高中会考等值方法的比较研究

本研究采用随机等组设计与铆测验相结合的方案。首先验证了两随机等组的平均数、方差和分布状态无显著差异，再用随机等组的等值分作为等值效标来检验其他等值方法的误差，然后比较了在铆测验设计中三种线性等值方法（在不同总体权重下）的误差值，以选出适合高中合考的等值方法及总体权重。经研究发现：会考等值宜采用Tucker观察分数线性等值方法，并宜选择总体权重W1＝1。

The Comparative Research on Equatig Method In the United Examination

The eperimental method in the research is characterized by the commonitem random equivalent groups design.In the design,it first proves that there exists no significant difference between the two means, variances and distributions, then uses the equated result using random groups as the equating criteria to calculate the errors of equating in the three linear methods and compares them.The main conclusions of the research as following: Tucker's method should be adapted to conduct equating functions of the United Examination,and the choice of weights should be W1 =1 and W2 = 0.