Validation of a selection and classification model: a case study of the faculty of education Kenyatta University

Abstract

This study investigates the efficiency in selection and classification. Selection is a decision making process to determine who gains admission to a course of study. Classification is done to assign applicants to different courses of study. To aid in the decision making strategy a model is used to combine predictor variables. These predictor for public university education in Kenya was considered.

Because of the stiff competition for places, the grades obtained in the university selection examination, the Kenya Certificate of Secondary Education (KCSE), were used in a conjunctive model. This model was utilised in two stages, the first is the screening for university admission and the second is classification.

The sample consisted of the Year 2000 graduating cohort of Bachelor of Education Arts and Science students at the Kenyatta University (n=320 and n=206 respectively). The study was carried out retrospectively, where performance in the university was traced back to the selection examination, the Kenya Certificate of Secondary Education (KCSE), done in 1994. The criterion variable was University performance, which is the average grade obtained after four years of study at the university. The predictors were the KCSE Aggregate Grade, used to set a cut off necessary for admission to university, and the KCSE Cluster Grade used in assigning qualified applicants to different courses.

The question of whether the KCSE English Grade can add to the prediction efficiency was explored. The influences of some non-academic variables, namely gender and high school background of a student on University Performance were also investigated.

Data was obtained from the offices of the Joint Admission Board (JAB) at the University of Nairobi and at the Kenyatta University. The data for two groups of students. Arts (BEd) and Science (BEd) were independently analyzed. Correlation analyses was carried out to explore inherent relationships between various variables. Multiple Correlations was carried out to determine the degree of relationship between various combinations of predictors with University Performance. One way Analysis of Variance (ANOVA) was used to determine the influence of gender and high school background on University Performance. The significance of results was tested atµ=05.

For both the Arts (BEd) and Science (BEd) sub samples, it was found that the KCSE Aggregate Grade had low but significant correlations (.22) with University Performance. The combination of the KCSE Aggregate Grade with the KCSE Cluster Grade yielded a multiple correlation coefficient (R) of 23, in both sub samples. Addition of the Arts (BED) sub sample, but lowered R to 22 for the Science sub sample. The most plausible explanation of these modest gains in R was the effects of restriction of range, a problem in any selection study, and the high and significant inter-correlations among the KCSE grades.

The KCSE Aggregate Grade was shown to contribute more to University performance in comparison to the other predictors. This provided support for its continued use in setting a cut-off score for university admission. The KCSE Cluster Grade had lower correlations with University Performance. Support for its continued use in classification arose from the fact that, apart from restriction of range, the high rate of inter-faculty transfers could have compromised the selection standards. Therefore the KCSE Cluster Grade remains the best option in classification due to ease of administration at no extra cost in the selection process.

In both sub samples female students performed better at the university than their male counterparts. Further analysis revealed that there ware no significant differences in university performance when students were grouped by gender. This implied that affirmative measures put in place by the JAB in an effort to bolster female populations were justified, as lowering of cut-off grade for female students did not compromise standards at the university.

Students with a Provincial Schools backgrounds performed better than their counterparts in the Arts (BEd), while those with a District Schools backgrounds performed better than their counterparts in the Science (BEd) course respectively, even though not significantly.

It was felt that there was nee to provide effective career guidance services in all schools to help students in making career choices, using some of relevant findings of this study. There is also need to improve record keeping at the university so as to keep track of student performance through the whole course of study. There is also need for further investigation on the effect of course choices on university performance