Undesirable behavior is prevalent among pet dogs, with 72–85% of dogs estimated to exhibit at least one type of problem behavior (1, 2). These behaviors may reflect normal canine behavior that owners simply find undesirable, such as barking, or they may arise from medical or behavioral pathology, such as generalized anxiety disorder or compulsive behavior (3). Canine behavior problems can often be indicative of a poor welfare state (4, 5) and lead to an increased risk of euthanasia. Undesirable behavior is the leading cause of mortality among privately owned dogs under the age of 3 years (6, 7). Problem behaviors have also been associated with reduced owner attachment and satisfaction (8–11) which may threaten the human-animal bond and place the dog at increased risk of relinquishment as behavioral issues are a leading cause of canine relinquishment worldwide (12–14). Many unsuccessful animal adoptions, in which animals are returned to the shelter post-adoption, can also be attributed to behavioral issues (15–17).

Animal shelters often assess the behavior of incoming dogs to help determine their suitability for adoption and match them with adoptive owners based on their behavioral needs and the adopter's ownership preferences (18). Concerns have been raised about the use of standardized behavioral assessments in shelters as some evidence suggests they have a poor ability to predict the future behavior of dogs in the home environment (19). Other researchers argue that standardized assessments provide useful behavioral information, particularly when combined with reports from previous owners (18). Survey instruments, such as the Canine Behavioral Assessment and Research Questionnaire (C-BARQ), have been used successfully to collect behavioral data from relinquishing owners (20, 21). Significant correlations have been documented between C-BARQ scores and canine behavior captured during shelter behavior assessments (22, 23). The shortened or “mini” version of the C-BARQ (42-items compared with the 100-item C-BARQ) has also been shown to discriminate between shelter dogs based on their outcomes (adopted/euthanized), and mirrors the assessment of shelter staff regarding the presence (or absence) of aggression, i.e., dogs that showed aggression in the shelter also had high scores on the aggression C-BARQ subscales as reported by the owner (20).

Despite these findings, widespread doubts exist within the animal sheltering community regarding the value and reliability of behavioral information provided by relinquishing owners (21, 24, 25). Such doubts have been partly reinforced by the conflicting results of previous research (20, 21). An initial study of 54 relinquishing dog owners found that owners who believed the information they provided was confidential (i.e., not shared with the shelter) reported higher rates of owner-directed aggression and stranger-directed fear on the C-BARQ than owners who were informed that the information would be shared with the shelter (21). The latter group may have misrepresented their dog's behavior, either due to social desirability biases or in an attempt to conceal the presence of behavior problems that might contribute to the pet's euthanasia. However, a second, larger study including 443 relinquishing owners failed to replicate these initial findings. Instead, confidential and nonconfidential groups of relinquishing owners had similar scores across all C-BARQ subscales (20). The authors of this study speculated that the difference in results may have arisen from the way in which the C-BARQ was administered (face-to-face interview in the preliminary study compared with self-completion in the latter). As relinquishing owners' reports appear to differ depending on the manner in which behavioral data is collected, we hypothesized that the quality of owner reports may also depend on how the topic of behavior problems is addressed at the time of pet relinquishment. For example, the C-BARQ, which was designed to reduce bias by asking owners to report the frequency or severity of behaviors in specific situations and contexts in the recent past (26), may evoke different responses from relinquishing owners compared with a direct question about the dog's behavior problems.

To test this hypothesis, the current paper utilizes previously unpublished data from the sample of relinquishing owners in the second of these studies (20) to compare owner's self-completed mini C-BARQ responses with their responses to a single direct written question regarding their dog's behavior problems. We also explore whether both sets of responses (mini C-BARQ and direct, written behavioral question) were influenced by the owner's perception of the confidentiality of the information they provided. The information provided by relinquishing owners may also be affected by their understanding of dog behavior as previous studies have found significant differences between owners in their perceptions of dog behavior (27–29). To characterize relinquishing owners' perceptions of dog behavior and elucidate possible differences between relinquishing owners and current dog owners, we compared the responses of the former with a matched sample of dog owners gleaned from the C-BARQ database at the University of Pennsylvania.

Four-hundred-forty-three relinquishing owners completed the mini C-BARQ, although 16 owners were excluded as they had a length of ownership <2 months. The final sample included 427 relinquished dogs and 427 pet dogs that were successfully matched for sex, breed, and source of acquisition (in most cases). There were significant differences in neuter status (X² = 5.53, p = 0.02) and age (X² = 10.82, p = 0.01) between the groups, although post-hoc analyses based on standardized residuals were not statistically significant. Pet dog owners had owned their dogs for significantly longer at the time of completing the C-BARQ compared with relinquishing owners (U = 59,274.50, Z = −4.85, p < 0.001, Table 1).

