Investigating canine personality structure using owner questionnaires measuring pet dog behaviour and personality

Investigating canine personality structure using owner questionnaires measuring pet dog behaviour and personality

Accepted Manuscript Title: Investigating canine personality structure using owner questionnaires measuring pet dog behaviour and personality Author: D...

509KB Sizes 0 Downloads 79 Views

Accepted Manuscript Title: Investigating canine personality structure using owner questionnaires measuring pet dog behaviour and personality Author: Diana J. Rayment Richard A. Peters Linda C. Marston Bert De Groef PII: DOI: Reference:

S0168-1591(16)30091-0 http://dx.doi.org/doi:10.1016/j.applanim.2016.04.002 APPLAN 4233

To appear in:

APPLAN

Received date: Revised date: Accepted date:

5-8-2015 11-1-2016 4-4-2016

Please cite this article as: Rayment, Diana J., Peters, Richard A., Marston, Linda C., Groef, Bert De, Investigating canine personality structure using owner questionnaires measuring pet dog behaviour and personality.Applied Animal Behaviour Science http://dx.doi.org/10.1016/j.applanim.2016.04.002 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Investigating canine personality structure using owner questionnaires measuring pet dog behaviour and personality Diana J. Rayment*, Richard A. Peters, Linda C. Marston, Bert De Groef School of Life Sciences, La Trobe University, Bundoora VIC 3086, Australia

Author e-mail addresses: [email protected] (DJR), [email protected] (RAP), [email protected] (LCM), [email protected] (BDG)

* Corresponding author:

Diana Rayment Rm 234, Biological Sciences 1, La Trobe University Bundoora, Victoria 3086 Australia Tel. +61 3 9479 1481 E-mail: [email protected]

1

Highlights    

We examined relationships between context-specific behavioural traits and personality traits in pet dogs. Several trait clusters were found, which included personality traits and context-specific traits. ‘Neuroticism’, ‘Extraversion’ and ‘Behavioural Regulation’ were analogous to traits previously described in animals. Some trait correlations varied significantly between owner groups with different levels of dog-related work experience.

2

Abstract Questionnaire-based personality and behavioural assessments for domestic dogs are widely used by professional dog handlers, researchers and those working in welfare. In order for behaviour tests to provide accurate, useful information about the future behaviour of test subjects, they need to reflect aspects of personality despite the limited context and time in which they are conducted. It is therefore important for questionnaires used to assess external validity of behaviour tests, to measure aspects of canine personality, and for assessors to understand how context-specific behavioural traits, such as those measured in battery tests, relate to broader personality traits in domestic dogs. Furthermore, the reliability of questionnaire-based tools using behavioural ratings can be significantly affected by questionnaire design and application, particularly when owners are asked to rate their own dogs and when the experience of raters is varied. This paper investigates the hierarchical structure of canine personality by assessing correlations between factor scores from three published owner questionnaires, two of which were designed to assess aspects of canine personality and one designed to assess narrow, behavioural traits that are somewhat contextspecific. Between-factor correlations were also compared between sub-groups of owners and dogs based on demographic factors, to investigate the stability of these relationships in a mixed sample of volunteer pet dog owners and pet dogs. Several broad clusters of traits were identified, which each correlated with a number of context-specific behavioural traits. Three of these traits, ‘Neuroticism’ and ‘Extraversion’ in the Monash Canine Personality Questionnaire-Revised and ‘Behavioural Regulation’ in the Dog Impulsivity Assessment Scale, show similarities to personality traits identified in other species. However, notable differences were identified in the relationships between some factors in owner groups which varied in dog-related work experience, which calls in to question the validity of structure identified. These variations may reflect a lack of reliability in some measures when experience level within a rater sample is mixed. Further investigation of the hierarchical structure of canine personality is required, using tools which behave more reliably with a mixed group of raters. 3

Key words: canine; temperament; behaviour; questionnaire

Abbreviations: CBARQ: Canine Behavioural Assessment and Research Questionnaire; DIAS: Dog Impulsivity Assessment Scale; MCPQ-R: Monash Canine Personality Questionnaire-Revised; PCA: principal components analysis

4

1. Introduction Questionnaire-based personality and behavioural assessments for domestic dogs are widely used in both research and practice by professionals working with canines. Questionnaires often fulfil multiple roles, functioning as stand-alone tools for suitability assessments for canine working, breeding, pet and service roles (Duffy et al., 2014), as well as tools through which the nature of canine personality and behaviour are investigated (Asp et al., 2015; Tamimi et al., 2015). In addition, questionnaires are often used to assess the concurrent validity of battery-style behaviour tests, such as those used in animal shelters and by working dog professionals (Haverbeke et al., 2009; Siracusa et al., 2010). Even though battery-style tests are strongly affected by the context in which they are performed, the primary aim of such tests is to reveal personality traits that allow the assessor to predict responses outside of the test environment (Taylor and Mills, 2006; Rayment et al., 2015). As such, questionnaires specifically designed to assess canine personality are desirable for the assessment of concurrent validity of behavioural tests.

A number of questionnaires specifically designed to assess canine personality exist, although these typically measure broad, higher-order traits (Jones, 2008; Ley et al., 2009; Fratkin et al., 2013). In order for these questionnaires to be useful for behaviour test developers, the relationship between broad personality traits and observable behavioural tendencies must be understood. Personality can be thought of as “those characteristics of individuals that describe and account for temporally stable patterns of affect, cognition, and behaviour” (Gosling, 2008, pg. 986). Personality characteristics can be understood at a number of levels within a hierarchical structure of personality, which varies from directly observable and measureable behavioural patterns in individuals within specific contexts, to abstract constructs used to explain consistent patterns of variation seen between individuals within a population and across a number of contexts (Gosling, 1999; Uher, 2011). The constructs within the upper most level of the hierarchy are sometimes referred to as 'super traits', being those traits that explain some degree of co-variance between recognised personality traits (Svartberg, 2002). In dogs, 5

