An Evaluation Of The Validity Of Discriminant Analysis: The Case Of MBA Projects, Faculty Of Commerce, University Of Nairobi

Abstract

The primary objective of this study was to evaluate the validity of the discriminant method and results as used in Master of Business and Administration (M.B.A.) projects in the University of Nairobi. This was done by testing whether' data used in the analysis met the underlying assumptions of discriminant analysis, testing the significance and stability of the coefficients and lastly by determining the error of classification. Two M.B.A. projects were analyzed, one written by Chirchir (1989) and the other by Nderitu (1989) For Chirchir's project both assumptions for multivariate normality and equality of group dispersion held true. It was also found out that the error of classification was quite high. It was also found that Economics, Fundamentals of Accounting and Quantitative Methods were significant. In the case of Nderitu's project, the variables used were dummy variables and therefore the two most important assumptions did not hold true. In fact in her project she agrees that her model lacked in validity. For dummy variables it would have been more appropriate to develop a quadratic discriminant function (Dillon 1979). Therefore, given the importance of her study one could replicate it and this time develop a quadratic discriminant function. Both studies are quite applicable in the University of Nairobi. Chirchir's findings could be used to guide students on option selection. If Nderitu's study could be improVed on then it could also be useful to the M.B.A. admission board in selection of students.