Selection to medical school in Great Britain

British Medical Journal, April 3, 1999 by C Michael Steel, David Jackson, David W Sinclair, Stephen R Magee, D A Levison, D Parratt, J M Bland, Sheila M Gore, Chris McManus

Sheila M Gore Senior statistician Medical Research Council Biostatistics Unit, Cambridge CB2 2SR

[1] McManus IC. Factors affecting likelihood of applicants being offered a place in medical schools in the United Kingdom in 1996 and 1997: retrospective study [with commentaries by A Esmail and M Demetriou]. BMJ 1998;317:1111-7. (24 October.)

[2] Marshall EC, Spiegelhalter DJ. Reliability of league tables of in vitro fertilisation clinics: retrospective analysis of live birth rates. BMJ 1998;316:1701-4.

[3] Sanderson C, McKee M. Commentary: how robust are rankings? The implications of confidence intervals. BMJ 1998;316:1705.

[4] Goldstein H, Spiegelhalter DJ. Statistical aspects of institutional performance: league tables and their limitations (with discussion). J R Stat Soc 1996;159:385-444. (Series A.)

Author's reply

Editor--Steel et al discuss the deficiencies of my analysis and say that at St Andrews in 1997, applicants from ethnic minorities who met the school's entry requirements had a 98.5% chance of receiving an offer, compared with 97% for comparably qualified white applicants. Data cited by Steel elsewhere claimed, "Of the 452 eligible white applicants [for entry in 1997], offers were made to 428 (94.7%) while offers were made to 77 of the 87 eligible ethnic minority candidates (88.5%)" (personal communication). The relative risk of rejection for minority candidates seems to be 2.17 (11.5% v 5.3%; [chi square] = 4.72, df = 1, P = 030), with an odds ratio of 2.31 (95% confidence interval 1.06 to 5.04), which is similar to my own estimate. These data were calculated "by hand," based on predicted and achieved grades, and only considered candidates achieving the academic threshold with selection strongly weighted towards academic performance. The poorer outcome of ethnic minority applicants therefore requires proper explanation, rather than empty reassurances.

Levison and Parratt claim that my methodology is flawed. There is no paradox or error; Dundee met the criteria for significance, whereas Aberdeen, Oxford, and Glasgow did not. Nor was it reasonable to assume that the mean odds ratios were used since my paper clearly said otherwise.

Bland refers to the omission of GCSE and predicted A level grades. They would have helped interpretation but had not been collected electronically, and deans of medical schools in the United Kingdom agreed to proceed in their absence. They were, however, available in our 1991 cohort,[1] which resulted in similar findings to those in 1996 and 1997. By citing 1988-90 data, Bland acknowledges that previous studies help in interpreting the 1996 and 1997 data--despite critical claims that everything has changed since 1991 (albeit without numerical evidence).

Bland also comments on the frame of reference. Since I analysed applications to medical school in 1996 and 1997, I made no claims about universities as a whole or other sectors. Modood and Shiner's broader study of university applicants in 1992 did, however, find similar effects.[2] The remark, "It seems unlikely that staff concerned with admissions to medical school are universally racially prejudiced while those in the other university sectors are not," relates to nothing in my paper. Bland's idiosyncratic quasi-postmodernist analysis of the word "prejudice" I suggest as an exercise for those interested in discourse analysis and narrative based medicine.