文章基本信息

标题：多面評価法 (360度フィードバック法) に関する多特性多評価者行列分析
其他标题：Multitrait Multirater Matrix Analysis of 360 Degree Feedback Programs
本地全文：下载
作者：高橋潔
期刊名称：経営行動科学
印刷版ISSN：0914-5206
电子版ISSN：1884-6432
出版年度：2001
卷号：14
期号：2
页码：67-85
DOI：10.5651/jaas.14.67
语种：Japanese
出版社：The Japanese Association of Administrative Science
摘要：This study examines convergent and discriminant validity of the 360 degree feedback systems in which managerial behavior and performance are evaluated by multiple raters including bosses, peers, subordinates, and self. Multitraitmultirater (MTMR) correlation matrix was constructed from the data of 177 Japanese managers and 1, 187 coworkers. The matrix was investigated by zero-order correlations and confirmatory factor analyses (CFA). Results of correlation analyses found that 1) in any domain of managerial behavior (traits), ratings within the same rater were correlated highly, proving that discriminant validity was low; and that 2) ratings of the same trait provided by different raters were not associated highly, suggesting that convergent validity is questioned. Confirmatory factor analyses found that among three models, i.e., rater model, trait model, and rater-trait combined model, the combined model fit best, suggesting that both rater effect and trait effect were interwoven in the actual ratings of the 360 degree feedback.
其他摘要：This study examines convergent and discriminant validity of the 360 degree feedback systems in which managerial behavior and performance are evaluated by multiple raters including bosses, peers, subordinates, and self. Multitraitmultirater (MTMR) correlation matrix was constructed from the data of 177 Japanese managers and 1, 187 coworkers. The matrix was investigated by zero-order correlations and confirmatory factor analyses (CFA). Results of correlation analyses found that 1) in any domain of managerial behavior (traits), ratings within the same rater were correlated highly, proving that discriminant validity was low; and that 2) ratings of the same trait provided by different raters were not associated highly, suggesting that convergent validity is questioned. Confirmatory factor analyses found that among three models, i.e., rater model, trait model, and rater-trait combined model, the combined model fit best, suggesting that both rater effect and trait effect were interwoven in the actual ratings of the 360 degree feedback.