Development and rater reliability of COPE: A behavioral assessment tool to monitor shelter dog coping behavior

Abstract

n shelter settings, dogs exhibit a range of stress-related behaviors and coping styles such as anxious-avoidance (e.g., despondency) or excessive-arousal (e.g., lack of impulse control). The novel COPE (Canine Observed Provision of Enrichment) assessment was designed to identify coping styles and communicate welfare needs. COPE is comprised of eight human-animal interactions (HAI) that simulate activities and stimuli out of kennel. HAIs include kennel presentation, play, settling with treat, acoustic and visual startle response, conspecific behavior, returning to kennel, and additional behavioral observations. For each HAI-associated score, coping is quantified numerically such that zero indicates adaptive coping, negative numbers anxious-avoidance, and positive excessive-arousal. COPE was piloted with 91 dogs total with n = 72 assessed in real-time by two raters to analyze inter-rater reliability. Assessments were recorded, and a subset (n = 20) was randomly selected for intra- rater reliability. Agreement metrics included percent agreement (Po), proportion positive and negative agree- ment (Ppos, Pneg), and Cohen’s kappa (k); scores were further assessed using quadratic weighted kappa (QWK). Scores associated with the visual and acoustic startle tests were identified for removal due to low agreement (Po < 0.5), poor reliability (QWK ≤ 0.4), and lack of enrichment provision. All other scores had adequate agreement (Po > 0.6), of which approximately half had acceptable reliability (QWK ≥ 0.41). The remaining were identified as candidates for refinement (QWK ≤ 0.4). After refinement informed by subsequent criterion validity analysis (forthcoming manuscript), analysis will demonstrate that COPE can be reliably used to assess shelter dogs’ current coping state.