基于GT和多面Rasch模型的结构化面试分析

该研究应用GT和多面Rasch模型对结构化面试数据进行分析，并提出一些建议针对某辅导员招聘面试数据，运用GT从宏观上分析应聘者、考官和项目所带来的总体误差大小，在此基础上，运用多面Rasch模型从微观上进一步探查考官严厉度、应聘者能力差异、项目难易度及侧面偏差．结果表明：1）GT分析表明应聘者产生的变异较大（90．65％），说明面试可靠性较高，且当考官数为2时可靠性已较好.2）多面Rasch模型分析出了各侧面效应中的非拟合因素及交互效应中的偏差因素，表明面试误差主要来自考官间严厉度的差异及其自身一致性的不稳定。将GT与多面Rasch模型相结合分析面试数据不仅能测查出评价过程各方面的问题因素，并能更好地作整体把握。

Based on Generalizability Theory and Many-facet Rasch Measurement To Analyze the Data of Structured Interview

Objective： Use GT and Many - facet Rasch measuement to analyze the structured interview data and give some suggestions. Methods： The data is from a school assistant interview. First, use GT to analyze the total variance of the interviewers, raters and items macroscopically. Second, use Many - facet Rasch measurement to analyze the bias of all facets microcosmicly. Results： （1） The Generalizability Theory analysis indicates the greatest percentage of the total variance（90.65% ）is attributable to persons. The dependability is quite respectable. And dependability is high when using the two raters. （2）Many- facet measurement analyzes the misfitting in all facets and the bias in the interactions.The results indicate that the error is mainly from the differences among raters and the inconsistency of themselves. Conclusion： Using this method not only detect the problem factors, but also analyze the whole performance rating better.