Discriminant analysis on university students’ reading skills

Experiments using class learning outcomes and certificate exam scores

抄録

Scores of English certificate exams have been drawing increasing attention as one of the indexes for English learning activities. The certificate exams can evaluate proficiency of test takers objectively. However, the exams are usually expensive to take, making it difficult for students to take ones frequently. If students’ proficiency based on certificate exams’ result can be estimated by using learning outcomes of English course, their objective proficiency can be assessed without additional cost. This study used learning outcomes of 76 students in a public university in Ishikawa. Their learning outcomes (weighted average scores of mid-term and final exams, assignment average scores, assignment average study hours, times of assignment submission) in an English reading course in 2018 fall semester and reading scores of TOEIC Bridge in February of 2019 were used for discriminant analysis. A model using weighted average sores of mid-term and final exams and assignments average scores yielded 89.04% accuracy to classify the students into upper and lower groups. The threshold of the two groups are set on 80 % correct score of TOEIC Bridge reading scores. The results also showed a necessity to give students appropriately calibrated assignments on regular basis.