Since 1997 I have been Educational Advisor to the MRCP(UK) examination which is one of the largest, oldest and most distinguished of the postgraduate medical examinations held in the UK, being first sat in 1859. Postgraduate examinations have in the past been much criticised on a number of grounds. In the late-1990s the MRCP(UK) was completely redesigned, with the introduction of a new clinical examination (PACES), and the conversion of the Part 1 and Part 2 Written examinations from the old multiple true-false format, to the more respected best-of-five format, these changes being introduced in 2002.
Research into the MRCP(UK) has not involved the collection of new data, but instead has concentrated on detailed analysis of the large amounts of data that is available from routine statistical analyses. Publications in recent years have looked at:
- The reliability of the Part 1 examination
- An analysis of changes in the standard of candidates passing the Part 1 examination
- The tendency of clinical examiners in the MRCP(UK) PACES examination to be 'Hawks' or 'Doves'
- Differences in performance of candidates on all parts of the MRCP(UK) examination in relation to sex and ethnicity, and the inter-relation of candidate sex and ethnicity with examiner sex and ethnicity in the PACES examination
- Differences between graduates of UK undergraduate medical schools in performance on the Part 1, Part 2 and PACES examinations of the MRCP(UK)
- The process of standard setting using the Angoff method in the Part 1 and Part 2 examinations
- Whether the standard error of measurement is a better measure of quality for postgraduate examinations than is reliability.
In collaboration with the Royal College of Paediatrics and Child Health I have also described a statistical method, entitled Acinonyx, for detecting anomalous pairs of candidates who are taking multiple choice examinations and whose answers are more similar than chance might expect, and therefore require further investigation.
An example of an MRCP paper from 1869