Interviews should be structured or semistructuredBMJ 1996; 313 doi: https://doi.org/10.1136/bmj.313.7065.1149b (Published 02 November 1996) Cite this as: BMJ 1996;313:1149
- Mairead Boohan,
- Robert W Stout
- Lecturer in medical education Dean Faculty of Medicine, Queen's University of Belfast, Belfast BT7 1NF
EDITOR,—Sandra Goldbeck-Wood refers to the fact that using A level results allows medical schools to attenuate “numbers sufficiently to make individual interviews practicable.”1 This suggests that the interview is an appropriate technique to augment other selection procedures. There are, however, numerous problems associated with using the interview as a selection tool. Most of these problems pertain to interviewer bias, which results in low validity and interrater reliability coefficients. The sources of this bias have been studied extensively by occupational psychologists and include the following: the effects of information available before the interview (particularly unfavourable information) on both the questioning style and evaluation of the candidate's performance; errors associated with judgment (principally those relating to leniency and severity of rating); bias due to contrast effects (that is, an average candidate may seem better or worse if interviewed immediately after a succession of either very good or very poor applicants); the interviewer's own prototype of the “ideal” candidate; and prejudice on the grounds of race and sex.2
Research by psychologists has shown that much of this bias can be eliminated by use of either a structured or a semistructured interview. A meta-analysis of the literature on selection interviews reported average validity coefficients of 0.35 (uncorrected) increasing to 0.63 (corrected) for structured interviews.3 This dropped to 0.20 (corrected) for unstructured interviews.
Medical schools use either semistructured or unstructured interviews when selecting medical students.2 In recent years a small number of medical schools have attempted to improve the validity and reliability of the selection interview by using structured interviews; most, however, still use an unstructured format.4 This implies that neither the attributes to be assessed nor the required levels of attainment have been defined appropriately before the interview. Furthermore, evaluation of the effectiveness of the interview as a selection technique tends to be carried out on an ad hoc basis. None of this is surprising in the absence of agreement about the most appropriate criteria on which to select medical students.1 5 Until there is concordance on this issue it is naive to suggest that one selection technique is better than another.