The separation anxiety statistic that was nine times too high
Doges Editorial · 2026-06-21 · 4 min read
A re-analysis of canine behavioral data from the University of Pennsylvania finds that previous estimates of anxiety and aggression in dogs were dramatically overstated. The actual separation anxiety rate is closer to 9%, not 86%.
A widely cited study reported that 86% of US dogs had moderate to severe separation anxiety. A new re-analysis in the Journal of Veterinary Behavior puts the actual rate closer to 9%.
The original figures came from a study by Bonnie Beaver of Texas A&M University, published in the Journal of Veterinary Behavior. Beaver analyzed data from 43,517 dogs enrolled in the Dog Aging Project using the mini C-BARQ questionnaire and found that more than 99% of US dogs showed at least one moderate-to-severe behavior. [1] The headline categories: separation and attachment behaviors 85.9%, aggression 55.6%, fear and anxiety 49.9%. Those numbers described a population in significant, chronic distress and shaped how millions of owners interpreted their animals' behavior.
How the original numbers were calculated
The new re-analysis, by James Serpell and Lauren Powell at the University of Pennsylvania, went back to the same 43,517-dog dataset. Serpell co-designed the C-BARQ with Yuying Hsu at Penn in 2003, making this partly a case of the instrument's creator correcting its misuse. Serpell and Powell found that Beaver's alarming figures stemmed largely from how the data had been analyzed and how certain behaviors had been categorized. [2]
The C-BARQ asks owners to rate how often their dog shows specific behaviors on a five-point scale. It captures frequency, not clinical severity or context. Behaviors associated with separation anxiety are split across two questionnaire categories, 'separation-related behavior' and 'attachment and attention-seeking,' while fear-related aggression is tracked in a third. A dog that whines briefly when its owner puts on a coat scores on the separation anxiety subscale alongside one that cannot eat or sleep for hours after a departure. Where the researcher draws the threshold determines what the data shows.
What the re-analysis found
Serpell and Powell ran the same dataset with different statistical approaches. Moderate to severe fear- and anxiety-related behaviors came in at approximately 14 to 28%, depending on the specific behavior. Separation-related behaviors: roughly 9%. Aggression varied considerably by context and situation.
Beaver's own 2026 paper, published in Veterinary Research Communications, broke the same 43,517-dog cohort into specific behavior categories. [3] That analysis found that 22.3% of dogs showed at least mild anxiety toward unfamiliar people, consistent with five previous international surveys, and that about 10% showed extreme responses to noise. Neither figure approaches 86%.
These rates align with what previous literature found and with what veterinarians, trainers and behavioral specialists report observing in practice. Certified animal behavior consultant Steve Dale covered the re-analysis on June 11, noting that the earlier study's methods created a distorted baseline that overstated clinical disorder across the board. [4]
When studies report that most dogs are anxious or aggressive, owners may begin to view normal canine behaviors through a pathological lens.
— Steve Dale, certified animal behavior consultant, summarizing the Serpell and Powell findings
How behaviors get classified as disorders
A dog that barks when a stranger approaches the house may be doing something thousands of years of cohabitation with humans shaped it to do. One that growls when startled may be communicating clearly and reasonably. One that gets upset when its owner leaves for work may be expressing normal attachment rather than a clinical condition.
All of those behaviors can become genuine problems depending on severity, frequency and whether they interfere with the dog's welfare or the household's functioning. Classifying them as disorders by default has real consequences for how owners interpret their dogs, how they respond to normal friction and whether they decide a dog can be kept at all.
Not every unwanted behavior should automatically be considered a behavioral disorder.
— James Serpell and Lauren Powell, Journal of Veterinary Behavior, University of Pennsylvania
Why inflated statistics affect shelters and veterinary care
Behavioral problems are one of the leading reasons dogs are surrendered to shelters. When owners believe their dog has a clinical anxiety disorder, it shapes how they interpret everything the animal does. It can lead to unnecessary interventions, unnecessary medication and the conclusion that the dog simply cannot be helped. Overstated statistics raise the likelihood of those outcomes.
The problem extends into veterinary appointments. A client who arrives convinced their dog has clinical separation anxiety, because they read that 86% of dogs do, enters the consultation differently than one where the clinician can assess the actual behavior. Statistics that precede the appointment distort what both owner and veterinarian observe.
Serpell and Powell are clear that real anxiety exists and matters. Severe separation distress that leaves a dog unable to function, phobias that prevent eating or sleeping, aggression that poses genuine risk: these warrant veterinary attention and treatment. Their argument is against conflating normal canine behavior with pathology.
A more accurate baseline
Separation anxiety affects roughly 9% of dogs, according to the re-analysis. Aggression and fear responses vary by context but fall far below the near-universal figures the original study suggested. Those corrected numbers give clinicians, trainers and shelter staff a more accurate starting point: a recognizable minority with a genuine condition, and a larger majority whose behavior is inconvenient but not pathological.
References
[1] Beaver BV. The prevalence of behavior problems in dogs in the United States. Journal of Veterinary Behavior, 2025. https://www.sciencedirect.com/science/article/abs/pii/S155878782400090X
[2] Serpell JA, Powell LR. Prevalence and Severity of Behavior Problems in Dogs in the United States: A Re-assessment. Journal of Veterinary Behavior, 2025. https://www.sciencedirect.com/science/article/pii/S1558787825001078
[3] Beaver BV. Owner-reported prevalence and severity of fear and anxiety in dogs. Veterinary Research Communications, 2026;50(4):284. https://pmc.ncbi.nlm.nih.gov/articles/PMC13110192/
[4] Dale S. New Analysis Finds Dog Behavior Problems Less Common Than Previously Reported. StevedalePetWorld.com, June 11, 2026. https://stevedalepetworld.com/blog/new-analysis-finds-dog-behavior-problems-less-common-than-previously-reported/