Correlation between different PBL assessment components and the final mark for MB ChB III at a rural South African university

  • ME Garcia-Jardon
  • EV Blanco-Blanco
  • VG Bhat
  • SD Vasaikar
  • EN Kwizera
  • A Stepien

Abstract

Background. Problem-based learning (PBL) is now an accepted component of many medical school programmes worldwide. Our university also follows the PBL ‘SPICES’ model for MB ChB III. The assessment modalities used are the modified essay questions (MEQ), objective structured practical examination (OSPE), individualised process assessment (IPA) and tutorial continuous assessment (TUT). This study was done to compare the students’ performances in individual assessment components with the final mark to determine the correlation between these parameters. Materials and methods. The study was retrospective, descriptive and analytical, based on the integrated marks of all the MB ChB III students at Walter Sisulu University (WSU) in 2007. Assessment marks were stratified according to blocks and different types of assessment (MEQ, TUT, OSPE, IPA). Regression analysis was used to compute and scrutinise these vis-à-vis their correspondence with the final marks for each block. Results. Three hundred and seventy-nine block assessment marks of 96 students from 4 blocks of MB ChB III were analysed and the correlation between the assessment components and final mark were compared. Regression analysis showed good correlation when analysing the assessment modality versus the final mark for the MEQs (r=0.93, 0.93, 0.94, 0.96), followed by OSPEs (r=0.71, 0.70, 0.76, 0.77) and IPAs (r=0.62, 0.51, 0.68, 0.77). However, correlation was not significant with the TUT. Conclusion. There was good correlation between the students’ performance in the majority of assessment modalities and the final mark in the different blocks of the MB ChB III examination. There may be a need to make tutorial assessment methods more objective, partly by additional tutor training.
Section
Articles

Journal Identifiers


eISSN:
print ISSN: