摘要
本文运用多元概化理论的方法分析由6名评分者和77名考生组成的无领导小组讨论测评的误差控制问题。结果表明:测验中5个测评维度的概化系数为0.89-0.93之间,说明总体上6名评价者对被试评分较为一致;增加评价者数量和对评分者进行相关培训均可有效减少测量误差;从测评成本和控制误差两方面考虑,评价者人数控制在3-6人较为合适。概化理论在无领导小组测评中的应用能有效分析误差来源,改进人事测评的设计和实施。
This paper has made data analysis of leaderless group discussion with 6 Raters and 77 candidates.The results reveal as follows:Generalizability coefficient for each dimension is above 0.89.On the whole,the six raters have rated the candidates rather consistently.Considering evaluation costs and Control error,the number of raters controlled in 3 to 6 is more appropriate.Multivariate Generalizability Theory can also handle testing of multiple dimensions more scientifically and effectively.
出处
《中国人力资源开发》
北大核心
2013年第13期57-61,共5页
Human Resources Development of China
基金
2013年广州市哲学社会科学发展"十二五"规划课题(13G59)资助
关键词
概化理论
无领导小组讨论
误差控制
Generalizability Theory
Leaderless Group Discussion
Error Control
