog Impulsivity Assessment Scale (DIAS), a tool with limited validation. Furthermore, it did not control for environmental and training differences, especially among working lines, making it impossible to determine whether reported behavioral differences were inherited traits or learned responses. The findings highlight both the complexity of defining personality traits like impulsivity and the challenges of using owner surveys to draw meaningful conclusions about breed behavior.
This 2016 study by Stone et al. illustrates the persistent challenges in attempting to tie canine behavior to breed-related physical traits without rigorous, directly matched data. The authors explore potential associations between dog morphology—such as skull shape, height, and weight—and behavior, using data from the Swedish Dog Mentality Assessment (DMA) rather than owner surveys. While the use of standardized behavioral testing and registered purebred dogs avoids some common pitfalls like owner bias or misidentified breeds, the study is critically undermined by its reliance on morphological averages drawn from unrelated, unmeasured samples. Specifically, the physical characteristics used for analysis were taken from small groups of Australian show dogs or an unofficial breed website, while the behavioral data came from over 67,000 Swedish dogs tested years earlier. No actual morphological data were collected from the dogs whose behaviors were studied, making any claimed correlations between body type and behavior purely speculative. Although the authors reported links between shorter or lighter dogs and increased fearfulness or aggression, and taller or heavier dogs and sociability or boldness, the flawed methodology renders such findings questionable.