The BioMedical Admissions Test for medical student selection: Issues of fairness and bias

Joanne L. Emerya*b, John F. Bellac & Carmen L. Vidal Rodeiroada Cambridge Assessment, UKb JOANNE EMERY, PhD, is a research officer at Cambridge Assessment and has a PhD in Psychology from the University of Newcastle. Her background is in psychology, research design and statistics. Her research interests include admissions to higher education and the predictors of academic success.c JOHN BELL, MSc, is Head of Statistics at Cambridge Assessment. His research interests include educational statistics and multilevel modelling.d CARMEN VIDAL RODEIRO, PhD, is a senior research officer at Cambridge Assessment and has a PhD in Statistics from the University of Aberdeen. Her research interests are subject provision and uptake, subject choice, year to year stability of GCSE and A-level examinations and predictors of academic attainment in secondary school.* Correspondence: Joanne L. Emery, Research Division, Cambridge Assessment, 1 Regent St., Cambridge CB2 1GG, UK 44 1223 553841 44 1223 552700 emery.j@cambridgeassessment.org.ukBackground: The BioMedical Admissions Test (BMAT) forms part of the undergraduate medical admission process at the University of Cambridge. The fairness of admissions tests is an important issue.Aims: Aims were to investigate the relationships between applicants’ background variables and BMAT scores, whether they were offered a place or rejected and, for those admitted, performance on the first year course examinations.Method: Multilevel regression models were employed with data from three combined applicant cohorts. Admission rates for different groups were investigated with and without controlling for BMAT performance. The fairness of the BMAT was investigated by determining, for those admitted, whether scores predicted examination performance equitably.Results: Despite some differences in applicants’ BMAT performance (e.g. by school type and gender), BMAT scores predicted mean examination marks equitably for all background variables considered. The probability of achieving a 1st class examination result, however, was slightly under-predicted for those admitted from schools and colleges entering relatively few applicants. Not all differences in admission rates were accounted for by BMAT performance. However, the test constitutes only one part of a compensatory admission system in which other factors, such as interview performance, are important considerations.Conclusion: Results are in support of the equity of the BMAT.