Predicting patterns of change and stability in student performance across a medical degree

Barbara Griffin*, Piers Bayl-Smith, Wendy Hu

*Corresponding author for this work

    Research output: Contribution to journalArticlepeer-review

    10 Citations (Scopus)


    Context: Evidence of predictive validity is essential for making robust selection decisions in high-stakes contexts such as medical student selection. Currently available evidence is limited to the prediction of academic performance at single points in time with little understanding of the factors that might undermine the predictive validity of tests of academic and non-academic qualities considered important for success. This study addressed these issues by predicting students’ changing performance across a medical degree and assessing whether factors outside an institution's control (such as the uptake of commercial coaching) impact validity.

    Methods: Three cohorts of students (n = 301) enrolled in an undergraduate medical degree from 2007–2013 were used to identify trajectories of student academic performance using growth mixture modelling. Multinomial logistic regression assessed whether past academic performance, a test of cognitive ability and a multiple mini-interview could predict a student's likely trajectory and whether this predictive validity was different for those who undertook commercial coaching compared with those who didn't.

    Results: Among the medical students who successfully graduated (n = 268), four unique trajectories of academic performance were identified. In three trajectories, performance changed at the time when learning became more self-directed and focused on clinical specialties. Scores on all selection tests, with the exception of a test of abstract reasoning, significantly affected the odds of following a trajectory that was consistently below average. However, selection tests could not distinguish those whose performance improved across time from those whose performance declined after an average start. Commercial coaching increased the odds of being among the below-average performers, but did not alter the predictive validity of the selection tests.

    Conclusion: Memory for melodies in unfamiliar tuning systems: Investigating effects of recency and number of intervening items
    Original languageEnglish
    Pages (from-to)438-446
    Number of pages9
    JournalMedical Education
    Issue number4
    Publication statusPublished - Apr 2018


    Dive into the research topics of 'Predicting patterns of change and stability in student performance across a medical degree'. Together they form a unique fingerprint.

    Cite this