“互联网+测评”中的数据有效性分析

Data Validation for Online Assessments

"互联网+测评"是当今特别热门的话题之一。但是,在线测评在测量信度、效度和公平性方面的问题也十分突出,而导致这些问题的一个重要因素就是测评数据的有效性不够。文章认为,在线测评中数据的有效性问题主要包括数据缺失和作答失真两种类型,并分别从统计学和教育测量学角度,对数据缺失和数据失真问题提出了具体的处理办法,同时提供了一个实例分析和一个估计缺失数据的SAS程序。

“Internet ＋ testing” ,or taking assessments online,is a new and fast growing trend in China. However, the reliability, validity and fairness of the online assessments may be poor due to invalid item responses, which are typically caused by missing data or fake item responses. This paper discusses solutions to data verification and missing data imputation using traditional statistical methods and the item response theory. One empirical study is introduced, and a SAS macro program for imputing missing data is provided.