The purpose of this study was to determine which types of student application information, as well as demographic information obtained through a questionnaire after matriculation, best predicted later academic performance in an international English-medium university in Japan, and to examine the "big picture" of how cognitive and non-cognitive variables interact over time in accounting for student success in an English for Academic Purposes (EAP) program and in the regular university program. The study was divided into three parts that separately examined student application information, university entrance examinations, and the larger picture of student success.
In the first part of the study, a hierarchical multiple regression was employed to determine the extent to which a variety of variables derived from application information predicted grade point average (GPA) in the EAP program, as well as first-year GPA and final GPA in the regular university program. The independent variables examined in the main regression analysis were: high school grade point average (HSGPA); ITP TOEFL scores obtained in April of the students' first year at the university; and hensachi rankings of the students' high schools. Results indicated that HSGPA was a consistently significant predictor of all levels of university GPA. ITP TOEFL scores significantly predicted EAP GPA, and hensachi rankings were significant predictors of first-year GPA and final GPA. In pullout studies, additional variables were examined, including high school grade factor scores, ITP TOEFL section scores, and non-cognitive variables derived from student responses to a questionnaire (e.g., gender and parents' education). Results indicated that high school grade factor scores and ITP TOEFL section scores varied in the degree to which they predicted GPA, female gender predicted HSGPA and EAP GPA, and the father's education predicted HSGPA (though the correlation was negative).
In the second part of the study, university entrance examinations were examined to determine the extent to which they predicted EAP GPA and first-year GPA. The Center Examination section scores, four types of university entrance examinations, and numerical scores from the examinations (e.g., English test, Japanese test, English essay, and Japanese and English interview) were examined in further hierarchical multiple regression analyses to determine how valuable each type and measure was for predicting university success. Results indicated that the Center Examination English test scores and the university entrance examination English essay scores significantly predicted EAP GPA, and that the Center Examination Math test scores and the university entrance examination English test scores were significant predictors of first-year GPA.
In the third part of the study, the larger picture of student success was examined. Logistic regression was first employed in order to determine to what degree HSGPA, high school grade factor scores, ITP TOEFL scores, hensachi rankings, gender, and parents' education predicted timely and exemplary completion of program requirements (e.g., finishing the EAP program on time (FOT), graduating on time (GOT), and graduating with honors (GWH)). Results indicated that HSGPA, the Numerical Ability factor scores, ITP TOEFL scores, and the father's education were significant predictors of FOT. Although only the ITP TOEFL scores were significant predictors of GOT, HSGPA, Language Ability factor scores, hensachi rankings, and ITP TOEFL scores significantly predicted GWH. Path analysis was then used to examine the path of success, from before matriculation, through the first year of university, and on to graduation. Well-fitting models were produced for both first-year GPA and final GPA in which all levels of GPA were stable predictors of later GPAs, and hensachi rankings and ITP TOEFL scores contributed most to the model up to the first-year GPA.