the 'shy-bold axis' (Svartberg, 2002; Horvath et al., 2007; Starling et al., 2013a) and 'Consistency' (Fratkin et al., 2013) have been suggested as two such super-traits. Directly below the super-trait level, are those constructs generally referred to as personality traits, the most commonly accepted model of which is called the Big Five or Five Factor Model (McCrae and John, 1992; Gosling, 1999; Freeman and Gosling, 2010). There appears to be growing agreement within the literature that in domestic dogs, there are most likely five or six traits at this level, including Extraversion/Activity, Aggression/Self-assuredness, Fearfulness/Neuroticism, Responsiveness to Training, Sociability/Amicability and Submissiveness (Svartberg and Forkman, 2002; Jones and Gosling, 2005; Ley et al., 2008; Fratkin et al., 2013). Within a hierarchical model of personality, a number of narrow behavioural traits or tendencies cluster under each broad personality trait. For example, neophobia (non-social fear), poor resilience to stress and xenophobia (social fear) should all correlate with measures of Fearfulness/Neuroticism (McCrae et al., 1998). Consensus within the literature on the structure of canine personality below the broad-trait level is not clear, despite the constructs becoming less abstract and more directly measurable. As battery-style behavioural tests can only measure directly observed behaviours, this lack of consensus is problematic for test developers wishing to use personality questionnaires to assess test validity. As discussed by Uher (2011), one can examine the relationships between narrow behavioural tendencies and personality traits by examining correlations between measures of each and identifying clusters of traits that correlate.

Recently, in addition to whole-personality models, the topic of impulsivity in dogs has also garnered research interest, due to links to personality traits such as aggression, inattentiveness and activity (Hejjas et al., 2009; Wright et al. 2011; Wan et al., 2013). In humans, the four facets of impulsivity (Premeditation, Perserverence, Urgency and Sensation-Seeking) have been shown to correlate with three of the five traits in the Five Factor Model (Whiteside and Lynam, 2001). In dogs, impulsivity is thought to consist of at least three measures - Behavioural Control, Response to Novelty and

6

Responsiveness (Wright et al., 2012); however, the relationship between these measures and existing whole-personality models for canines has not yet been described.

In addition to questionnaires specifically designed to measure canine personality, questionnaires developed to assess narrow behavioural traits are also sometimes used as proxy measures of personality. One example is the Canine Behavioural Assessment and Research Questionnaire (CBARQ), which was developed to aid in identifying problematic behaviours in pet dogs within a clinical setting (Hsu and Serpell, 2003). CBARQ has since been used across a number of applications including the assessment of canine personality (Duffy and Serpell, 2012), broad-scale assessment of behavioural traits in dogs across different cultures (Nagasawa et al., 2011; Tamimi et al., 2015), to gather information from owners relinquishing their pets to a shelter (Duffy et al. 2014), and for concurrent validity assessment of battery tests for aggression (Siracusa et al., 2010). If CBARQ reflects narrow, lower-order personality traits that correlate with existing broad personality traits already identified in dogs, these relationship can provide information on the nature of canine personality. Knowledge of how CBARQ factors relate to broader traits could also increase its usefulness as a tool for the assessment of external validity of battery-style behaviour tests.

This paper aims to investigate the relationship between behavioural traits and personality traits in pet dogs, by investigating correlations between the calculated factor scores from three published owner questionnaires: CBARQ, the Dog Impulsivity Assessment Scale (DIAS) (Wright et al., 2011) and the Monash Canine Personality Questionnaire-Revised (MCPQ-R) (Ley et al., 2009a). As the design of behavioural questionnaires can have a significant impact on findings, particularly when owners are asked to rate their own dogs (Blackwell et al., 2013) or raters vary in their experience with dogs (Wan et al., 2012; Mirko, et al., 2013; Fratkin et al., 2015), our second aim was to explore whether these between-factor relationships are robust to variation in the relative experience of the observer,

7

by re-examining correlations between factor scores after splitting the sample according to owner (and dog) demographic variables.

2. Materials and methods This study is compliant with the Australian National Statement on Ethical Conduct in Human Research, and was approved by the La Trobe University Human Ethics Committee (approval number FHEC14/R15).

2.1. Survey procedure We constructed an online survey by combining published questionnaires with questions relating to dog and owner demographics as follows: 1. Custom-written questions relating to dog and owner demographics such as age, sex and training history of dog, and education and dog-related work experience of owners (full details given in Supplementary Materials); 2. An open-ended question asking owners to describe their dog’s personality in their own words; 3. DIAS (Wright et al., 2011), which includes measures of self-control, neophobia and response to novelty, impulsive aggression and responsiveness to humans; 4. CBARQ (Hsu and Serpell, 2003), which features 13 factors designed to assess behavioural responses to a broad range of contexts commonly encountered by pet dogs; 5. MCPQ-R (Ley et al., 2009a), a personality assessment for pet dogs derived from the psycholexical approach.

Using social media websites and Australian-based discussion forums for animal enthusiasts, we recruited 732 pet dog owners that were over 18 years of age and had owned their dogs for a minimum of six months. The questionnaire was conducted using Qualtrics online survey software 8

(Qualtrics Labs, www.qualtrics.com) and took approximately 30-40 minutes to complete. Of the initial 732 participants, 188 did not provide answers to all required sections of each questionnaire; these surveys were removed from the sample, resulting in 544 completed questionnaires that were used for analysis.

2.2. Statistical analysis Results were analysed using IBM SPSS Statistics 22.0. Scores for each factor assessed in the questionnaires (Table 1) were calculated according to published guidelines. Correlation analyses between survey factor scores were conducted using Spearman rank order comparisons; due to the large sample size, only correlations with rs > 0.3 and significance values of P < 0.01 were considered significant for the purposes of the study. Correlation analysis was also conducted on subsets of the sample split according to owner and dog demographics, including formal work experience of owner, nature of formal work experience of owner, sex of dog, and formal training history of dog. Fisher’s ztransformation was used to compare Spearman’s correlation coefficients, as described by Myers and Sirois (2006). Those pairs of correlations with statistically significant differences at the P < 0.01 level were examined. Exploratory principal component analysis (PCA) using direct oblimin rotation was used for each questionnaire, to assess how well the data fitted the respective published model. Visual analysis of scree plots and Monte Carlo parallel analysis were conducted for each PCA to determine the most appropriate number of factors for each questionnaire.

