RESUMO
Docimology has allowed the development of evaluative processes assuring valid, reliable and objective assessments. It was adopted within the faculty of Medicine of Tunis since 2007. The aim of this study was to analyze the docimological survey results of hematology-oncology exams, to evaluate the interest of this analysis in the elaboration of exams and the construction of an item bank, and propose some corrections in order to improve assessment. Methods :We have analyzed the hematology-oncology exams of SCMS1 (Second cycle of Medical Studies 1) from educational year 2008-2009 to 2013-2014. The data input was already done with Excel. The test includes 4 disciplines (Hematology, Oncology, Genetics and the Anatomic Pathology). We have calculated docimological parameters allowing global analysis, by discipline and by item. Results : A total of 3281 papers and 1004 questions were analyzed. The average success rate per year was 91,54% ± 7,12. The highest average success rate was found in hematology (80,51% ± 10,18). The lowest rate was found in the anatomic pathology (51,61% ± 23,76). The average rate of students succeeding the test without having average note in hematology was 5,36%. It was 42,29% in the anatomic pathology. Average difficulty index was 0,57 ± 0,05. Items analysis showed that 38,04% were easy and 19,02% were difficult. Average discrimination index was 0,25 ± 0,02. Discrimination was very good in 20,51% of items and good in 17,13%. Useless and bad discrimination items were about 40,53%. The average of Cronbach Alpha coefficient was 0,84 ± 0,03, showing a good internal-consistency. Conclusion :This study allowed an objective evaluation of "contributive disciplines" in multidisciplinary evaluation and showed the interest of integrating questions. Question analysis with teachers would be important to reevaluate and improve these items.