Multiply or average? A method of exploring interrelations between different tests

Authors
Citation
M. Hobsley, Multiply or average? A method of exploring interrelations between different tests, TEACH L MED, 11(4), 1999, pp. 249-253
Citations number
1
Language
INGLESE
art.tipo
Article
Categorie Soggetti
General & Internal Medicine
Journal title
TEACHING AND LEARNING IN MEDICINE
ISSN journal
1040-1334 → ACNP
Volume
11
Issue
4
Year of publication
1999
Pages
249 - 253
Database
ISI
SICI code
1040-1334(199923)11:4<249:MOAAMO>2.0.ZU;2-3
Abstract
Background: The Professional and Linguistics Assessment Board of the Genera l Medical Council examines foreign candidates wishing to practice medicine in the United Kingdom. Several separate tests and I conjoint test of each c omponent (medicine and English) are used. Purpose: This investigation aimed to determine how much the candidate's kno wledge of English affected the ability to pass the medical component, and v ice versa, and how much dependence between tests there was within each comp onent. Methods: The 14 examinations held during 1993 for 1,940 foreign candidates were analyzed. The technique used was to note whether, by and large, the ca ndidates who failed I test were the same as or different from those who fai led the other test(s). If the same candidates failed 2 tests, the overall p ass rate was approximately the average of the separate pass rates of the 2 tests. If the candidates who failed I test passed the other and vice versa, then the product of the 2 pass rates approximated to the observed over-all pass rate. Results: Using this method, it was found that the language component of the examination was one-third dependent, two-thirds independent of the medical component that the one-third dependence was due to the Oral Language Test, and that the two-thirds independence was due to the Oral Medical Test. Mor eover, the objective tests of medicine were independent of the subjective O ral Medical, and there was a similar incongruity between the subjective and oral tests of language. Conclusions: Two hypotheses are advanced that might explain these results. They are not mutually exclusive, and data do not exist at present to demons trate their relative importance. The method used should be applicable to ot her examinations using more than one test procedure.