Introduction

The human-dog relationship is thought to be the oldest domestic animal partnership [1], serving the needs of both the human and the dog in a wide variety of ways [2]. However, these relationships are complex and can become problematic for humans and dogs when they become dysfunctional. One of the most common signs of dysfunctional human-dog partnerships are behaviour problems that, when unidentified and uncorrected, can present a clear and present danger to both species. Dogs with unidentified behavioural problems tend to be the ones that bite humans and other animals [3], that are returned more frequently after adoption [4,5] and are most likely to be euthanized at the owners’ request. In fact, it has been suggested that behaviour problems represent the single most cited reason for the relinquishing and euthanasia of dogs [4,6,7]. As such, identifying behaviour problems before they become larger issues is important in guaranteeing both dog and human health and safety. Once identified, most of these problems can be corrected, helping to change dysfunctional human-dog dyads into functional ones.

To identify problem behaviours and understand their origin, the dog’s behaviour must be evaluated. In general, direct behavioural observation by trained behaviourists is the preferred form of assessing and classifying dog behaviour. Various tests have been developed to do so, mainly in the form of test batteries, ratings of individual dogs, expert ratings of breed prototypes, and observational tests [8]. These tests are often time consuming, require specific settings [9], their results may depend on the experimental conditions [10], and they may be difficult to conduct on a larger and more varied population, making generalization across populations difficult [8]. One way around these issues is by using the knowledge an owner possesses about the dog to evaluate an individual dog’s behaviour and temperament [11]. Although not specifically trained to observe canine behaviour, simply by virtue of their co-habitation, an owner may be knowledgeable about their pet’s behaviour. As such, owners may represent a reliable source of information regarding their dog’s behaviour.

One way to quantify owner knowledge is through questionnaires such as the widely-used Canine Behavioural Assessment and Research Questionnaire (C-BARQ), a 100-item instrument originally developed in the USA [12,13]. So far, the C-BARQ has been used to evaluate canine behaviour and screen for appropriate temperament in dogs in guide dog programs [14,15], to identify specific behaviours related to the dogs’ hormonal response to human contact [16], and even to classify behaviour phenotypes in morphological and genetic studies [17,18]. The psychometric properties of the C-BARQ have been studied in a variety of countries and validated for use in Mandarin [19], Japanese [20], Dutch [21], Swedish [11], Italian [22], Farsi [23], Latin American Spanish [24], and Brazilian Portuguese [25], making it a tool that has shown consistency and validity in assessing dog behaviour in a wide variety of cultures. Common canine behavioural problems in various populations may have common origins, or they may be unique to specific cultures; using the same validated instrument makes such comparisons possible [26]. By identifying behaviour problems present in a given population, it becomes possible to develop educational programs for owners which would focus on prevention of these issues. Through owner education, it should be possible to reduce problem behaviour, leading to a reduction in the relinquishment and euthanasia of dogs, as well as human-directed aggression [27,28].

In Portugal, dog ownership has gone through many changes in the past 20 to 30 years, since the revolution of 1974, when dogs started to become more common inside the home. It has only been very recently that dog training classes have been made available to the public which, along with the increased availability of pet insurance, demonstrates a gradual cultural shift in how the Portuguese view the family dog. Despite this shift, Portugal continues to have a dog abandonment problem, with official numbers from 2017 citing 24,079 dogs accepted in municipal kennels, of which 31% were euthanized (personal communication, National Authority for Animal Health, Government of Portugal). With the approval of the new Animal Welfare Act of 2016 [29], in which the euthanasia of healthy dogs under municipal care has been prohibited, it is likely that the importance of correct rehoming of relinquished dogs will become even more important. As such, having a reliable and valid tool, such as the C-BARQ, to assess and correctly classify a particular dog’s behavioural characteristics in a quick, easy, and consistent way could greatly benefit municipal kennels. The C-BARQ could also serve to help clarify the behavioural characteristics of the Portuguese dog population, thereby helping to direct public education campaigns that may contribute to more responsible dog ownership. In a clinical setting, the use of the C-BARQ could help veterinarians to clearly identify problems and, as such, better help owners when behaviour issues begin to appear.

