ISSN 0439-755X
CN 11-1911/B
主办:中国心理学会
   中国科学院心理研究所
出版:科学出版社

心理学报 ›› 2013, Vol. 45 ›› Issue (9): 1039-1049.doi: 10.3724/SP.J.1041.2013.01039

• 论文 • 上一篇    下一篇

无领导小组讨论的多侧面Rasch模型应用

姚若松;赵葆楠;刘泽;苗群鹰   

  1. (1广州大学教育学院, 广州 510006) (2广州大学外国语学院, 广州 510006)
  • 收稿日期:2012-12-27 发布日期:2013-09-25 出版日期:2013-09-25
  • 通讯作者: 姚若松
  • 基金资助:

    广东省哲学社会科学“十一五”规划项目(GD10CGL08)和广州市哲学社会科学发展“十二五”规划项目(13G59)资助。

The Application of Many-Facet Rasch Model in Leaderless Group Discussion

YAO Ruosong;ZHAO Baonan;LIU Ze;MIAO Qunying   

  1. (1Department of Education, Guangzhou University, Guangzhou 510006, China) (2 School of Foreign Studies, Guangzhou University, Guangzhou 510006, China)
  • Received:2012-12-27 Online:2013-09-25 Published:2013-09-25
  • Contact: YAO Ruosong

摘要: 采用项目反应理论(IRT)的多侧面Rasch模型(MFRM), 分析评价中心技术中无领导小组讨论(LGD)的测评结果, 探讨被试能力水平、评委评分宽严度、评分内部一致性、维度难度和评定等级等问题, 进而讨论各种偏差。通过MFRM分析人事测评结果, 可深入了解被试能力的真实差异、甑别维度难度、探查测评误差源, 从而完善测评试题编制、评估或诊断评委合格性、提高测评维度与测评目的匹配性, 为拓展项目反应理论在人事测评中的应用提供独特视角。

关键词: 无领导小组讨论, 多侧面Rasch模型, 项目反应理论, 人事测评

Abstract: Many-Facet Rasch model (MFRM) of Item Response Theory (IRT) is applied to performance assessment. Domestic and foreign researches applied MFRM in many fields such as analysis of various examinations, medical diagnosis, judgments of life quality and so on. In these assessment tests, ratings were influenced by a variety of factors among which judges played the most important part. This thesis mainly probed into issues covering subjects, judges, rating scales and rating deviation in Leaderless Group Discussion (LGD) of personnel assessment center in personnel assessment to improve the effectiveness and stability of assessment. This study adopted the FACETS software, a MFRM computer statistics program, to establish 3 facets of subjects, judges and rating dimensions to analyze subjects’ abilities, rater severity, inter-rater reliability, dimension difficulty and rating scales. Meanwhile, this study got results of deviation analysis of subjects and judges, judges and dimensions, deviation among judges, subjects and dimensions. The results illustrated significant differences existed among levels of subjects’ ability, rater severity, dimension difficulty and the rating scale. Differences of rater severity generally did not affect the test scores of subjects. Except some judges, other judges’ ratings had good internal consistency. Dimension difficulty could better distinguish subjects’ ability but judges tended to concentrate on using an intermediate rating scale; The results of deviation analysis of judges and subjects, judges and dimension showed that untrained judges E, F had more rating deviations, so it was necessary to monitor their scores and strengthen the training of the two judges. The application of MFRM, IRT’s expansion, to assessment center evaluation could enable evaluators to make the employment decision by estimated ability level of subjects, design tests according to dimension difficulty, set the standards for training and selection referring to examine judges’ ratings rater severity and inter-rater reliability, improve the assessment process based on a variety of deviation analysis, and finally promote scientific, standardized and precise development of evaluation system of assessment center.

Key words: leaderless group discussion, many-facet Rasch model, item response theory, personnel assessment