ABSTRACT
Many English-medium universities employ a compensatory model to establish cutscores on English language proficiency tests for student admissions. In this model, students can have different scores on different sections of the test provided their overall score meets the admission cutscore. This practice raises questions regarding potential variation in academic achievement among students with diverse score profiles. To address these questions, this study compared the demographic characteristics and academic achievement of 3,694 undergraduate students who met the required cutscore on the IELTS-Academic for admission to a Canadian English-medium university but have different scores on different sections of the test. The findings indicated that, generally, students with medium reading scores combined with low or medium writing scores exhibited lower academic achievement compared to those with high scores on all the IELTS sections or high reading scores combined with medium or high writing scores. The profile groups demonstrated significant differences in certain demographic characteristics, potentially explaining why they have different proficiency levels in different language skills. The implications of the findings are discussed, including whether to maintain the compensatory model or switch to a mixed one, and the implications for providing English language support to students with diverse proficiency profiles.
Acknowledgments
I would like to thank LAQ reviewers and Elvis Wagner for comments and feedback on earlier versions of the paper.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Similar concerns and questions have been raised in relation to the use of a compensatory model with the TOEFL iBT too. These are addressed in a related study.
2 There were no changes in the admission requirements in relation to scores on the IELTS during this period.
3 Landed immigrant are students who have recently immigrated to Canada but have not acquired citizenship yet.
4 Students who report English as their L1 but are from a non-English speaking country are required to submit ELP test scores. Similarly, students who are landed immigrants or Canadian citizens but did all their previous studies abroad and/or in non-English speaking schools must submit ELP test scores.
5 Each term consists of 12 three-hour classes. In the fall (September-December) and winter (January-April) terms, classes meet once a week over 12 weeks, while in the spring (May-June) and summer (July-August) terms, classes meet 2 or 3 times a week over 6 or 4 weeks.
6 Assumptions for RM ANCOVA include normality in variable distribution, linearity, and homogeneity of variance.
7 For significant effects, effect size was estimated using partial eta-squared (partial ἠ2). Cohen (Citation1969) suggested that partial ἠ2 values of .01, .06, and .14 indicate small, medium, and large effects, respectively (pp. 278–280, cited in Richardson, Citation2011, p. 140).
8 IBM (Citation2023) recommends using the following benchmarks for interpreting Cramer’s V: V ≤ .20 indicates a small effect; .20 ≤ V ≤ .60 indicates a medium effect; and V ≥ .60 indicates a large effect.