摘要
语言测试中的偏差是指由于与测试构念无关的因素导致某个或某些考生群体获得系统性高分或低分,从而影响测试结果的准确解释和使用,降低测试效度。自20世纪60年代以来,偏差研究作为高风险测试效度验证的必要环节已受到国外语言测试界的关注,但国内在这方面的研究几乎是空白。本文在综述国外研究成果的基础上,着重介绍偏差的概念、偏差的来源和偏差的主要侦测方法,以期对国内大规模高风险外语测试的偏差研究提供一定的理论依据和方法指导。
Test bias arises when a factor or factors inherent within a test that systematically prevents access to valid estimates of candidates' ability or when a test score is misinterpreted or misused. Investigation of test bias is central to the construct validation process. Therefore, test bias has been widely studied throughout the world since the 1960s. Due to lack of such studies in China, the paper attempts to focus on the following fundamental aspects: definition of bias, source of bias, and method of detecting bias. The purpose of this paper is to provide the rationale and methods for investigating bias in large-scale high-stake foreign language tests in China.
出处
《外语界》
CSSCI
北大核心
2008年第1期81-89,共9页
Foreign Language World
基金
上海外国语大学2007年研究生基金项目资助
关键词
测试偏差
效度验证
测试公平性
项目功能差异
test bias
validity validation
test fairness
differential item functioning