Objective: This paper describes the development of the tests of competence used as part of the General Medical Council's assessment of potentially seriously deficient doctors. It is illustrated by reference to tests of knowledge and clinical and practical skills created for general practice. SUBJECTS AND TESTS: A notional sample of 30 volunteers in 'good standing' in the specialty (reference group), 27 practitioners referred to the procedures and four practitioners not referred but who were the focus of concern over their performance. Tests were constructed using available guidelines and a specially convened working group in the specialty.
Methods: Standards were set using Angoff, modified contrasting group and global judgement methods, as appropriate.
Results: Tests performed highly reliably, showed evidence of construct validity, intercorrelated at appropriate levels and, at the standards employed, demonstrated good separation of reference and referred groups. Likelihood ratios for above and below standard performance based on competence were large for each test. Seven of 27 doctors referred were shown not to be deficient in both phases of the performance assessment.