A Social Network Approach to Peer Assessment: Improving Predictive Validity
研究提出基于社会网络分析的新同伴评价指标（提名者-被提名者、未回应的提名），在249名士兵样本中验证其比传统提名数更能预测6个月后的工作表现。
Abstract Our premise is that the simple measure used in peer assessment (i.e., number of peer nominations) does not capture the complexity of social information processing and therefore has limited predictive validity. Based on indicators derived from social network analysis and social information processing theories, we suggest new measures (nominations‐by‐nominees and nominations‐not‐returned) to enhance the predictive validity of peer assessment. We then compare the validity of existing measures with ours, using a longitudinal sample of 249 soldiers, divided into 18 groups. The soldiers first assessed each other on friendly behavior and instrumental contribution to the team. More than six months later, the commanders of the 132 soldiers in the unit under review provided evaluations of their performance in regard to stress, engagement, and leadership. We found that our new, complex measures predicted performance above and beyond the traditional measure. Theoretical and applied implications are discussed.