An empirical study into the pass rates in KASNEB CPA examinations

Abstract

The study aimed at determining factors that influence performance in CPA examinations conducted by KASNEB. Two approaches were used: FIRST was to determine students characteristics that influence completion period of CPA examinations. A sample of 190 candidates who qualified was selected from the entire population of 1,865 candidates who ~ad qualified as at December, 1996. SECOND, an effort was made to determine factors that influence whether a candidate would pass the CPA section 6 examination or fail. A tota1 of 112 candidates who sat the December, 1995 and June, 1996 examinations and passed was analyzed alongside a sample of 146 candidates drawn from a total population of 1007 candidates who did not pass in the two sittings. On each of the approaches, Correlation, Multiple Regression, Stepwise Regress,ion",Multiple Discriminant and Stepwise Discriminant analyses were conducted. In both cases, mode of study, educational background and occupation were found to be very sig~ificant variables. Age appeared also but with the least significance. Analysis II identified Kenya College of Accountancy as a college that positively influenced passin~.- Analysis Igave emphasis on the manner of attempting the examination. Eng1ish and , mathematics at '0' level also showed a positive association with completion period. Regression analysis revealed that variables identified explained 80% of the reasons influencing completion period. Discriminant analysis showed that the same variables constituted over 90% of the discriminating attributes between those candidates who finish the course within a short time and those who take along time to comp1ete. Variables covered in influencing passing however could only apcount for 30% of the reasons for passing or not passing under regression analysis and had only 35% discriminative abi1ity between those candidates who fai1 and those who pass. This means that over 70% of the reasons for passing or not passing were not captured in this set of variables. Consequent 1y, there is need to search and estab1ish the factors that constitute the remaining 70%. In all the analyses, Regression and Discriminant Analyses strongly agreed on the findings.