3. Results 3.1. Participants Of the 544 completed surveys, 504 participants (93%) were female with the majority of respondents (75%) aged between 26 and 55 years. Most participants (88%) lived in Australia, had owned at least one dog prior to the dog on which they reported (95%), and were educated beyond secondarycollege level (83%). The sample included a large number of owners experienced with dogs, with 52% 9

having owned at least four dogs prior to the one they were reporting on, and 48% having had formal work or volunteer experience with dogs, including experience in the veterinary (10%), training (22%) and welfare (18%) sectors.

3.2. Dogs Male and female dogs were equally represented in the sample group, with the majority of dogs of both sexes being desexed (neutered males 80%, spayed females 86%). Almost two thirds of the sample group (62%) were reported as purebreds representing all 10 Fédération Cynologique Internationale groups (http://www.fci.be/en/Nomenclature/). The remaining dogs were of unknown or mixed breeding. The vast majority of dogs were micro-chipped (95%) and vaccinated (88%). A small percentage of dogs (16%) lived with no other animals, while more than half (56%) lived in multi-dog households and 40% lived with cats. Dogs were primarily obtained by their owners as companions or pets (92%), with a small number of dogs sourced for working, service or breeding roles (8%). Dogs were most commonly reported as purchased from pedigree breeders (35%), the welfare sector (26%) or non-pedigree breeders (16%), with smaller numbers of participants sourcing their dogs from family and friends (10%), ‘Free to good home’ advertisements (5%), pet shops (5%) and ‘Other sources’ (< 5%).

3.3. Principal component analyses of questionnaires Principal component analysis of the CBARQ questions revealed a 13-factor solution that explained 56.1% of the variance (Table 1). The factors mostly resembled those described by Duffy and colleagues (2008), although a number of fear- and aggression-related questions loaded together onto individual factors; the ‘Energy’ and ‘Excitability’ scales combined into one factor, and ‘OwnerDirected Aggression’ split over two separate factors (see Table S1 in Supplementary Materials for more details). Principal component analysis of the DIAS questionnaire revealed a three-factor solution that explained 43.3% of the variance (Table 1). However, a number of individual questions 10

did not load onto any of the factors (see Table S2 in Supplementary Materials). Principal component analysis of the MCPQ-R reproduced the factors described by Ley and colleagues (2009b) and explained 61.9% of the variance (Table 1, Table S3).

3.4. Relationships between factors A number of statistically significant correlations of moderate (0.3 < rs < 0.5) or large (rs > 0.5) size were found between factor scores from the different questionnaires (Table 2). The MCPQ-R factor ‘Neuroticism’ showed positive correlations with three CBARQ fear-related factors and DIAS 'Response to Novelty', as well as a negative correlation with MCPQ-R 'Amicability’. MCPQ-R ‘Amicability’ showed negative correlations with several CBARQ factors relating to fearful and aggressive behaviour, as well as two measures of impulsivity. MCPQ-R ‘Training Focus’ showed a moderate negative correlation with DIAS ‘Behavioural Regulation’ and a moderate positive correlation with DIAS ‘Responsiveness’. Lastly, MCPQ-R ‘Extraversion’ showed positive correlations with DIAS ‘Behavioural Regulation’ and CBARQ ‘Excitability’.

Within the CBARQ questionnaire, a number of factors correlated with each other. Measures of social and non-social fear correlated with measures of aggression towards unknown people and dogs. CBARQ ‘Excitability’ correlated with both measures of separation-related problems. The factors ‘Trainability’ and ‘Chasing’ did not correlate with any other CBARQ factors, while ‘Touch Sensitivity’ only correlated with ‘Dog-Directed Fear’. Two inter-factor correlations were observed in the MCPQ-R results, with 'Extraversion' positively correlating with 'Motivation', and 'Amicability' negatively correlating with 'Neuroticism'. No correlations of greater than rs = 0.267 were noted within DIAS.

3.5. Factor relationships and owner demographics In order to determine whether owner experience with dogs affected between-factor correlations, the sample group was split accordingly and the relationships between all factor scores were re11

examined. When data were analysed this way, a number of significant differences were seen in the correlations between factors, depending on the amount and type of dog-related work experience reported by owners (Table 3; also see Table S4 in Supplementary Materials). Between-factor relationships differed notably when owners with no dog-related work experience and those that had formally worked or volunteered with dogs were compared. When the sample was broken down further according to nature of the owner’s dog-related work experience, it was found that factor relationships varied considerable between inexperienced owners, compared to dog trainers and those with veterinary experience (Table S4).

4. Discussion Our PCAs indicate that the questionnaires worked mostly as defined by the published models, although a number of variations were seen in the factors of both CBARQ and DIAS. We also found that two out of the three DIAS factors and four of the five MCPQ-R factors were moderately correlated with three or more other factors. These clusters aligned with the personality factors MCPQ-R ‘Neuroticism’, MCPQ-R ‘Extraversion’, MCPQ-R ‘Amicability’, MCPQ-R ‘Training Focus’, DIAS ‘Behavioural Regulation’ and DIAS ‘Response To Novelty’. These relationships can be interpreted within a hierarchical structure (see Figure S1 in Supplementary Materials for a suggested structure), although correlations between some factors were influenced strongly by the presence and nature of formal dog-related work experience of reporting owners. A number of factors showed no or only one correlation with factors from other questionnaires, possibly indicating that these factor scores reflect tendencies that are highly influenced by environmental variables.

4.1. Relationships between factor scores In the present study, a number of factor clusters were identified which align largely with recognised personality traits identified in the literature. DIAS ‘Response To Novelty’, which measures aspects of both neophobia and aggression (Wright et al., 2011), correlated with both MCPQ-R 'Amicability' and 12

MCPQ-R 'Neuroticism'. The MCPQ-R trait ‘Amicability’ is defined by Ley and colleagues (2009a;b) as how a dog reacts to other individuals and includes both pro-social behaviours and a lack of aggressive behaviour. In the present study, MCPQ-R ‘Amicability’ negatively correlated with a number of factors related to social fear, aggression towards people or dogs, and poor behavioural control. While this appears logical as aggressive or fearful dogs are generally not considered friendly, and supports the findings of Ley and colleagues (2009a), this finding contradicts that of other work assessing sociability in dogs. Using a variety of assessment methods, several authors have found that sociability and human-directed aggression are best described as two separate traits, which are independently reliable and valid (Svartberg and Forkman, 2002; Jones and Gosling, 2005; Mirko et al., 2012; Fratkin et al., 2013). As CBARQ is primarily a tool for assessing problematic behaviours and behavioural traits in pet dogs in a clinical setting (Hsu and Serpell, 2003), it is expected that measures of companionable or affiliative behaviours in dogs would not be included within the questionnaire. This limits analysis of these traits in the current dataset. However, calmness and a friendly, social disposition are important considerations for pet dog owners (King et al.,2009) and therefore also for assessors wanting to determine the suitability of dogs for pet roles. As such, it is vital to investigate whether it is most useful to treat amicability and aggression as separate personality traits for this purpose, and if so, how to best assess these traits separately in questionnaires and behavioural tests.

