基于随机交叉设计的试题计分误差来源及可靠性分析

The Analysis of Randomized Crossover Test's Scoring Error Sources and Reliability Basing on Generalizability Theory

概化理论在标准化参照系测验、非标准化测验、教师教学评价和人事测评等领域得到了广泛应用。通过对某高校运筹学试题进行随机交叉设计,探讨考试试题的计分误差来源和试题可靠性。研究表明,考生通过试题所获得的分数与掌握课程的真实水平之间的差异来自于考生本身掌握知识的水平、试题难易度以及考生和试题两者的交互效应。通过计算得出在不同题型下的变异分量估计值和试题可靠值,为测评试题的区分度和稳定性提供了一种工具。

Generalizability theory is widely applied to standardized tests, non-standardized tests,teaching evaluation and personnel evaluation. Basing on this theory, the paper sets up a randomizedcrossover design and analyzed the test＇s error sources and reliability. It puts forward that the differ-ence between examination results and true grades comes from how well the students have mastered theknowledge, the degree of difficulty of test and interaction of students and test. At last, it calculatesthe value of difference and reliability in different question types and provides a tool which can reviewthe test＇s differentiation and stabilization.