Behavioral traits detected in shelter dogs by a behavior evaluation

Abstract

Animal shelters and other organizations caring for and placing dogs regularly use behavioral test batteries to evaluate a dog's behavioral tendencies in order to make safe and appropriate placements. Behavioral tendencies, or behavioral traits, are driven in part by personality. Therefore, test batteries should be able identify patterns of behaviors which reflect elements of canine personality. This study investigated the ability of one behavioral test battery currently in use at a shelter by examining the results from 668 behavior evaluations of shelter dogs. The test was composed of 19 sub-tests, one of which was excluded because it did not use the same coding scheme as the others. The 18 sub-tests included in the analysis used the same coding scheme where 38 behaviors were coded dichotomously (observed/not observed). Scores reflecting the number of sub-tests in which each behavior was observed were calculated and subjected to a principal components analysis with Varimax rotation. Analysis yielded a four factor solution which accounted for 45.3% of the total variance and included 26 of the behaviors. Three of the components, fearfulness (accounting for 12.3% of the variance), friendliness (11.7%), and aggressiveness (10.7%), were readily interpretable and consistent with other research. The fourth component, interest (10.5%), was unique and seemed to reflect a neutral or anticipatory state, perhaps related to the testing situation. Results provide evidence that this test battery does indeed detect elements of canine personality, which may enable identification of stable behavioral tendencies and so facilitate placement decisions.

Introduction

Each year, millions of unwanted dogs arrive in animal shelters after abandonment, abuse or relinquishment by their previous owners. In order to screen out dangerously aggressive dogs, identify the need for behavior modification, and successfully match dogs to new homes, shelter organizations first need to collect information about the dog's behavior. This information is used to identify stable behavioral tendencies in order to predict future behavior in similar situations (Ledger and Baxter, 1997, Svartberg, 2007, Taylor and Mills, 2006). Behavioral tendencies, or behavioral traits, reflect a consistent pattern of behaviors exhibited in similar situations and are driven, at least in part, by personality (Svartberg, 2007). Personality, a combination of genetic, cognitive, and environmental factors (Krueger and Johnson, 2008), may be defined in a very general way as the characteristics (i.e., traits) of an individual that describe and motivate consistent patterns of thoughts, feelings, and behaviors that persist across time and situations and that differentiate one individual from another (Pervin and John, 1997, Weinstein et al., 2008). In animals, we may infer personality through the observation of behavior (Svartberg, 2007). In dogs, personality may be measured in a variety of ways: questionnaire-based instruments completed by knowledgeable observers (usually owners) which may use adjective ratings similar to techniques used in human personality research (Gosling et al., 2003, Ley et al., 2008) or behavioral reports describing the actual behavior, frequency, intensity, etc. (Hsu and Serpell, 2003, Stephen and Ledger, 2007); and observational studies including those based on ethological techniques where observations are made in “natural” settings (De Palma et al., 2005) or behavioral tests which record a dog's reaction to standardized stimuli (Christensen et al., 2007, Hennessy et al., 2001, Jones and Gosling, 2005, Ledger and Baxter, 1997, Miklósi, 2007, Svartberg and Forkman, 2002, van der Borg et al., 1991, Svartberg, 2007). In shelters, the more common technique is the behavioral test, also called a behavioral evaluation or test battery (Jones and Gosling, 2005).

Test batteries examine and record a dog's reaction to specific stimuli presented through a series of standardized scenarios or sub-tests. The dog's reaction at the time of the sub-test is recorded in a standardized manner, typically using one of two methods: behavioral coding or behavioral rating scales (Miklósi, 2007, Svartberg, 2007, Weinstein et al., 2008). Behavioral coding focuses on individual behaviors such as their presence, frequency, duration or latency (Christensen et al., 2007, Hennessy et al., 2001, Ledger and Baxter, 1997, van der Borg et al., 1991). Noting if the dog approaches or sniffs a stranger or how many times the dog jumps up are examples of behavioral coding, the first (approaches or sniffs) being an example of a dichotomous data structure recording the presence or absence of the behavior, the second (number of jumps) being an example of a numeric data structure. Behavioral rating scales, in contrast, group behaviors along a pre-defined dimension (e.g., aggression) and then define each level of the scale in terms of escalating behaviors (Netto and Planta, 1997, Svartberg and Forkman, 2002). For example, Netto and Planta (1997) used a 5-point scale for aggressive behavior in dogs where 1 was no aggression, 2 was growling and/or barking, 3 was baring teeth, 4 was snapping, and 5 was biting and/or attacking with bite intention. Regardless of the coding method, test batteries are more frequently used in shelters than other methods because they can be conducted by trained staff unacquainted with the dog in a minimal amount of time and still cover a range of common situations. These are important advantages because any test battery used in a shelter must be feasible for practical use (Taylor and Mills, 2006); moreover, any test battery used in an applied setting should reveal something about canine personality.

In fact, previous research (Goddard and Beilharz, 1984, Hennessy et al., 2001, Svartberg and Forkman, 2002, Wilsson and Sundgren, 1997) has demonstrated that test batteries can indeed detect underlying personality structures, even if that was not the express intent of the study. Using statistical data reduction techniques called factor or principal components analysis to find correlated measures (such as behaviors or ratings) in the data set (Tabachnick and Fidell, 1989), researchers have found that test batteries can identify common canine personality traits such as aggressiveness, sociability, and fearfulness (Goddard and Beilharz, 1984, Hennessy et al., 2001, Svartberg and Forkman, 2002, Wilsson and Sundgren, 1997). Therefore, shelter test batteries, which seek to identify stable behavioral tendencies, should reveal elements of canine personality.

The goal of the present study was to examine the ability of one test battery using dichotomous behavioral coding to detect behavioral traits in shelter dogs. Traits which are consistent with other research will provide evidence that this test battery does reveal elements of canine personality and is likely to be a useful tool to help shelters evaluate and place dogs in their care.