Development of a Computer Assisted Testing System for Measuring Mathematics Achievement at the Junior High School Level.

本研究的目的是开发一个计算机辅助测试系统，以测量初中生的数学成绩。首先，我们收集了学生从入学到毕业进行的测试的反馈数据。我们得到168名一年级学生195项、135名二年级学生304项、124名三年级学生329项的反应数据，没有任何嘶嘶值。然后，我们根据一个人的正确或错误的答案输入项目响应数据。然后从矩阵数据中计算出2个题项之间的四分相关系数，通过成分分析得到因子加载分数，最后得到各年级试题的题项难度和题项判别度。最后，我们得到了每个测试的信息功能，并分析了教师自编测试与能力测试的差异特征。另一方面，我们在主机上设计了一个测试项目数据库。该数据库采用INQ（INformation Query）语言描述，由两个文件组成；一个数学部分文件和一个测试项目文件。section文件的记录包括section名称、section代码、学校和等级。测试项目文件的记录由项目代码、分段代码、项目难点、项目鉴别、说明和测试项目组成。我们使用区段代码作为密钥获取两个文件。我们通过FORTRAN语言访问它，以便我们能够根据项目难度，项目辨别和一个人的测试项目反应来计算学生的成绩。

The purpose of this research is to develop a computer assisted testing system in order to measure the student's mathematical achievement in junior high schools. Firstly, we have collected students' response data from the tests carried out from their entrance to their graduation. We have got 195 item response data of 168 first graders, 304 item of 135 second graders and 329 item of 124 third graders without any hissing value. Then, we have input item response data according to one's correct or wrong answer. After that, we calculated tetrachoric correlation coefficient between 2 items from the matrix data, and got factor loading scores by a component analysis, finally we have got item difficulties and item discriminations of the tests given to each grade. Finally, we have got information functions on each test, and analyzed characteristics of differences between teacher-made tests and proficiency tests.On the other hand, we designed a data base of test items on host computer. This data base is described by INQ(INformation Query) language and consists of two files ; a section file of mathematics and a test item file. The records of the section file consist of section names, section codes, schools and grades. The records of the test item file consist of item codes, section codes, item difficulties, item discriminations, instructions and test items. We serge two files using the section codes as the key. We access it by FORTRAN language so that we are able to calculate students' achievement depending on item difficulties, item discriminations and one's test item responses.