Assessing pig body language: agreement and consistency between pig farmers, veterinarians, and animal activists

This study investigates the interobserver and intraobserver reliability of qualitative behavior assessments (QBA) of individual pigs by 3 observer groups selected for their diverging backgrounds, experience, and views of pigs. Qualitative behavior assessment is a "whole animal" assessment approach that characterizes the demeanor of an animal as an expressive body language, using descriptors such as relaxed, anxious, or content. This paper addresses the concern that use of such descriptors in animal science may be prone to distortion by observer-related bias. Using a free-choice profiling methodology, 12 pig farmers, 10 large animal veterinarians, and 10 animal protectionists were instructed to describe and score the behavioral expressions of 10 individual pigs (sus scrofa) in 2 repeat sets of 10 video clips, showing these pigs in interaction with a human female. They were also asked to fill in a questionnaire gauging their experiences with and views on pigs. Pig scores were analyzed with generalized procrustes analysis and effect of treatment on these scores with ANOVA. Questionnaire scores were analyzed with a χ(2) test or ANOVA. Observers achieved consensus both within and among observer groups (P < 0.001), identifying 2 main dimensions of pig expression (dim1: playful/confident-cautious/timid; dim2: aggressive/nervous-relaxed/bored), on which pig scores for different observer groups were highly correlated (pearson r > 0.90). The 3 groups also repeated their assessments of individual pigs with high precision (r > 0.85). Animal protectionists used a wider quantitative range in scoring individual pigs on dimension 2 than the other groups (P < 0.001); however, this difference did not distort the strong overall consistency of characterizations by observers of individual pigs. Questionnaire results indicated observer groups to differ in various ways, such as daily and lifetime contact with pigs (P < 0.001), some aspects of affection and empathy for pigs (P < 0.05), and confidence in the validity of personal QBA descriptors (P < 0.02). The main finding of this study is that despite such differences in background and outlook, the 3 observer groups showed high interobserver and intraobserver reliability in their characterizations of pig body language. This supports the empirical nature of QBA in context of the wider anthropomorphism debate.