An Accurate Estimation Method for Abilities in Online Adaptive Testing Based on Item Response Theory


Recently, universities in Japan, in particular, private universities, are inclined to accept a wide variety of students because the number of high school students is becoming smaller in contrast to the increase of the number of enrollment of students. In response to this situation, universities have been gradually providing pre-classes before enrollment, various levels of classes, or follow-up classes after regular classes. However, it becomes very difficult to serve adequate learning chances to each student because the distribution of students' skills is spreading like a uniform distribution. We may need many teaching assistants if exhaustive learning classes are required. Instead, we have developed new learning systems to assist classes, called the follow-up program systems, consisting of learning check testing, follow-up program testing, and collaborative work testing; they have been working successfully in Hiroshima Institute of Technology. In these online testing systems, we adopt the item response theory to evaluate students' learning skills fairly. Although students' abilities can be estimated accurately by using a large number of responses to tests, we can also obtain the estimates for the abilities accurately with the small number of responses to tests if we use a method proposed in this paper, the EM-type IRT. This can estimate the response values to the empty elements during the estimation process. The incomplete matrix can be modified to the complete matrix. We may expect that this makes the estimated ability values more reliable. This can lead us to reconfigure the class design in the early stages.

Author Information
Hideo Hirose, Hiroshima Institute of Technology, Japan

Paper Information
Conference: ACP2017
Stream: Qualitative/Quantitative Research in any other area of Psychology

This paper is part of the ACP2017 Conference Proceedings (View)
Full Paper
View / Download the full paper in a new tab/window

Comments & Feedback

Place a comment using your LinkedIn profile


Share on activity feed

Powered by WP LinkPress

Share this Research

Posted by James Alexander Gordon