Relinquishing owners and pet owners differed in their perceptions of their dogs' behavior (X² = 108.03, p < 0.001, Figure 1). More than two-thirds of relinquishing owners reported their dog had no behavior problems (69.3%), 23.7% reported their dog had minor behavior problems, 5.9% reported moderate problems and 1.2% reported serious problems. Comparatively, 34.5% of pet owners reported their dogs had no behavior problems, 41.1% reported minor problems, 19.5% reported moderate problems and 4.9% reported serious problems. There were no differences in relinquishing owner's perceptions of their dog's behavior based on the confidentiality of their responses (X² = 1.44, p = 0.97, Figure 2) or the shelter at which they relinquished their dog (X² = 9.17, p = 0.14). The perceived confidentiality of relinquishing owners' responses was not associated with their dogs' mini C-BARQ subscale scores or miscellaneous items as described by Duffy et al. (20).

Relinquished dogs scored significantly higher than pet dogs on most C-BARQ subscales, including excitability [t₍₈₃₀₎ = −5.76, p < 0.001], energy [t₍₈₄₇₎ = −7.87, p < 0.001], training difficulty [t₍₈₂₄₎ = −11.16, p < 0.001], stranger-directed aggression (U = 96,870, z = −3.92, and p < 0.001), owner-directed aggression (U = 100,314.50, z = −4.43, and p < 0.001), dog-directed aggression (U = 85,436.50, z = −2.69, and p = 0.01), stranger-directed fear (U = 94,441, z = −3.37, and p < 0.001), touch sensitivity (U = 69,390.50, z = −3.33, and p < 0.001), and separation-related behavior (U = 93,334, z = −2.04, and p = 0.04; Figure 3). There were no significant differences between relinquished and pet dogs in dog rivalry (U = 58,034, z = −1.36, and p = 0.18), dog-directed fear (U = 82,682, z = −1.83, and p = 0.07), nonsocial fear (U = 89,589, z = −1.78, and p = 0.07) and attachment/attention-seeking behavior (U = 82,186.50, z = −1.79, and p = 0.07), while pet dogs scored higher than relinquished dogs for chasing [t₍₇₉₉₎ = 3.94, p < 0.001]. Relinquished dogs had significantly higher scores for all miscellaneous C-BARQ items, except for escaping behavior (p = 0.83; Figure 4).

The binary logistic regression models, including adjustment for sex, neuter status, age group and length of ownership, produced largely similar results (Table 2). Relinquished dogs scored significantly higher than pet dogs in excitability, stranger-directed aggression, owner-directed aggression, dog-directed aggression, touch sensitivity, training difficulty, energy, pulls on leash, tail chasing, urine marking, urinates when alone, and defecates when alone. Again, pet dogs had a significantly higher score than relinquished dogs on the chasing subscale, although they also scored higher for attachment/attention-seeking behavior.

The goal of this study was to investigate whether relinquishing owners' reports of their dogs' behavior were influenced by the manner in which the topic was addressed at the time of relinquishment. We also examined how the perceived confidentiality of the information provided by relinquishing owners influenced their responses to different behavioral screening tools. Finally, we used a matching sample of currently owned dogs to determine if relinquishing and current (non-relinquishing) owners perceive the behavior of their dogs differently.

The results suggest that the perceived confidentiality of the behavioral information provided by relinquishing owners did not affect their responses to either the direct question regarding the dogs' behavior problems or the mini C-BARQ (20). Owners who believed their responses would be shared with the shelter staff and used for adoption decision-making were just as likely to report behavioral problems as owners who believed their data were confidential and would not be shared with the shelter. In contrast to the C-BARQ, which is designed to reduce subjective biases by asking owners to report the frequency or severity of behaviors in specific situations (26), the direct question might offer relinquishing owners greater opportunity to misrepresent their pet's behavior. However, as the perceived confidentiality did not impact relinquishing owners' responses to the direct question, our findings challenge the notion that relinquishing owners deliberately deceive shelter staff or provide biased information at the time of relinquishment.

At the same time, highly significant differences emerged in how relinquishing and current dog owners responded to the two types of behavioral assessments. Irrespective of the perceived confidentiality of their reports, relinquishing owners were far less likely to report experiencing minor, moderate, or serious behavior problems when asked directly compared with a matched sample of current dog owners. Conversely, the mini C-BARQ scores of currently owned dogs were significantly and consistently lower (more desirable) for the majority of subscales and miscellaneous items compared with those of the relinquished dogs. In other words, based on their mini C-BARQ scores, relinquished dogs had significantly less desirable behavior than the matching sample of currently owned dogs, but their owners were less likely to report behavioral issues when asked directly than pet owners, despite being unaffected by the perceived confidentiality of their responses.

