ANALYZING STUDENTS’ COURSE EVALUATIONS USING TEXT MINING: VISUALIZATION OF OPEN-ENDED RESPONSES IN A CO-OCCURRENCE NETWORK

Abstract

Japan’s Standards for Establishment of Universities states, “A university shall conduct organized training and research to improve the content and methodology used in courses at said university.” Based on this, most of Japan’s universities have recently implemented course evaluations by students. Student course evaluations are intended to quantify and provide an understanding of students’ satisfaction with their courses, and all universities are implementing them as one way to objectively evaluate courses. These course evaluations often combine computer-graded multiple-choice items with open-ended items. Computer-graded multiple-choice items are easy to assess because the responses are quantifiable. However, open-ended items’ responses are text data, and objectively grasping the students’ general tendencies is challenging. Moreover, it is difficult to avoid risking arbitrary and subjective interpretations of the data by the analysts who summarize them. Therefore, to avoid these risks as much as possible, the so-called “text-mining” method or “quantitative content analysis” approach might be useful. This study shares our experiences using text mining to analyze students’ course evaluations through the visualization of their open-ended responses in a co-occurrence network, and we discuss the potential of this method.