Neuroticism is generally considered to reflect the degree of nervousness, negative affect and emotional stability exhibited by a person (McCrae and John, 1992; McCrae et al., 1998) or dog (Ley et al., 2008; 2009b). The three CBARQ factors relating to fear, ‘Stranger-Directed Fear’, ‘DogDirected Fear’ and ‘Non-Social Fear’, correlated with MCPQ-R ‘Neuroticism’, as well as several CBARQ factors related to aggression. This possibly indicates that the CBARQ factors may be different behavioural manifestations of an underlying pre-disposition towards emotional reactivity, neophobia and fearfulness. Similarly, Wright and colleagues (2011) noted that questions related to 13

impulsive aggression and a poor response to novelty co-vary. The two CBARQ factors relating to attachment and separation distress only weakly correlated with MCPQ-R ‘Neuroticism’, but positively correlated with each other. This result is consistent with the findings of Temesi and colleagues (2014), who examined several questionnaire-based measurements of fearfulness and anxiety in pet dogs; the authors found four traits that were labelled ‘Neuroticism’ (including items indicating non-social fears and neophobia), ‘Human-Directed Fear’ (reflective of responses towards unknown people), ‘Dog-Directed Fear’ (fear responses to unknown dogs), and ‘Separation-Related Anxiety’. The first three items loaded onto one higher-order factor labelled 'Fear-Related Behaviour', while ‘Separation-Related Anxiety’ remained independent. This trend for behaviours indicative of emotional reactivity, curiosity (or lack thereof) and neophobia to co-vary was also noted by Svartberg and Forkman (2002), who labelled this trait 'Curiosity/Fearlessness' in experiments using direct observation of dogs during battery-style tests. As noted by Ley and colleagues (2008), questionnaire items related to curiosity and investigation in novel circumstances seem to correlate with items that assess fear in dogs, but not in humans. Rather, curiosity and an interest in novel ideas is a key component of the Five Factor Model trait 'Openness to Experience' in people and is assessed independently of a person's response to novelty or their tendency to be fearful (McCrae and John, 1992; Jackson et al., 2012). As it is difficult to assess internal states and motivations in animals, curiosity in dogs is generally measured as short approach latency and lack of overt behavioural signs of fear towards startling or novel stimuli, such as loud noise, suddenly appearing items or new places (Svartberg and Forkman, 2002). As such, the relationship between curiosity, neophobia and emotional stability in these results could simply be reflective of the novel settings in which curiosity and fearfulness in dogs are generally assessed.

In agreement with the findings of Ley and colleagues (2009a), MCPQ-R ‘Amicability’ also negatively correlated with MCPQ-R ‘Neuroticism’, indicating that these factors have a moderate relationship in this sample. Both MCPQ-R 'Amicability' and MCPQ-R 'Neuroticism' cluster with DIAS 'Response To 14

Novelty' in this dataset. As discussed by Ashton and Lee (2005) when considering human personality assessments, this situation may be suggestive of both traits measuring some aspects of a separate, third factor. The authors suggest that for humans a three-factor structure is most suitable to describe variations in social behaviour and emotional stability, with separate traits to describe anger-hostility, fearfulness and emotional sensitivity, and deceit and pretentiousness. In non-human animal studies investigating coping strategies, interactions between emotional reactivity and the predisposition of an animal to behave either proactively or reactively in response to a perceived threat, have been used to describe a variety of coping strategies that are characterised by either active aggression, flighty behaviour, calmness, or fearful passivity (Koolhaas et al., 2007). In dogs, it has been proposed that these behavioural predispositions, or coping styles, may also be related to sociability in non-threatening contexts (Svartberg, 2005). This concept, termed the ‘shy-bold axis’, appears to show similarities to the cluster of traits identified here, and perhaps gives a good starting point at which to consider how traits reflecting sociability, emotional stability and aggression could be conceptualised in a useful way in domestic dogs.

According to Ley and colleagues (2008; 2009b), the MCPQ-R ‘Extraversion’ factor is related to MCPQR ‘Motivation’ and reflects the energetic, lively and excitable nature of dogs. The MCPQ-R ‘Extraversion’ factor is analogous to the 'Activity' trait identified in a variety of non-human animals (Gosling, 1999; Gosling and John, 1999), including dogs (Jones and Gosling, 2005). It is therefore not surprising that MCPQ-R ‘Extraversion’ positively correlated with the factors MCPQ-R ‘Motivation’, CBARQ ‘Excitability’ and DIAS ‘Behavioural Regulation’ in our study, as these traits reflect activity levels and behavioural indicators of impulse control. According to Ley and colleagues (2008), MCPQR ‘Motivation’ reflects self-assuredness and descriptors sometimes indicative of a tendency towards competitive aggression and social confidence in dogs, such as tenacity, perseverance, and assertiveness. As both CBARQ ‘Owner-Directed Aggression’ and ‘Dog Rivalry’ are thought to reflect some degree of competitive aggression or resource protection, the lack of correlation between 15

these factors and MCPQ-R ‘Motivation’ is unexpected. This perhaps indicates that the behaviours scored in these CBARQ factors are heavily influenced by context-specific factors.

