Objectives: To assess the reliability of the MRCP(UK) Part I Examination over the period 1984-2001, and to assess how the reliability is related to the difficulty of the examination (mean mark) and to the spread of the candidates' marks (standard deviation).
Methods: Retrospective analysis of the reliability (KR20) of the MRCP(UK) examination recorded in examination records for the 54 diets between 1984 and 2001.
Results: The reliability of the examination showed a mean value of 0.865 (SD 0.018, range 0.83-0.89). There were fluctuations in the reliability over time, and multiple regression showed that reliability was higher when the mean mark was relatively high, and when the standard deviation of the marks was high.
Conclusions: The reliability of the MRCP(UK) Examination was maintained over the period 1984-2001. As theory predicted, the reliability was related to the average mark and to the spread of marks.