DOI: 10.36347/sjams.2016.v04i11.003

Pages: 3868-3875

The aim is to describe the psychometric properties of the July 2015 Paediatrics OSCE scores of 19, 3rd year Clinical Medicine students in a resource-limited medical school in Tanzania, East Africa, with the goal of improving assessment. This descriptive and cross-sectional study used literature review and structured observation as data collection methods. Students’ performances were assessed using checklists by 4 examiners in 27 OSCE stations. Stations 24-27 were manned and stations 1-23 were in written format. Statistical analysis was carried out using SPSS computer packages. The written stations were too many in proportion to the clinical stations for an OSCE. The mean scores in the stations were between 5-78% and pass mark was set between 0-97percent. The tasks were either too easy or too difficulty in 26% (7stations, n=27) of the stations, the variance was generally too high for a criterion referenced test and none of the means was centrally located. Station 19 was extremely difficult. Examiners’ error was moderately high at 66%, stability was too high (Alpha was 0.6, ideal is up to 0.5) and internal consistency ranged from -0.4 to 0.6 (alpha). G-coefficient was low at 0.22. Item analysis was good in 30% (8) but poor in 19% (5) of the stations. The analysis recommended that 19% of the stations was to be discarded, 52% of the stations to be reviewed and 30% of the station (whose properties were good) were to be banked for future use.