The DIAS factor ‘Behavioural Regulation’ reflects a general lack of behavioural control (i.e. the dog does not appear to think before acting) and ease of arousal (Wright et al., 2011). It is therefore logical that it would correlate positively with CBARQ ‘Excitability’ , CBARQ ‘Separation-related problems’ and MCPQ-R ‘Extraversion’, and negatively with both MCPQ-R factors ‘Training Focus’ and ‘Amicability’, as these factors all encompass some degree of proneness to overt behavioural displays of arousal and lack of inhibition. The correlation between DIAS ‘Behavioural Regulation’ and MCPQ-R ‘Extraversion’ is noteworthy; Eysenck and Eysenck (1978) reported a similar correlation between the factors ‘Impulsiveness’ (reflective of tendencies towards risk-taking, lack of planning and liveliness) and ‘Extraversion’, indicating that these traits are perhaps analogous in the two species.

Few other moderate to strong relationships were identified between factors, possibly indicating that these scores are highly affected by context and learning history. DIAS ‘Responsiveness’, which is described by Wright and colleagues (2011) as a general responsiveness to the environment resulting in ease of training, fast reactions and a prolonged interest in novel items, correlates positively with MCPQ-R ‘Training focus’, indicating that DIAS 'Responsiveness' may specifically relate to trainingrelated contexts. CBARQ 'Trainability' also weakly correlates with these factors, lending support to this interpretation. The CBARQ factors ‘Chasing’ and ‘Owner-Directed Aggression’ show no correlations with other factors in this dataset, possibly indicating that these tendencies are very context-specific and not highly affected by broader personality traits. This would mean that behavioural tests for these tendencies may be problematic if assessed outside of the contexts in which these behaviours normally occur, and that management of exposure to those contexts likely plays a significant role in how often they are reported by owners. Svartberg (2005) came to a similar conclusion, finding that behaviour tests measures originally thought to reflect predatory behaviour 16

correlated instead with human-directed play interest, and that aggressive tendencies towards people did not correlate with other behavioural measures.

4.2. Factor relationships and owner demographics The relatively large portion of owners in this sample who reported formal work or volunteer experience with dogs, allowed for examination of the effect of this experience on the correlations found between factor scores. Previous studies using DIAS, CBARQ and MCPQ(-R) rarely report owner demographics and experience levels, and no authors have cautioned against the use of their questionnaires with rater groups with varied levels of dog-related experience. While the literature on the effect of experience on inter-rater reliability is mixed, previous reports have suggested that formal and informal experience with dogs does not significantly affect rater ability to accurately rate overt behaviours (Tami and Gallagher, 2009; Fratkin et al., 2015). In this sample, eight factor relationships proved to be unstable when rater experience varied. CBARQ ‘Dog Rivalry’, CBARQ ‘Owner-Directed Aggression’, DIAS 'Behavioural Regulation' and DIAS 'Response To Novelty' all showed varying correlations with at least two other factors, indicating that they may be comparatively more sensitive to rater experience levels than other factors. Despite the lack of previous literature to suggest that these questionnaires may be sensitive to observer work experience with dogs, these differences in ability to subjectively rate dog behaviour are not entirely unexpected. Recent research using functional magnetic resonance imaging has shown that dog experts perceive socially relevant behavioural cues in dogs differently to non-experts, but similarly to how humans interpret social cues from conspecifics (Kujala et al., 2012). Observer experience as a potential source of bias in questionnaire-based personality assessments needs to be further explored before these tools can be confidently used to investigate personality structure in dogs.

17

5. Conclusion The relationships between factors from the three questionnaires used in this study indicate that canine personality may be usefully described using a hierarchical model; this model includes at least six broad personality traits in the upper levels and a number of context-specific behavioural traits below this. At least three of these broad trait clusters, MCPQ-R ‘Neuroticism’, MCPQ-R ‘Extraversion’ and DIAS ‘Behavioural Regulation’, show similarities to traits identified in other species. However, the relationship between sociability, emotional stability and aggression remains unclear due to methodological limitations. A small number of factors were also identified which appear to be highly context-specific and not related to higher-order traits; these may be difficult to assess in batterystyle tests performed in limited contexts, such as those conducted in shelters. Notable differences in the relationships between some factors based on the degree of dog-related work experience of owners were also found, indicating that these factors are not reliable when dog-related work experience varies between observers. This could prove problematic for applications in which behavioural information is sought from owners and owner experience is not controlled for, as is common in studies recruiting volunteers with internet-based methods. This study highlights the need to control for observer experience with dogs when investigating canine personality through questionnaires, and provides some support for the current consensus that canine personality may be usefully described using a hierarchical structure of traits, some of which relate to behavioural traits which could be measured in a battery test environment.

18

References Ashton, M.C., Lee, K., 2005. The lexical approach to the study of personality structure: toward the identification of cross-culturally replicable dimensions of personality variation. J. Pers. Disord. 19, 303-308. Asp, H.E., Fikse, W.F., Nilsson, K., Strandberg, E., 2015. Breed differences in everyday behaviour of dogs. Appl. Anim. Behav. Sci. http://dx.doi.org/10.1016/j.applanim.2015.04.010 Blackwell, E.J., Bradshaw, J.W.S., Casey, R.A., 2013. Fear responses to noises in domestic dogs: Prevalence, risk factors and co-occurrence with other fear related behaviour. Appl. Anim. Behav. Sci. 145, 15-25. http://dx.doi.org/10.1016/j.applanim.2012.12.004 Duffy, D.L., Hsu, Y., Serpell, J.A., 2008. Breed differences in canine aggression. Appl. Anim. Behav. Sci. 114, 441-460. http://dx.doi.org/10.1016/j.applanim.2008.04.006 Duffy, D.L., Kruger, K.A., Serpell, J.A., 2014. Evaluation of a behavioral assessment tool for dogs relinquished to shelters. Prev. Vet. Med. 117, 601-609. doi: http://dx.doi.org/10.1016/j.prevetmed.2014.10.003 Duffy, D.L., Serpell, J.A., 2012. Predictive validity of a method for evaluating temperament in young guide and service dogs. Appl. Anim. Behav. Sci. 138, 99-109. http://dx.doi.org/10.1016/j.applanim.2012.02.011 Eysenck, S.B.G., Eysenck, H.J., 1978. Impulsiveness and venturesomeness: Their position in a dimensional system of personality description. Psychol. Rep. 43, 1247-1255. Fratkin, J.L., Sinn, D.L., Patall, E.A., Gosling, S.D., 2013. Personality consistency in dogs: A metaanalysis. PLoS ONE 8, 1-19. http://dx.doi.org/10.1371/journal.pone.0054907 Fratkin, J.L., Sinn, D.L., Thomas, S., Hilliard, S., Olson, Z., Gosling, S.D., 2015. Do you see what I see? Can non-experts with minimal training reproduce expert ratings in behavioral assessments of working dogs? Behav. Processes 110, 105-116. http://dx.doi.org/10.1016/j.beproc.2014.09.028 Gosling, S.D., 1999. Personality dimensions in animals. Dissertation Abstracts International: Section B: The Sciences and Engineering 60, 1341. 19

