基于IRT模型的数学学业成就水平测试分析  

An Empirical Analysis on Mathematical Academic Achievement Test Based on IRT

作　　者：沈南山[1] 

机构地区：[1]皖西学院数学系,安徽六安237012

出　　处：《安徽师范大学学报（社会科学版）》2012年第1期67-73,共7页Journal of Anhui Normal University(Hum.&Soc.Sci.)

基　　金：安徽省哲学社会科学规划项目(AHSK09-10D188)

摘　　要：对不同类型学校的774名有效被试实施数学学业成就水平测试,并应用IRT参数模型方法进行分析,得出四点判断:(1)测验分数、最优分数呈负偏态分布;(2)测验信息函数负向偏移,大体呈现双峰波形;(3)主观性试题与逻辑斯蒂模型的拟合性较差;(4)不同类型学校学生的数学学业成就水平存在显著性差异。Mathematics academic achievement of proficiency test is based on the national mathematics curriculum standards,which is criterion-referenced evaluation that constructed professionally in order to assess student mathematics learning quality in basic education.Through stratified sampling,774 valid subjects distributed in the different types of schools are to be for mathematics academic achievement testing.We analysis Analyse the test dates in applying the IRT parameter model and obtain four judgments：（1） Test scores and the best scores should be positively distributedshow negative skewness;（2） the test information function should be offset to the negative direction,so that there would be generally show a double waveform.（3）The subject item is poorly fit with the logistic models;（4） There is an obvious difference among the mathematics academic achievement levels of different types of schools.

关 键 词：数学学业成就 标准参照性评估 项目反应理论 信息函数 实证分析 

分 类 号：G44[哲学宗教—心理学]