Clark Corinna C A, Sibbald Nicola J, Rooney Nicola J
Department of Life Sciences, University of Warwick, Coventry, United Kingdom.
Defence Science and Technology Laboratory, Salisbury, United Kingdom.
Front Vet Sci. 2020 Sep 10;7:612. doi: 10.3389/fvets.2020.00612. eCollection 2020.
Self-assessments of performance are commonly used in the human workplace, although compared to peer or supervisor ratings, they may be subject to positive biases or leniency. The use of subjective ratings scales in animal sciences is also common, although little consideration is usually given to possible rater bias. Dog handlers, work very closely and form strong relationships with their dogs and are also best placed to monitor dog performance since they often work in isolation. Previous work found ratings of search dog performance correlated well between experienced dog trainers, instructors, and scientists; but until now, there has been no investigation into ratings made by a dog's own handler. We compared handlers' subjective assessment of their own dog's search performance to scores given by other handlers and in a second study, to scores made by impartial raters. We found that handlers generally showed leniency; for example scoring their own dogs more favorably for Control (responsiveness to commands) and Strength of Indication. But the degree of bias varied with the trait being scored and between raters. Such differences may be attributable to greater desirability or importance of favorable scores for certain traits, or a lack of clarity of their precise meaning. Handlers may vary in susceptibility to bias due to differing levels of experience and the extent to which they view their dog's ability as dependent on their own. The exact causes require further investigation. We suggest working dog agencies provide rater-training to overcome leniency, improve reliability and validity, and to increase handler's motivation to provide accurate assessments. This study represents one of a series of steps to formulate robust, validated and evidence-based performance rating systems and has relevance to any situation where raters assess their own performance or others (particularly where they may have a vested interest in, or loyalty toward, the ratee).
在人类工作场所中,绩效的自我评估很常见,尽管与同事或上级的评分相比,自我评估可能存在积极偏差或宽松倾向。在动物科学中,主观评分量表的使用也很普遍,不过通常很少考虑评分者可能存在的偏差。训犬员与他们的狗密切合作并建立了牢固的关系,而且由于他们经常单独工作,所以最有条件监测狗的表现。先前的研究发现,经验丰富的训犬师、教员和科学家对搜救犬表现的评分相关性良好;但到目前为止,还没有对狗的主人给出的评分进行过调查。我们将主人对自己狗的搜救表现的主观评估与其他主人给出的分数进行了比较,在第二项研究中,还与公正评分者给出的分数进行了比较。我们发现主人通常表现出宽松倾向;例如,在服从性(对命令的反应)和示警强度方面给自己的狗打出更有利的分数。但偏差程度因所评分的特征以及评分者而异。这种差异可能归因于某些特征的有利分数更具可取性或重要性,或者其确切含义不够清晰。由于经验水平不同以及他们认为狗的能力在多大程度上依赖于自己,主人在偏差易感性方面可能存在差异。确切原因需要进一步调查。我们建议工作犬机构提供评分者培训,以克服宽松倾向,提高可靠性和有效性,并增强主人提供准确评估的积极性。本研究是制定稳健、经过验证且基于证据的绩效评估系统的一系列步骤之一,适用于评分者评估自己或他人绩效的任何情况(特别是在他们可能对被评估者有既得利益或忠诚度的情况下)。