Gosling, S. D. 2008. Personality in non-human animals. Social and Personality Psychology Compass, 2(2), 985-1001. http://dx.doi.org/10.1111/j.1751-9004.2008.00087.x Gosling, S.D., John, O.P., 1999. Personality dimensions in nonhuman animals: A cross-species review. Curr. Dir. Psychol. Sci. 8, 69-75. http://dx.doi.org/10.1111/1467-8721.00017 Haverbeke, A., De Smet, A., Depiereux, E., Giffroy, J.-M., Diederich, C., 2009. Assessing undesired aggression in military working dogs. Appl. Anim. Behav. Sci.117, 55-62. http://dx.doi.org/10.1016/j.applanim.2008.12.002 Hejjas, K., Kubinyi, E., Ronai, Z., Szekely, A., Vas, J., Miklosi, A., Sasvari-Szekely, M., Kereszturi, E. (2009). Molecular and behavioral analysis of the intron 2 repeat polymorphism in the canine dopamine D4 receptor gene. Genes, Brain & Behavior, 8(3), 330-336. http://dx.doi.org/10.1111/j.1601-183X.2008.00475.x Hsu, Y., Serpell, J.A., 2003. Development and validation of a questionnaire for measuring behaviour and temperament traits in pet dogs. JAVMA. 223, 1293-1300. http://dx.doi.org/10.2460/javma.2003.223.1293 Jackson, J.J., Hill, P.L., Payne, B.R., Roberts, B.W., Stine-Morrow, E.A., 2012. Can an old dog learn (and want to experience) new tricks? Cognitive training increases openness to experience in older adults. Psychol. Aging 27, 286-292. http://dx.doi.org/10.1037/a0025918 Jones, A. C., 2008. Development and validation of a dog personality questionnaire. (Doctor of Philosophy), University of Texas, Austin, Texas. Jones, A.C., Gosling, S.D., 2005. Temperament and personality in dogs (Canis familiaris): A review and evaluation of past research. Appl. Anim. Behav. Sci. 95, 1-53. http://dx.doi.org/10.1016/j.applanim.2005.04.008 King, T., Marston, L.C., Bennett, P.C., 2009. Describing the ideal Australian companion dog. Appl. Anim. Behav. Sci. 120, 84-93. http://dx.doi.org/10.1016/j.applanim.2009.04.011

20

Koolhaas, J., De Boer, S., Buwalda, B., Van Reenen, K., 2007. Individual variation in coping with stress: A multidimensional approach of ultimate and proximate mechanisms. Brain Behav. Evol. 70, 218. http://dx.doi.org/10.1159/000105485 Kujala, M. V., Kujala, J., Carlson, S., Hari, R., 2012. Dog experts’ brains distinguish socially relevant body postures similarly in dogs and humans. PLoS ONE, 7(6). http://dx.doi.org/10.1371/journal.pone.0039145.g001 Ley, J., Bennett, P., Coleman, G., 2008. Personality dimensions that emerge in companion canines. Appl. Anim. Behav. Sci. 110, 305-317. http://dx.doi.org/10.1016/j.applanim.2007.04.016 Ley, J.M., Bennett, P.C., Coleman, G.J., 2009a. A refinement and validation of the Monash Canine Personality Questionnaire (MCPQ). Appl. Anim. Behav. Sci. 116, 220-227. http://dx.doi.org/10.1016/j.applanim.2008.09.009 Ley, J.M., McGreevy, P., Bennett, P.C., 2009b. Inter-rater and test–retest reliability of the Monash Canine Personality Questionnaire-Revised (MCPQ-R). Appl. Anim. Behav. Sci. 119, 85-90. http://dx.doi.org/10.1016/j.applanim.2009.02.027 McCrae, R.R., Costa, P.T., Del Pilar, G.H., Rolland, J.-P., Parker, W.D., 1998. Cross-cultural assessment of the Five-Factor Model the revised NEO Personality Inventory. J. Cross Cult. Psychol. 29, 171188. McCrae, R.R., John, O.P., 1992. An introduction to the Five‐Factor model and its applications. J. Pers. 60, 175-215. Mirko, E., Kubinyi, E., Gacsi, M., Miklosi, A., 2012. Preliminary analysis of an adjective-based dog personality questionnaire developed to measure some aspects of personality in the domestic dog (Canis familiaris). Appl. Anim. Behav. Sci. 138, 88-98. http://dx.doi.org/10.1016/j.applanim.2012.02.016 Mirko, E., Doka, A., & Miklosi, A. (2013). Association between subjective rating and behaviour coding and the role of experience in making video assessments on the personality of the domestic dog

