Final Test Analysis Of Post Graduate Medical Residents

Background and purpose: Multiple choice questions are the most frequent test for medical students. It is important to analysis the overall response to individual  questions in the test.The aim of this study is to analyse questions of post graduate medical residency  tests.Methods: Final annual local (Ramadan medical school) and national tests given to three Residency groups  including  17 Obstetrics  and gynecology testees,  7 pediatrics  and  12 internal  medicine  in 2004 were studied. In local tests residents answered to 148, 150 and 144 and in national  tests to ISO MCQS. Questions were  evaluated regarding cognitive domain level, Difficultly index and Discriminative index  and finally to evaluate  the optimal,  proper, acceptable and  ''must  omitted" questions.Results: Questions of local Obstetrics and gynecology, pediatrics and internal medicine tests evaluated the "recall" level in 72%, 72% and 51% and in national  tests 71%,  35% and 19%,  respectively. Questions  with  Discriminative indices  of 0.7 or more (proper)  were 3 and  5% in  Obstetrics  and gynecology, 3.5% and 1% in pediatrics and 1% in local and national tests. Proper difficulty indices (30-70)  were shown in 53% and 54% in Obstetrics  and gynecology, 34% and 43% in pediatrics and 40% and  42% in internal  medicine.  Generally  evaluating,  "must  omitted" questions in local and national tests were 76% in Obstetrics and gynecology, 81% and 79% in pediatrics and 91% and 85% in internal medicine. The most common causes making the questions to be considered  "must omitted" in studied tests were negative, zero or less than 0.2 Discriminative indices.Conclusion: Test analysis  of final  annual  local  (Ramadan medical  school)  and national  tests  of Obstetrics  and gynecology, Pediatrics and internal medicine residency  programs  in 2004 revealed that most of the questions  are planned  in  "recall" level, harbor  improper  Discriminative indices, Difficulty indices and generally  evaluating are "must omitted".