事业单位考试简版测验组成方法的比较研究?

分别基于因素分析法和逐步回归法,按照因子载荷和对误差解释率的大小,依次组成包含3、4、5、6个题型的简版测验.计算每种简版测验与全版测验的相关大小,以及在给定误差的情况下,每种简版测验的覆盖率.分别比较在给定及格分数线60和70分时,每种简版测验的总命中率和正命中率.结果为:1)基于因素分析方法开发的4种简版测验与全版测验的相关均达到显著水平,但重复测量的方差分析表明,这些简版测验与完整版测验分数的差异达到显著水平;2)基于逐步回归方法开发的4种简版测验与全版测验的相关均达到显著水平,但重复测量的方差分析表明,包含3和4个分测验的简版测验分数与全版测验的分数差异并不显著;3)当将60分作为及格线时,基于回归分析方法开发的简版测验的总命中率高于相应的基于因素分析方法开发的简版测验的总命中率,但正命中率却低于相应的简版测验.当将70分作为及格线时,基于回归分析方法开发的简版测验的总命中率高于相应的基于因素分析方法开发的简版测验的总命中率,但正命中率却没有表现出类似趋势.总体上讲,基于回归分析方法开发的简版测验与全版的相关,覆盖率以及命中率优于基于因素分析方法开发的简版测验.

An exploratory study on the optimal composition method of the General Abilities Test for new staffs

Exploratory-factor analysis revealed that only one factor was found by principle component analysis.Four short forms of tests were created,the number of subtests increased by 3,4,5,6 corresponding to factor loading of subtests.The same procedure was conducted using stepwise regression.It was found that correlation between four short forms based on factor analysis,and the full form was significant.It was also found that difference between all four short forms,and the full form was significant.Correlation between the four short forms based on stepwise regression,and the full form was significant.It was found that difference between the short forms which contained three and four subtests,and the full form was not significant.When 60 was used as the pass line,total hit ratio of short forms based on factor analysis was comparatively higher than those based on stepwise regression.However,positive hit ratio was comparatively lower than those based on stepwise regression.When 70 was used as the pass line,total hit ratio of short forms based on factor analysis was comparatively higher than those based on stepwise regression.It can be concluded that features of short form including correlation with full form,coverage and hit ratio based on stepwise regression are better than that based on factor analysis.