21

(Canis familiaris). Applied Animal Behaviour Science Oct(Pagination), No Pagination Specified. doi:http://dx.doi.org/10.1016/j.applanim.2013.10.003 Myers, L. and Sirois, M. J. 2006. Spearman Correlation Coefficients, Differences between. Encyclopedia of Statistical Sciences. 12. Nagasawa, M., Tsujimura, A., Tateishi, K., Mogi, K., Ohta, M., Serpell, J.A., Kikusui, T., 2011. Assessment of the factorial structures of the C-BARQ in japan. The Journal of Veterinary Medical Science 73, 869-875. Rayment, D.J., De Groef, B., Peters, R.A., Marston, L.C., 2015. Applied personality assessment in domestic dogs: Limitations and caveats. Appl. Anim. Behav. Sci. 163, 1-18. http://dx.doi.org/10.1016/j.applanim.2014.11.020 Siracusa, C., Di Nardo, A., Serpell, J., Iannetti, L., Podaliri, M., Trentini, R., Villa, P.D., 2010. A comparison between a direct and an indirect method for assessing aggressiveness in dogs. J. Vet. Behav.: Clin. Appl. Res. 5, 55. http://dx.doi.org/10.1016/j.jveb.2009.09.016 Svartberg, K., 2005. A comparison of behaviour in test and in everyday life: evidence of three consistent boldness-related personality traits in dogs. Appl. Anim. Behav. Sci. 91, 103-128. http://dx.doi.org/10.1016/j.applanim.2004.08.030 Svartberg, K., Forkman, B., 2002. Personality traits in the domestic dog (Canis familiaris). Appl. Anim. Behav. Sci. 79, 133-155. http://dx.doi.org/10.1016/s0168-1591(02)00121-1 Tami, G., Gallagher, A., 2009. Description of the behaviour of domestic dog (Canis familiaris) by experienced and inexperienced people. Appl. Anim. Behav. Sci.120, 159-169. http://dx.doi.org/10.1016/j.applanim.2009.06.009 Tamimi, N., Jamshidi, S., Serpell, J.A., Mousavi, S., Ghasempourabadi, Z., 2015. Assessment of the CBARQ for evaluating dog behavior in Iran. J. Vet. Behav.: Clin. Appl. Res. 10, 36-40. http://dx.doi.org/10.1016/j.jveb.2014.09.064

22

Taylor, K.D., Mills, D.S., 2006. The development and assessment of temperament tests for adult companion dogs. J. Vet. Behav.: Clin. Appl. Res. 1, 94-108. http://dx.doi.org/10.1016/j.jveb.2006.09.002 Temesi, A., Turcson, B.l., Miklosi, d.m., 2014. Measuring fear in dogs by questionnaires: An exploratory study toward a standardized inventory. Appl. Anim. Behav. Sci. 161, 121-130. http://dx.doi.org/10.1016/j.applanim.2014.09.009 Uher, J., 2011. Individual behavioral phenotypes: an integrative meta-theoretical framework. Why "behavioral syndromes" are not analogs of "personality". Dev. Psychobiol. 53, 521-548. http://dx.doi.org/10.1002/dev.20544 Wan, M., Bolger, N., Champagne, F.A., 2012. Human perception of fear in dogs varies according to experience with dogs. PLoS ONE 7, 1-10. http://dx.doi.org/10.1371/journal.pone.0051775.t001 Wan, M., Hejjas, K., Ronai, Z., Elek, Z., Sasvari-Szekely, M., Champagne, F. A., Miklosi, A., Kubinyi, E., 2013. DRD4 and TH gene polymorphisms are associated with activity, impulsivity and inattention in Siberian Husky dogs. Animal Genetics, 44(6), 717-727. http://dx.doi.org/10.1111/age.12058 Wright, H.F., Mills, D., Pollux, P.M.J., 2011. Development and validation of a psychometric tool for assessing impulsivity in the domestic dog (Canis familiaris). Int. J. Comp. Psychol. 24, 210-225. Wright, H.F., Mills, D.S., Pollux, P.M.J., 2012. Behavioural and physiological correlates of impulsivity in the domestic dog (Canis familiaris). Physiol. Behav. 105, 676-682. http://dx.doi.org/10.1016/j.physbeh.2011.09.019

23

Tables Table 1. Published factors and factors indicated by principal component analysis (PCA) for the Dog Impulsivity Assessment Scale (DIAS), the Canine Behavioural Assessment and Research Questionnaire (CBARQ), and the Monash Canine Personality Questionnaire-Revised (MCPQ-R). Numbers in parentheses indicate how many questions load onto the relevant factor. Questionnaire Published factors DIAS

CBARQ

MCPQ-R

1. 2. 3. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

Behavioural Regulation (10) Response to Novelty/Aggression (5) Responsiveness (5) Trainability (8) Stranger-Directed Aggression (10) Owner-Directed Aggression (8) Dog-Directed Aggression (4)* Dog Rivalry (4) Stranger-Directed Fear (4) Non-Social Fear (6) Dog-Directed Fear (4)* Touch Sensitivity (4) Separation-Related Problems (8) Excitability (6) Attachment/Attention Seeking (6) Chasing (4) Energy (2) (Not scored)

1. 2. 3. 4. 5.

Extraversion (6) Motivation (5) Training Focus (6) Amicability (5) Neuroticism (4)

Factors from our PCA 1. 2. 3. 1. 2. 3. 4.

Behavioural Regulation (8) Response to Novelty/Aggression (2) Responsiveness (2) Trainability (7) Stranger-Directed Aggression (9) Owner-Directed Aggression (4) Dog-Directed Aggression AND DogDirected Fear (8)* a 5. Dog Rivalry (4) a 6. Stranger-Directed Fear (8) 7. Non-Social Fear (6) a 8. Touch Sensitivity (5) 9. Separation-Related Problems (8) a 10. Excitability AND Energy (8) 11. Attachment/Attention Seeking (6) a 12. Chasing (4) a 13. Owner-Directed Aggression When Punished (2) 1. Extraversion (6) a 2. Motivation (4) 3. Training Focus (6) a 4. Amicability (5) a 5. Neuroticism (4) a

* Note: CBARQ ‘Dog-Directed Aggression’ and ‘Dog-Directed Fear’ can be calculated as both separate and combined factors according to the published model. a

Superscript letter indicates that published and calculated factors were identical.

24

Table 2. Spearman’s rho correlation analysis of questionnaire factors for the Dog Impulsivity Assessment Scale (DIAS), the Canine Behavioural Assessment and Research Questionnaire (CBARQ) and the Monash Canine Personality Questionnaire-Revised (MCPQ-R).

