The differential predictive validities of University entrance examinations in Tanzania.
No Thumbnail Available
Date
2004
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Dar es Salaam
Abstract
This study investigated whether the matriculation examination had better selection validities than traditional form six examinations at the University of Dar es Salaam. The predictor variables were examination scores obtained from students' A- level examinations and matriculation examinations. The intervening variables assumed to have the potential to influence students' performance were gender and student background, whereas the criterion variables were final university student GPA. The sample consisted of the 1999 students' cohort from the Faculty of Arts (462) and the Faculty Science (141), and the 2001 cohort from the same faculties, Arts (484) and Science (168). Correlation analysis was carried out to explore the relationships between predictor variables and criterion variables. The findings showed that in the 1999 cohort, Form Six results were better predictors for Science students, with a correlation coefficient of r = 0.370, p < 0.01, compared to university matriculation examination which yielded a correlation of r = 0.292, p< 0.01. Again, in the 2001 cohort, Form Six results gave better predictive validities for Science students, with r = 0.439, p< 0.01 as compared to university matriculation examination result which gave a correlation coefficient of r = 0.274, p< 0.01. However, the matriculation examination was a better predictor for Arts students in the 1999 cohort with r = 0.316, p< 0.01 and gave a weak correlation in 2001 of r = 0.002 which was insignificant. The 1999 cohort was used to test the efficiency of the matriculation examination in students' selection decision. Among 61 students who performed below the cutoff point of 40 percent, only 12 percent indeed failed in the final university examinations, as compared to a failure rate 5 percent for those who had above cutoff point. In the two combined faculties, the results showed that the matriculation examination rejected 53 students who would have otherwise benefited from the university education. In other words, 87 percent of failures in matriculation examination were good university candidates. Two models (conjunctive versus compensatory) were tested in order to determine the best selection model. The compensatory model yielded stronger correlation of r = 0.361, p<0.01, than conjunctive model that gave r = 0.242, < 0.01. It was found that the cost of attempting the matriculation examination of Tshs 20,000 may be prohibitive to poor households. It is recommended that if the matriculation examination was to be retained in selection, the compensatory model should be used, and the battery should contain an aptitude test as opposed to syllabus based tests.
Description
Keywords
Ability, Testing, Higher education, Tanzania, University of Dar es Salaam
Citation
Lema, G. S. (2004). The differential predictive validities of University entrance examinations in Tanzania. Masters dissertation, University of Dar es Salaam. Available at (http://41.86.178.3/internetserver3.1.2/detail.aspx)