If relinquishing owners were deliberately under-reporting their dog's behavior problems, we would anticipate the perceived confidentiality of their responses would impact the owner's reports, particularly to the direct question. However, this was not the case. The discrepancy between owners' responses to the direct question and the mini C-BARQ (where they had to describe specific behavioral responses to stimuli) could therefore indicate that relinquishing owners' were “honest” in their reporting or that they under-reported their dog's behavior irrespective of confidentiality because, (a) they failed to read or comprehend the confidentiality statements, or (b) they did not trust the stated claim that their responses would not be shared with the shelter.

If relinquishing owners were indeed reporting “honestly,” our findings may indicate that relinquishing owners were less informed about normal canine behavior and therefore did not always recognize their dog's behavior as a behavioral problem. Thus, when asked to describe specific behaviors using the mini C-BARQ, relinquishing owners reported behaviors that produced less desirable scores for their dogs than current pet owners. But, when asked directly about their experience of undesirable behavior, a general lack of understanding of what constitutes normal dog behavior could have meant they were less likely to report behavior problems than the sample of dog owners who voluntarily completed the C-BARQ. It is conceivable that relinquishing owners with a limited understanding of canine behavior were less likely to access behavioral resources due to a perceived lack of need. Previous research has found owners who do not attend obedience classes are at greater risk of relinquishing their pets (14, 16). Owners who do not attend puppy training classes are also less likely to report behaviors, such as pulling on a leash or separation-related behavior, as a problem (34).

It is also possible that some relinquishing owners faced difficulties accessing behavioral resources which may have hindered their understanding of canine behavior. Equity of access is a significant issue in veterinary medicine (35), and barriers, such as cost, transportation, and opening hours, can also affect pet owners' ability to access behavioral resources and knowledge (36). A recent study in the U.K. investigated the utilization of free training classes by low socioeconomic dog owners and found dog owners with lower household incomes had higher drop-out rates notwithstanding the removal of the cost barrier (36). The authors suggested that non-financial barriers likely prevented these underserved dog owners from attending training courses. Interventions that aim to increase access to behavioral resources must go beyond simply offering free services to combat other issues of inequality, such as transportation, job flexibility and timing of classes. Owners' demographic characteristics can also affect their perceptions of canine behaviors. Lord et al. (34) found owners who were employed or students were less likely to recognize canine behavioral problems than retired and unemployed owners. Demographic characteristics were not included in the present study to reduce the response burden on participants, but the possible influence of demographic characteristics on relinquishing owners' perceptions of dog behavior presents an interesting avenue for future research. The risk of relinquishment may also be affected by owner characteristics, such as gender and age (37, 38).

The possibility that relinquishing owners are unable to recognize canine behavior problems also presents an opportunity for animal shelters to develop intake diversion programs. Many shelters implement proactive relinquishment reduction programs to help pet owners who are considering relinquishment to retain their pets, thereby reducing the number of animals entering shelters. Community support programs (i.e., programs that aim to support general pet owners, including those who are not considering relinquishment) are also common among shelters. Examples include food banks and free/low-cost spay and neuter services (39–42). Our data suggest behavioral intake diversion programs might help to increase retention rates by assisting at-risk owners to recognize and manage their dog's behavioral problems, and could provide a crucial resource to improve equity of access for some dog owners. Behavioral intake diversion programs can take many forms from phone counselling to basic obedience training classes and one-on-one behavioral consultations. One example is the Richmond SPCA's behavioral helpline (41, 43). Community dog owners can call or email the helpline to receive behavioral support, enroll in behavioral classes and/or organize a face-to-face behavioral consultation. Although the program has not been formally evaluated, shelter staff report the helpline is heavily used (43).

In a recent survey of 111 animal shelters regarding community support and intake diversion programs, behavior training classes were listed as the top-ranking program that shelters believed would most help dog owners to retain their dogs, followed by behavior consultations (42). Only 27% of shelters in the study offered training classes and many indicated that financial and staffing restraints prevented them from implementing such programs (42). Research into relinquishment prevention programs is limited and studies focusing on behavioral diversion programs targeting at-risk dog owners are almost non-existent (44, 45). Considering the potentially costly nature of behavioral support programs, the need for animal shelters to efficiently utilize their resources and the considerable benefits that may be attained, the effectiveness of behavioral intake diversion programs should be considered a priority for future research.