CBARQ SeparationRelated Problems CBARQ Excitability CBARQ Attachment / Attention Seeking CBARQ Chasing MCPQ-R Extraversio n MCPQ-R Motivation MCPQ-R Training Focus MCPQ-R Amicability MCPQ-R Neuroticis m

MCPQ-R Amicability

CBARQ Touch Sensitivity

MCPQ-R Training Focus

CBARQ DogDirected Fear

MCPQ-R Motivation

CBARQ Non-social Fear

MCPQ-R Extraversion

CBARQ StrangerDirected Fear

CBARQ Chasing CBARQ Attachment/ Attention Seeking CBARQ Excitability CBARQ SeparationRelated Problems CBARQ Touch Sensitivity

CBARQ Dog Rivalry

CBARQ DogDirected Fear

CBARQ StrangerDirected Aggression CBARQ OwnerDirected Aggression CBARQ DogDirected Aggression

CBARQ Nonsocial Fear

CBARQ Trainability

CBARQ Strangerdirected Fear

DIAS Responsiv eness

CBARQ Dog Rivalry

CBARQ DogDirected Aggression CBARQ OwnerDirected CBARQ Aggression StrangerDirected Aggression CBARQ Trainability DIAS Responsiveness DIAS Response To Novelty DIAS Behavioural Regulation DIAS Response To Novelty

0.24 9** 0.01 0.26 2 7** 0.01 0.24 0.07 8 9** 0 0.13 0.32 0.05 0.08 8* 2** 6 5 0.17 0.18 0.03 0.10 0.06 ** ** 2 3 2 6 6 0.20 0.31 0.02 0.03 0.36 0.11 2** 7** 7 8 8** 3* 0.14 0.19 0.02 * ** 8 7 1 0.09 0.48 0.07 ** 8 8 4 0.18 0.31 0.06 ** ** 2 4 5 0.23 0.36 0.01 ** ** 4 4 6 0.17 0.24 0.04 ** ** 3 2 3

0.07 0.21 0.29 0.42 6 9** 8** 4** 0.42 0.08 0.09 0.09 0.03 5** 7 4 5 5 0.26 0.08 0.17 0.21 0.40 0.06 0** 2 8** 8** 2** 2 0.01 0.19 0.11 0.42 0.27 0.39 0.38 3 6** 5* 7** 4** 2** 4**

0.18 0.18 0.20 0.21 0.21 0.28 0.30 0.08 4** 6** 4** 3** 6** 6** 8** 8 0.34 0.15 0.04 0.17 0.13 0.14 0.16 0.16 0.25 0.20 0.18 0.06 ** * 3 0 8 4** 9** 7* 8** 0** 3** 2** 8** 5 0.48 0.03 0.22 0.13 0.21 0.09 0.17 0.06 0.08 0.13 0.16 0.33 0.14* 6** 7 5** 5* 9** 5 6** 7 1 7* 5** 2** 0.29 0.05 0.13 0.12 0.20 0.04 0.16 0.11 0.09 0.13 0.11 0.12 0.36 0.35 0** 3 2* 5* 5** 9 5** 8 0 4* 8* 7* 4** 4** 0.21 0.10 7** 8 0.33 0.05 7** 4 0.10 0.08 6 3 0.43 0.16 2** 3** 0.33 0.59 2** 8**

0.05 0.08 0.23 0.09 0.23 0.17 0.10 0.12 0.15 0.14 2 0 0** 4 1** 8** 2 3* 5** 6* 0.29 0.27 0.13 0.06 0.12 0.07 0.00 0.09 0.07 0.00 4** 8** 1* 5 2* 1 1 9 8 4 0.22 0.18 0.06 0.09 0.08 0.09 0.10 0.05 0.04 0.03 0** 2** 9 2 7 9 9 0 9 8 0.32 0.29 0.01 0.05 0.07 0.05 0.05 0.08 0.11 0.16 3** 8** 9 8 5 5 0 2 1* 9** 0.14 0.06 0.40 0.13 0.38 0.23 0.49 0.23 0.33 0.25 2* 5 2** 7* 5** 9** 2** 5** 1** 7** 0.10 0.40 0.22 0.00 0.03 0.10 0.54 0.50 0.45 0.25 0.04 0.10 7 0** 6** 3 3 7 8** 4** 5** 1** 2 3

0.11 0.18 0.07 3* 1** 3 0.21 0.45 0.26 0.19 7** 7** 8** 6** 0.00 0.18 0.09 0.06 0.39 8 7** 8 7 4** 0.00 0.13 0.24 0.17 0.14 0.10 0 8* 7** 1** 8* 4 0.08 0.26 0.23 0.12 0.09 0.19 0.11 1 5** ** * ** * 8 2 8 9 9 0.17 0.04 0.13 0.08 0.01 0.24 0.06 0.32 8** 2 7* 7 8 8** 0 6**

Blue shading indicates moderate correlation between factors, while dark green shading indicates strong correlation between factors. Level of significance: * P <0.01; ** P <0.001 (2-tailed)

25

Table 3. Between-factor Spearman’s rho correlations for Dog Impulsivity Assessment Scale (DIAS), Canine Behavioural and Research Questionnaire (CBARQ) and Monash Canine Personality Questionnaire (MCPQ) factors, from samples of owners with varying amounts of dog-related experience.

Factor 1

Factor 2

Whole sample (n=544 )

Formal work experienc e (n=259)

No formal work experienc e (n=284)

Dog trainer (n=12 2)

Veterinar y worker (n=54)

Overall varianc e

CBARQ Dog Rivalry

CBARQ StrangerDirected Aggressio n

0.219** *

0.123

0.311***

-0.005

0.08

0.316a

DIAS Behavioural Regulation

MCPQ Motivation

0.106

0.207***

-0.003

0.13

0.372**

0.375 a

DIAS Response To Novelty

DIAS Behaviour al Regulation

0.249** *

0.118

0.364***

0.139

-0.035

0.399 a

CBARQ OwnerDirected Aggression

DIAS Response To Novelty

0.183** *

0.105

0.251***

0.044

-0.152

0.403 a

CBARQ Nonsocial Fear

CBARQ Dog Rivalry

0.218** *

0.171

0.262***

0.296* *

-0.132

0.428 a

DIAS Responsivene ss

CBARQ DogDirected Aggressio n

0.027

0.209***

-0.126

0.178

0.309*

0.435 a

CBARQ DogDirected Fear

MCPQ Neuroticis m

0.455** *

0.345***

0.533***

0.265* *

0.084

0.449 a

CBARQ OwnerDirected Aggression

MCPQ Extraversi on

0.065

0.137

-0.002

0.208*

0.456***

0.458 a

Level of significance of correlation: * P <0.05; ** P <0.01; *** P <0.001. a

Indicates significance of difference between correlations (Fishers Z transformation, P<0.01)

26