摘要
考试的公平性问题一直备受全社会的高度关注,但是影响考试公平的主要因素之一即测评的信度问题,却长期不被重视。社会上不仅出现了滥用信度极低的测试的问题,还出现了信度估计方法不当的问题。其中,多题型试卷和多门学科合成分数的信度被不恰当估计的问题最为严重。本文以举例的方式,分别讨论了分层α(αstrat)和成套测验信度估计方法在多题型试卷和多学科合成分数信度估计方面的应用情况,并指出了它们之间的关系、适用范围和注意事项,同时还为实践工作者提供了一个成套测验信度估计的Excel计算模板样例。
It is essential for tests to have high reliability to ensure their fairness. However, reliability has seldom been attracted the attention of the public, and its estimation methods are often inadequately used, especially for tests that contain mixed item formats. This paper discusses two methods for estimating reliability： the stratified a for a single test containing mixed item formats and an additional one for a battery composite score. The relationship between these two methods are explored and demonstrated with examples. The typical misuse of these methods is mentioned. An excel template is also provided for estimating the reliability of a battery composite score.
出处
《教育测量与评价》
2017年第4期5-9,15,共6页
Educational Measurement and Evaluation
关键词
测评信度
分层成套测验信度
reliability, stratified a , composite score seliability
