---

Author(s): Usoro Anthony E., Edeminam Desire Edeminam.

Volume/Issue: Volume 6 , Issue 2 (2023)

Abstract:

Every educational institution requires a sufficient number of qualified academic staff to deliver on the mandate, which includes, training, research and community development service. The quality of academic staff in a tertiary institution is expected to reflect on its graduates, who should compete favourably in the world labour market and add value to the society. The motivation behind this research was predicated upon the need to assess the productivity of academic staff in Akwa Ibom State University. The aim was to analyse academic staff profiles for possible reclassification on the basis of some performance factors. Information about the qualification, years of experience and research publications for 388 pensionable academic staff of the university was obtained from staff records. Firstly, goodness of fit tests for conformity of academic staff mix with the NUC proportional distributions of 20%, 35% and 45% for Professors/Associate Professors, Senior Lecturers and Lecturer1/Below categories were conducted. The tests results showed conformity of 26 out of 38 departments with the NUC proportional staff mix. 12 departments were affected with non-conformity with the NUC proportional academic staff mix. This is a challenge, not only to the 12 affected departments, but to the university as a whole, and this calls for concern. Secondly, Fisher’s and Bayesian Discrminant methods were adopted to analyse the staff profiles for possible reclassification. The analysis using Fisher’s method has revealed 100% correct classification of Professors/Associate Professors, 71% correct classification of Senior Lecturers, 68% correct classification of Lecturer1/Below and overall correct classification and misclassification probabilities as 0.71 and 0.29 respectively. Bayesian method has recorded 100% correct classification of Professors/Associate Professors, 61% correct classification of Senior Lecturers, 88% correct classification of Lecturer1/Below and overall correct classification and misclassification probabilities as 0.84 and 0.16 respectively. Comparing the two approaches, there is a higher value of correct classification probability in Bayesian Discriminant approach than Fisher’s approach, and a lower misclassification probability in Bayesian method than Fisher’s method. Bayesian approach gives more advantage in minimizing the classification error than the Fisher’s linear Discriminant method, and therefore, places Bayesian Discriminant Approach on higher comparative advantage than Fisher’s Discriminant method. The classification and misclassification probabilities presented in this paper are modifications of Usoro (2015). This paper recommends Bayesian Discriminant Analysis, especially, when carrying out discriminant analysis involving many groups or populations to avert the multiple pairwise Fisher’s Linear Discriminant analysis for multiple sample or population distributions. The outcome of this research is a good working instrument for staff assessment, planning and development of academic manpower in Akwa Ibom State University.

Keywords:

Academic Staff, NUC Staff-Mix, Discriminant Analysis

No. of Downloads: 0