The present study aims to establish the psychometric properties of an adapted and shortened 78-item European Portuguese version of the C-BARQ. Such a questionnaire may be useful in classifying dogs for rehoming as well as identifying possible behavioural problems in owned dogs before they become so serious as to lead to abandonment or euthanasia.

Materials and methods

Instrument

The version of the C-BARQ used in the current study was based on the 100-item version used in the study by Duffy and Serpell [13], itself an updated version of the original C-BARQ [12]. The questionnaire’s 100 items ask owners to assess their dog’s reactions in everyday situations and score them on a Likert-type 5-point scale of frequency (0 representing “never”, 4 representing “always”) and of severity (0 indicating “no sign of the behaviour” and 4 indicating “severe demonstrations of the behaviour”). The questionnaire was translated from English to Portuguese, corrected by three university professors, and back translated by a native English speaker (Canadian citizen). The questionnaire was then administered to a small test population of owners (N = 50) and, after frequent comments regarding the perceived excessive length, items labelled as “miscellaneous” (items 77 to 90), were removed to shorten the questionnaire. The result was a European Portuguese version of the C-BARQ containing 78 items (Table 1), maintaining the 7 sections of the original, but excluding 22 Miscellaneous items.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. CBARQ sections and items translated into European Portuguese.

https://doi.org/10.1371/journal.pone.0209852.t001

Participants were invited to complete the C-BARQ online using Google Forms ^TM or in person through paper questionnaires distributed throughout the Greater Lisbon Metropolitan Area to various small animal hospitals, clinics and anti-rabies vaccination programs. Owners were instructed to complete the questionnaire as thoroughly as possible, however if they had no experience with the behaviour described, they were given the option to select “non-applicable” or “not observed”; these responses were treated as missing values in statistical analyses. Questionnaires were made available for a period of 8 months, resulting in 344 completed questionnaires.

Results

Population

The canine population under study was varied and is detailed in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Demographic characteristics of canine population in study (N = 345).

https://doi.org/10.1371/journal.pone.0209852.t002

Factor analysis

Through analysis of the correlation matrix using the Kaiser-Meyer-Olkin measure of sampling adequacy, a value of 0.812 was obtained [32], and a significant Bartlettˋs test of sphericity (χ2 = 12071.958; df = 3003; p<0.001) confirmed that the sample size is adequate for analyses using principal components analysis [33,34].

The scree plot and eigenvalues suggested a 13-factor structure, which were extracted with item loadings presented in Table 3. This structure explained 58.42% of the total variance. Most of the items loaded onto the same factors as the original study [13], with the exception of two factors and two items (as shown in Table 3). In Duffy and Serpell´s [13] study Dog-directed Aggression (DDA) and Dog-directed Fear (DDF) had 4 items loading onto two different factors, whereas in the current study all 8 items loaded onto a single factor renamed Dog-directed Fear/Aggression [12]. Duffy and Serpell [13] loaded item 8 onto the factor “Trainability” (TR, factor 10) whereas in the current study the item loaded onto the factor “Energy” (EL). Item 43 in the Duffy and Serpell [13] study loaded onto the “Touch Sensitivity” (TS) factor, whereas in the current study the item loaded onto the “Non-social Fear” (NSF) factor.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Results of factor analysis on the European Portuguese CBARQ.

https://doi.org/10.1371/journal.pone.0209852.t003

In all cases, items loading values were above 0.412, with the exception of item 30 (loading 0.366). Despite this lower value, the item represented at least 9% of the variance accounted for in the factor [35] and as such was maintained.

Internal consistency and internal validity

The internal consistency of extracted factors was analysed using Cronbach’s alpha, with values above 0.70 considered to have adequate reliability. The Cronbach’s alpha values ranged from 0.902 and 0.721 (Table 3), showing excellent to respectable consistency [30].