Regardless of the underlying reasons for the discrepancy between relinquishing owners' responses to the mini C-BARQ and the direct question about their experience of behavior problems, the current findings indicate that the mini C-BARQ, or similar behavioral assessment tools, are likely to provide animal shelters with reliable and accurate information about a relinquished dog's behavior. The results of numerous previous studies suggest that behavior problems are a major reason why owners surrender or return dogs to shelters (12, 14–16). Our finding that relinquished dogs' mini C-BARQ scores were consistently poorer than those of a matched sample of pet dogs also suggests that behavioral issues might play a major role in relinquishment decisions.

This study is subject to several limitations that should be noted. Given the difference in recruitment strategies, the two groups of dog owners are likely to reflect different subsections of the dog owning population. Pet dog owners in this study were self-selected, meaning they voluntarily sought out and completed the C-BARQ online and likely had some level of interest in animal behavior. Relinquishing owners, on the other hand, were recruited to participate by shelter staff and did not necessarily have the same level of interest in understanding the behavior of their animals. The voluntary nature of the study meant some relinquishing owners may have declined to participate and the sample may not be representative of all relinquishing owners. As participants were recruited by animal shelter staff, we could not ascertain a response rate for this group to explore this issue. Similarly, the group of pet dog owners in this study may be subject to self-selection bias and may not be representative of the average dog owner given their aforementioned interest in animal behavior. Relinquishing and continuing pet dog owners also differed in terms of how they completed the mini C-BARQ (pencil-paper at the shelter compared with online) which may have impacted their responses. The groups of dogs differed in neuter status, age and length of ownership which could have affected their behavioral scores, although the binary logistic regression models including adjustment for these covariates produced similar results to the unadjusted analyses. The logistic regression models were subject to some missing values, particularly regarding the length of ownership, which resulted in the exclusion of some cases. Owners' responses to the direct question may have been impacted by order effects. All respondents answered the direct question prior to completing the mini C-BARQ/C-BARQ and it is possible that some owners would have answered the direct question differently had they completed the C-BARQ first. Finally, the reasons for relinquishment were not available to researchers which precluded an investigation of the level of agreement between owner's perceptions of dog behavior, their dog's C-BARQ scores and their relinquishment reasons. Interestingly, the 30% of relinquishing owners who reported canine behavioral problems mirrors several previous studies that found behavior problems were the primary relinquishment reason for ~30% of relinquishments (12, 13, 46). Nevertheless, relinquishment reasons are often multi-factorial (47) and future investigations of the associations between relinquishment reasons and C-BARQ scores could provide more information about relinquishing owners' perceptions of problem behavior.

The findings of this study suggest that relinquishing owners provide more accurate information about their dog's behavior when asked about particular behaviors in specific circumstances and provide support for the use of validated behavioral questionnaires, such as the mini C-BARQ, to collect behavioral data from relinquishing owners at the time of relinquishment. When considering the owner's perceptions of their dog's behavior through direct question, we found most relinquishing owners reported they were not experiencing problems with their dog's behavior. The comparative sample of pet dog owners, on the other hand, were more likely to indicate their dogs were exhibiting mild, moderate or serious behavior problems. In contrast, relinquishing owners' responses to the mini C-BARQ indicated their dogs were exhibiting higher rates of undesirable behavior compared with the matched sample of pet dogs. The disparity between relinquishing owners' reports of their dog's behavior based on the direct question and the mini C-BARQ does not appear to be attributable to deliberately biased reporting from relinquishing owners as the perceived confidentiality (or lack thereof) did not affect relinquishing owner's responses regarding their dog's behavior. It is therefore possible that relinquishing owners had less understanding of dog behavior and did not recognize their dog's behaviors as problem behaviors. Animal shelters could increase owners' retention of their animals by implementing behavioral intake diversion programs to support dog owners who are at risk of relinquishment and help them to manage their pets' behavior.

The data analyzed in this study is subject to the following licenses/restrictions: The data are freely available on request to the corresponding author. Requests to access these datasets should be directed to serpell@vet.upenn.edu.

The studies involving human participants were reviewed and approved by University of Pennsylvania Institutional Review Board. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

JS conceived and designed the study. KK collected and compiled the data. DD performed preliminary analyses. LP conducted the final statistical analyses and drafted the manuscript. JS, LP, and BW interpreted the data. All authors contributed to manuscript revision, read, and approved the submitted version.

This research was supported by a Morris Animal Foundation grant #D07CA-071 to JS. LP's position is supported by the Arnall Family Foundation.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.