事业单位新进人员通用能力考试最优简版探究  

An exploratory study on the optimal short form of institution General Abilities Test for new staffs

作　　者：陈海平[1,2] 姜亚丽[1,2] 车宏生[1,2] 

机构地区：[1]北京师范大学心理学院 [2]北京师范大学心理测量与人力资源研究所,北京100875

出　　处：《北京师范大学学报（自然科学版）》2015年第3期320-325,共6页Journal of Beijing Normal University(Natural Science)

基　　金：国家教育部人文社会科学研究规划基金资助项目(10YJA40003)

摘　　要：利用《北京市新进人员通用能力考试》专业技术岗位1 335名考生成绩,在主成分分析确定该考试仅含一个公因子的情况下,根据因子载荷值的大小,分测验数目从2~9依次增加分别组成8个简版测验.在给定简版考试成绩与全版考试(full form)成绩误差分布的前提下,计算各个简版最大分散度条件下的α错误的概率、统计检验力以及考生覆盖率.按照误差尽量小、α错误概率小于0.05、统计检验力大于0.90且考生覆盖率95%以上的标准,逐一比较各个简版的性能,结果发现由逻辑推理、阅读理解、图表解读与短文加工4个分测验组成的简版性能最优.由此提出考试简版测验开发中应综合考虑测量误差、α错误、统计检验力、覆盖率、考试时间5项标准.The principle of exploratory-factor analysis was followed to explore the optimal short form of the General Abilities Test for new staffs.Data of 1 335 candidates who applied for professional-skill positions were analyzed.For only one factor by principle component analysis,8short forms were created,the number of subtests increased respectively by 2~9according to their factor loadings of its subtests.Given the distribution of score errors between the full form and its short form,the probabilities of errorα,the power of statistical testing,and the covering percentage of candidates were calculated for each short form in maximum dispersion of a short form among its subtests.It was found that given the permitted value of divergence,the more types one test contained,the bigger the degree of dispersion,the bigger the rate of coverage,the bigger the statistical power,and the bigger the rate ofαerror.It was found that the 4-subtest version including Logic Reasoning,Reading Comprehension,Diagram Understanding,and Essay Processing was the optimal short form with the following standards:minimum error,probability of errorα < 0.05,power of statistical testing >0.90,and covering percentage of candidates> 95%.

关 键 词：简版测验 分测验 误差分布法 考试 

分 类 号：D630.3[政治法律—政治学]