Abstract
The landscape of higher education has changed considerably due to internationalisation and massification. Although a positive phenomenon, challenges concerning academic language proficiency have arisen. Concomitantly, rates of attrition have become concerning for all stakeholders; therefore, predictive studies identifying students likely to require additional support have been undertaken to enhance retention and success rates. Similarly, post-entry language assessments or PELAs assist in identifying students who may lack academic language skills. PELAs are ubiquitous in Australia, yet there is limited literature investigating the validity of PELA score interpretations and uses and the potential predictive qualities of the measures. This thesis aimed to determine whether academic and non-academic factors could predict academic performance (i.e., grade point average, [GPA]) of undergraduate students. The main factor investigated was Bond English Language Assessment (BELA), defined as a PELA screening tool measuring academic essay writing skills of undergraduate students. Via a mixed-method approach, this thesis also aimed to investigate the validity of the interpretation and use of BELA scores.Findings provided support for the validity argument for BELA in the context of its use at the research site. The research identified concerns regarding inter-rater reliability, academic integrity, and the provision of feedback. Recommendations in response to each concern, included more in-depth and frequent rater training for academic staff, with the development of an online rater training module, and routinely monitoring ratings. It was also suggested that further investigation be undertaken concerning the administration of BELA in an environment which allows invigilation, especially with the rapid development of generative artificial intelligence. Furthermore, it was recommended that all students receive general, online feedback on their essays. A final recommendation was to determine the appropriateness of widening the group of students required to attend Academic Skills Centre (ASC) consultations.
Turning to the main purpose of the research, via logistic regression, this thesis identified academic and non-academic factors predicting academic performance of undergraduate students. Regarding students likely to fail after two semesters, Indigenous Australian students were 4.06 times more likely to fail in comparison to non-Indigenous students; students with Below satisfactory academic writing were 3.77 times more likely to fail compared to students with Satisfactory writing; Bond University College students were 3.12 times more likely to fail than students from other faculties; students who did not attend ASC were 2.89 times more likely to fail in comparison to those who engaged; males were 1.93 times more likely to fail compared to females, English as an additional language or dialect (EAL/D) students were 1.89 more likely to fail in comparison to English-speaking background students; finally, Bond Business School students were 1.74 more likely, compared to students from other faculties, to fail after two semesters.
In terms of high academic achievement, the odds of maintaining a GPA above 3.0 (i.e., a Distinction average or above) after two semesters were: 5.16 times higher for students with Satisfactory academic writing, 4.63 times greater for Health Sciences and Medicine students, 1.67 times higher for English-speaking background students, 1.45 times greater for females, and 1.03 times more for older students.
The creation of a profile of students at risk of potential academic failure and, conversely, a profile of students likely to excel academically, assists decision makers in allocating resources to support students in an effort to reduce attrition and enhance retention and success rates. This thesis contributes to the literature on success and retention at university and evaluating the validity of PELAs. This research opens the door to further studies into the role of predictive factors, including PELAs, within other contexts.
| Date of Award | 28 Nov 2024 |
|---|---|
| Original language | English |
| Supervisor | Beata Webb (Supervisor), Masanori Matsumoto (Supervisor) & Gaelle Brotto (Supervisor) |