The Examination of Theory to Estimate the Amount of Extension Time for Test-Takers with Disabilities by analyzing Their Answering Processes

本研究的目的是探讨在我们以前的研究中开发的时间-反应率曲线的估计方法，以定量地估计公平量的测试，如大学入学考试的残疾考生的延长时间。该方法能够估计的时间限制的方法下，在普通测试的延长时间的量，虽然我们的实验下进行的工作限制的方法。我们发展了一种新的方法，通过比较非残疾学生和视力残疾学生的得分率的时间-得分率曲线来估计他们的得分率。在时间-得分率曲线中，横坐标表示回答时间，纵坐标表示相对得分率。在实验中，我们发现只有很少的差异，估计的延长时间的新方法（时间分数率曲线）和以前的方法（时间响应率曲线）的每个测试（日语，数学和英语）。因此，我们可以建议，如果两组的分数分布存在一些差异，则新方法适合于估计合理的延长时间量，而如果分数分布几乎没有差异，则前一种方法也是有用的。通过实验证实，试卷中回答过程的特征铅笔测试也可以通过我们使用笔式计算机开发的新的计算机化测试系统来实现，以使我们的研究成为可能。为了评估计算机化测试系统的性能，我们采用拉丁方设计进行了实验。测试方法（电脑化测试和普通纸笔测试）对测试的回答时间和得分几乎没有显著的主效应。因此，这种计算机化的考试系统可以代替纸笔考试。

This research is aimed at examining the estimation method by time-response rate curves developed in our previous study to estimate the fair amount of extension time for tests such as university entrance examinations for test-takers with disabilities quantitatively. The method enabled to estimate the amount of extension time in ordinary tests under the time-limit method, though our experiments were conducted under the work-limit method. We have developed a new method to estimate by comparing the time-score rate curves of nondisabled students to that of students with visual disabilities in their score rate. In the time-score rate curves, an abscissa means answering time and an ordinate means relative rate of accumulated score. In the experiments, we found only little difference between the extension time estimated by the new method (time-score rate curves) and the previous method (time-response rate curves) for each test (Japanese, mathematics and English). Therefore, we can recommend that the new method is apropriate to estimate the fair amount of extension time if there is some differences in the score distibutions of two groups, and that the previous method is aoso useful if there is almost no difference in score distributions.Moreover, it was established by experiments that the investigatins of answering processes in the paper-pencil tests were also enabled by the new computerized test system we had developed by using a pen computer to make possible our study. To evaluate the performance of the computerized test system, we conducted the experiments by a Latin square design. There were almost no significant main effects of test methods (computerized tests and ordinary paper-pencil tests) on the answering time and score of tests. Therefore, it is concluded that such a computerized test system can be used instead of paper-pencil tests.