分数表达的常见方式及其潜在风险的规避  

Scaling Methods and Risk-Free Strategies in Test Score Reporting

作　　者：杨志明[1] 李洋 丁港 Yang Zhiming;Li Yang;Ding Gang

机构地区：[1]湖南师范大学测评研究中心、外国语学院,长沙410081 [2]湖南师范大学教育科学学院 [3]湖南师范大学数学与统计学院

出　　处：《教育测量与评价》2021年第4期3-10,共8页Educational Measurement and Evaluation

基　　金：国家社会科学基金“十三五”规划2020年教育学重点课题“中西部地区推进高考综合改革研究”(AEA200013)的部分成果。

摘　　要：考试分数主要有常模参照和标准参照两种表达方式,美国新版高考SAT-I成绩报告单中的6种常模参照分数和3种标准参照分数对此提供了具体案例。在设计和使用分数时,需要警惕来自测评信度、效度、公平性、原始分使用、标准分或等级分的设计与使用条件、常模与等值的结合,以及量表“漂移”等7个方面的潜在风险。规避这些风险的办法主要有加大科普力度、设立行业标准、成立考试政策与技术咨询委员会、发掘考试大数据的价值、促进考试工作的专业化发展等。Test scores are typically scaled either in a norm-referenced or criterion-referenced way.A good example isthe redesigned SAT-I score report,in which 6 norm-referenced scores and 3 criterion-referenced scores are informed.The quality of scaling,however,may be impacted by several factors,such as poor reliability,insufficient evidence of validity,item bias,misuse of raw scores,violation of scaling assumptions,lack of equating,and(or)scale drifting.This paper recommends some risk-free strategies in score reporting.These include promoting public education on testing science,setting academic standards,organizing a national advisory committee on testing policy and technology,conducting data mining on the national college entrance exams,and promoting professional development of the testing industry.

关 键 词：常模参照分数 标准参照分数 标准分数 常模研发 测验等值 

分 类 号：G40-058.1[文化科学—教育学原理]