Estimating Frequencies of Misclassified Students： A Comparison Between Discriminant Analysis and the Intersection-o f-Distribution-Based Method  

作　　者：Florian Klapproth 

机构地区：[1]University of Luxembourg, Luxembourg City, Luxembourg

出　　处：《US-China Education Review(A)》2014年第9期603-610,共8页美中教育评论（A）

摘　　要：In several European countries, students are allocated to different tracks of secondary school depending on their achievements in primary school. However, the assignment of students lacks precision since often high achievers are allocated to lower tracks, and vice versa. To estimate the frequency of misclassified students, a method has been suggested that draws upon the distribution of scores of standardized achievement tests. In case of two distributions with equal variances, each representing achievements of students from either a lower or a higher track, the number of misclassified students has been shown as being equal to half the overlap of both distributions. However, the same estimate will be yielded when using discriminant analysis. In simulation studies, both the variances of the distributions and the group sizes were varied. It turned out that estimates of both methods differed substantially when the variances of the distributions and the group sizes were not equal. The results are discussed with respect to the characteristics of both methods.

关 键 词：student classification TRACKING discriminant analysis test scores 

分 类 号：O212.1[理学—概率论与数理统计] X928[理学—数学]