A welfare assessment system should be "high" in validity, robustness and feasibility - the latter both as regards time and costs. Therefore, observers must be able to perform the on-farm assessment with acceptable validity after some training. Based on empiric data this paper evaluates the consequences of operating with several observers. Animal based measures on 9 Danish mink farms were taken in November 2011. Eight observers individually, but in paris on herd level, carried out data collection on the measures involving subjective grading, e.g. mink "activity", "injuries" and "fur-chewing" on approximately 120 cages with mink per farm. The assessment of the two observers gave similar frequencies of welfare problems and thus similar welfare assessments. The individual problems observed were however, not the same leading to poor or fair, but rarely good inter observer reliability. Despite the skilled assessors, the short training was not sufficient to get highly reliable results. No overall difference was found between the inter observer reliability of cages with ≤2 or ≥3 mink in a cage. More training and better training material and, for some measures, observation procedures are needed in order to increase reliability of the difficult subjective measures.
Proceedings of the Xth International Scientific Congress in Fur Animal Production, 2012, p. 469-476
Neovison vison; Kappa; index of concordance; welfare