Significant correlations (p<0.01, p<0.05) were found between the 13 factors within the C-BARQ; the coefficients, which varied between 0.454 and 0.108, denoting mostly moderate or weak correlations [36]. However, some negative coefficients (in a few weak associations) and null associations were also detected. Item-factor correlations can be found in Table 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Item-factor correlation summary.

https://doi.org/10.1371/journal.pone.0209852.t004

Discussion

This paper set out to study the psychometric properties of the European Portuguese version of the C-BARQ to establish its validity for use in a European Portuguese context. The obtained results for this instrument suggest good validity and reliability indices, with a robust 13-item factor structure accounting for 58.42% of the total variance of the results. These findings reveal the important psychometric qualities of the instrument and highlight specific differences found in the current population compared to others studied.

The European Portuguese version of the C-BARQ very closely followed the structure of the original [13], with the extraction of almost all of the same subscales. The exception was the two subscales, DDA and DDF, each with 4 items loading strongly onto one factor that we renamed Dog Associated Fear/Aggression (Table 4). Although this result was similar to the results obtained by Svartberg [11], it contrasts clearly with studies carried out in other countries [19–21]. Portugal has only recently started to see the dog as a family member, and many dogs are still kept in yards. The importance of socializing dogs [37] is not widely acknowledged by Portuguese owners and, as a result, some dogs may show inappropriate behaviour when meeting an unfamiliar animal making the line between aggression and fear difficult to draw. This inexperience with dog behaviour could account for the grouping of DAF and DAA into a single factor.

When considering individual items on the European Portuguese C-BARQ, each loaded strongly on its expected subscale, except for two: items 8 and 43. The former, “will fetch or attempt to fetch sticks, balls, or objects” loaded onto the subscale EL instead of the original TR [13] as it did in a recent Mexican study [24]. In Portugal, dog training classes have only recently started to be regularly offered and, as in other countries, few owners attend [38]. It is possible that fetching is not considered to be an act of training but of playing. Dogs scoring high on the EL factor may tend towards a more extroverted personality [39] and may readily display fetch-like behaviours, but may not have been receptive to basic obedience commands such as “sit” and “stay”, items included in the TR factor. Item 8 is also the only TR subscale item that can be demonstrated by the dog when alone, making it more likely to be displayed by extroverted, high energy dogs.

The only other item that differed from the English C-BARQ was “when examined/treated by a veterinarian” (item 43), which loaded onto the subscale NSF instead of the original TS subscale [13]. As previously suggested [19], the reaction of a dog when examined by a veterinarian may not be an accurate measurement of touch sensitivity, but rather of fear, as the dog could be reacting as a result of a previous negative experience with veterinarians. It is the only item in the TS subscale that involves a potentially unfamiliar person, and the dog could be effectively reacting to fear of a novel person. This may be even more true in Portugal, where visits to veterinarians have traditionally been exclusively for obligatory rabies vaccinations instead of regular health care checks during the dog’s entire lifetime.

While great care was taken to try and obtain the most representative dog owner population possible by distributing the questionnaire in every parish in the Greater Lisbon Metropolitan Area, spanning a wide variety of socioeconomic classes, it must be noted that the experimental design required that owners volunteer to participate. As stated by various authors [19,38,40] these owners may be naturally more connected with their dogs, making them more observant than the general population. Although this effect can never be completely accounted for, the fact that almost identical factor structures where extracted from data in different countries [19–21,24,41] gives weight to the notion that the questionnaire does measure universal dog behaviours that are evident to most owners, regardless of individual characteristics, such as culture or attachment level.

Conclusion

The C-BARQ has been shown to be an effective instrument, both valid and reliable, that can be used cross culturally. Small differences that may arise between countries can be identified by validating new translated versions of the questionnaire before they are widely used [19]. This study has demonstrated that the European Portuguese version of the C-BARQ can confidently be used to help characterize the behaviour of the Portuguese dog population and, as such, direct any future public education endeavours. This is borne out by the excellent psychometric properties demonstrated both in terms of reliability and validity. The instrument is adequate for use in animal shelters to better match dogs with potential new owners and in clinical settings to identify behaviour problems in veterinary patients before they become unmanageable. The European Portuguese C-BARQ could be of vital importance to help resolve behavioural problems in owned dogs before they become so serious as to lead to abandonment or euthanasia, diminishing the pressure on municipal kennels and greatly improving canine welfare in Portugal.

Supporting information

Acknowledgments

The authors wish to thank all of the municipal kennels, veterinary hospitals, and clinics that participated in this study.