The relationship between selection scores and course outcomes for undergraduate medical students


  • Annette Mercer (UWA)
  • Margaret Hay (MU)
  • Katrina Simpson (MU)
  • Ian Puddey (UWA)
  • Ben Canny (MU)
  • Wayne Hodgson (MU)


10JJ Selection for admission to Medicine


Monash University (MU)
The University of Western Australia (UWA)


The selection processes into undergraduate medical courses at The University of Western Australia (UWA) and Monash University(MU) utilise three components: an academic score, an aptitude test and an interview.

The academic score is the Australian Tertiary Admissions Rank (ATAR), a composite score based on performance in state-wide examinations; and the aptitude test is the Undergraduate Medicine and Health Sciences Admission Test (UMAT). The UMAT is widely used in Australia and New Zealand and is representative of aptitude tests used for medical selection. A version of UMAT called the Health Professions Admission Test (HPAT) is used in Ireland.

The interview used at each university is specific to the institution, UWA using a structured interview with a focus on communication skills and Monash an 8-station multiple mini interview assessing a range of domains.


Summary of Work

Data were collated for students who entered the two medical courses from 2002 to 2011, consisting of demographics, entry scores and measures of performance in the course. The parts of the UMAT: logical reasoning and problem solving, understanding people and non-verbal reasoning were entered as separate variables.

Latent Growth Curve modelling was undertaken, to determine the profiles of the selection components across three time points as predictors of outcomes in the course. The time points were: T1, transition from campus-based to hospital-based; T2, transition to pre-internship; T3, end of course.

The M-plus program evaluated the trajectory of change (through latent intercept and slope) across the three time points for selection scores as predictors of achievement in the course.

Summary of Results
  • 1388 students had complete data across the three time points (51% female).
  • Good fit for the model was identified with RMSEA = 0.05, CFI = 0.99, TLI = 0.96, SRMR = 0.01.
  • Chi square = 35.64/df = 8, p = 0.00,acceptable due to the large sample size.
  • Intercepts across time were positive for AGE (z = 2.20, p = 0.03); ATAR (z = 14.01, p = 0.001); INTERVIEW SCORE (z = 3.861, p = 0.001) and negative for UMAT3 (z = -2.55, p = 0.01).
  • Although a general decline was seen in results (inter-individual change) across time, this was not significant and was not consistent for all students.
  • Covariance of intercept and slope of change (intra-individual change) showed that those who performed well at the beginning continued to do so ( despite the overall decline).
  • INTERVIEW SCORE was the only predictor to be positively associated with increased results over time (z = 2.446, p<0.001) i.e. there was an increase in slope from T1 - T2 to T2 - T3.
  1. Our model showed good fit.
  2. Age, ATAR and interview score were positive predictors across time.
  3. INTERVIEW SCORE showed a positive trajectory over time - an increase from T1 - T2 to T2 - T3, indicating greater association with achievement in the course in the later clinical years.
  4. High-achieving students generally remain as high achievers, in spite of a general decline in results for all students across the course.
Take-home Messages

Despite differences between each university’s interview format (panel and MMI), the value of interview scores was demonstrated by their increased positive association with higher results in the clinical years of both medical courses. This indicates good predictive validity of interview scores on course performance.

Therefore, this would suggest that even though resource intensive, interviews have an important role in medical course selection.


Hooper, D., Coughlan, J. & Mullen, M. R. (2008) “Structural Equation Modelling: Guidelines for Determining Model Fit.” The Electronic Journal of Business Research Methods 6(1) 53 – 60


The UMAT Consortium for funding this research project

Summary of Work
Summary of Results
Take-home Messages
Send ePoster Link