国家公务员遴选策论的多面Rasch模型应用  

The Application of Many-facet Rasch Model in the Administrative Strategy Test

作　　者：李英武[1] 张海丽 胡心约 Li Yingwu;Zhang Haili;Hu Xinyue

机构地区：[1]中国人民大学心理学系,北京100872

出　　处：《上海行政学院学报》2021年第6期89-99,共11页The Journal of Shanghai Administration Institute

基　　金：国家社科基金一般项目“公务员分类录用面试的科学性研究”(14BZZ087);中国人民大学科学预研委托(团队基金)项目:重点项目培育类(19XNLG20)的阶段性成果

摘　　要：基于多面Rasch模型(MFRM),对某中央直属系统公务员遴选策论的主观评分偏差进行分析,探讨考生的能力水平,评分者宽严度、评分内部一致性、维度难度和评分量尺等因素对遴选考试决策的影响。结果发现:评分者的宽严度差异显著;评分者对特定评分维度的使用差异显著;评分者与评分维度间交互作用显著,不同评分者在特定评分指标上评分偏差明显。通过MFRM分析公务员遴选策论的测评结果,可深入了解考生的真实能力差异,策论评分维度的难度,并对公务员遴选测评中的主观评分偏差来源进行甄别,以完善国家公务员策论试题命制,建立评分者培训体系,提高公务员遴选考试决策的科学性,夯实考试测量学的理论与方法基础。Based on the Many-facet Rasch model(MFRM)of Item Response Theory(IRT),this study analyzed the evaluation results of Administrative Strategy Test(AST)of China Civil Servant Examination.To explore the source of rater bias,This study discussed the examinee’s ability level,rater severity,inter-rater reliability,dimension difficulty and rating scales.The results showed that the rater’s leniency is significantly different;the differences between the different rater evaluation criteria are significant,and the ability of different candidates is examined separately;there is interaction between rater and AST rating dimensions,and specific raters will exhibit deviated rating scores in specific dimensions.The application of MFRM helped to deeply understand the real differences of examinees’abilities and track the sources of scoring deviation,so as to improve the civil servant strategy test system,to establish the selection and training system of evaluators,to improve the scientificity of civil servant’s employment decision,and finally,to provide a measurement perspective for expanding the application of item response theory in large-scale personnel selection.

关 键 词：公务员遴选 策论 评分者偏差 多面RASCH模型 

分 类 号：D630.3[政治法律—政治学]