|国家预印本平台
首页|心理与教育测验分类信度:分类一致性评估方法

心理与教育测验分类信度:分类一致性评估方法

英文摘要

lassification consistency reflects the probability that a participant will obtain the same classification on two parallel tests. It is widely used by test administrators to assess the reliability of psychological tests, educational tests, and medical diagnostic classifications. Since administering parallel tests is often challenging in practice and the internal consistency is not appropriate for classification tests or criterion-referenced tests, many methods are focused on estimating classification consistency based on results from a single test administration in psychological and educational measurement. These estimation methods are crucial for assessing and improving the reliability and fairness of tests. The study focused on the investigation of the general approach and representative methods based on criterion-referenced tests for estimating classification consistency under classical measurement theory, item response theory, cognitive diagnostic models, and machine learning models. The ideal and procedures of the representative methods under each model were introduced in details. A series of examples were illustrated about how to apply classification consistency indices for evaluating test reliability. Their advantages, disadvantages, and applicability in different testing contexts were also analyzed. Future research should consider the methods of estimating the confidence intervals of classification consistency. Researchers and practitioners should widely apply and report classification consistency to better evaluate the quality of test classification results.

陈静仪、宋丽红、汪文义

江西师范大学计算机信息工程学院江西师范大学教育学院江西师范大学计算机信息工程学院

教育信息传播、知识传播

分类信度分类一致性决策规则认知诊断机器学习

classification reliabilityclassification consistencydecision rulescognitive diagnosismachine learning

陈静仪,宋丽红,汪文义.心理与教育测验分类信度:分类一致性评估方法[EB/OL].(2025-04-11)[2025-08-18].https://chinaxiv.org/abs/202504.00155.点此复制

评论