15 Evaluating Repeatability and Scorer Differences When Assigning Feet and Legs Scores in Simmental Heifers

Abstract

Abstract The objective of this paper was to determine the relationship and repeatability of between-evaluator scores for claw set (CS), hoof angle (HA), and rear leg side view (RSV). Repeatability of feet and legs scoring has implications on scorer bias and subsequently on the estimation of genetic parameters, expected progeny differences (EPD), and accuracy. Data were collected on 112 13-15-month-old Simmental heifers by 4 evaluators, where scorer 1 (S1) was a trained and experienced evaluator and subsequently trained scorers 2 (S2), 3 (S3), and 4 (S4) for approximately 20 minutes before all 4 evaluators sequentially scored every female at the same time. Training was conducted using feet and leg rubrics provided by the American Simmental Association (ASA) on a 1-9 scale, with 5 being optimum for each trait. Scorers were provided an online evaluation instrument, requiring responses for each evaluator for each animal, and included identification, CS, HA, and RSV. Pearson correlations were computed pairwise, with each new scorer compared to the trained scorer. For partial linear regression, the three new evaluators were fit individually to S1. Pearson correlation values ranged from 0.088 to 0.397. The correlation between S1 and S3 was the strongest for all traits. The weakest correlation of 0.088 was in rear leg side view between S1 and S2. Partial linear regression adjusted R-squared values ranged from 0.00 to 0.15. The correlation between S1 and S3 was the strongest for all traits. The weakest correlation of 0.00 was in rear leg side view between S1 and S2. The correlations between S1 and S2 was significant for CS and HA (P = 0.002, P = 0.039), but not for RSV (P = 0.355). The correlation between S1 and S3 was significant for CS, HA, and RSV [P = 1.5x10^(-5), P = 0.006, P = 0.02]. The correlation between S1 and S4 was not significant for CS, HA, and RSV (P = 0.276, P = 0.085, P = 0.223). Overall, the scores had weak correlations, which illustrates the differences in variability across scorers. The weak correlation also illustrates the need for more robust and intensive training of evaluators. Overall, while the scores are not highly correlated between evaluators, scorer bias should not affect estimation of genetic parameters long as each contemporary group is scored by one evaluator.