Estimating TOEIC scores using G-TELP scores: A Bayesian Model in a Japanese National University

Abstract

The purpose of this study is to examine (a) whether students’ TOEIC scores will increase as they gain higher English proficiency measured by G-TELP, a surrogate English proficiency test and (b) how and what kind of factors (such as individual, faculty, and year) affect their scores most. The target group is students in a Japanese national university from nine departments, sampled from school year 2011 to 2015. To explore the objectives, we generated hierarchical Bayesian models to estimate the tendency and differences together. We obtained the following three results. First, students’ TOEIC scores rise as their English proficiency improves. Second, the three factors affect the TOEIC scores. Finally, the similarly-proficient students with lower TOEIC scores raise their TOEIC scores more than those with higher scores if the students improve their own proficiency. We conclude that the students’ TOEIC scores will improve if the students raise their proficiency. Moreover, if we judge their English proficiency in terms of their TOEIC scores, we should consider the three factors carefully.