Challenges of student selection: Predicting academic performance

Abstract

Finding accurate predictors of tertiary academic performance, specifically for
disadvantaged students, is essential because of budget constraints and the need of
the labour market to address employment equity. Increased retention, throughput
and decreased dropout rates are vital. When making admission decisions, the
under preparedness of students necessitates that their potential cognitive abilities
should be assessed rather than their current abilities. In predicting their academic
performance, it is argued that conventional psychometric tests are less suitable for
the selection of students from disadvantaged backgrounds, because they are a
static measure of current abilities which gives no indication of the student's potential
to learn when in an optimum environment. The predictive validity of the Potential
Index Battery, the Learning Potential Computerised Adaptive Test and schoolleaving
results in selection, were determined by calculating the correlation of these
measures with academic performance over the full duration of the students' studies.
Statistically significant correlations were found, thus indicating that the learning
potential test had higher predictive powers than static measures of cognitive ability
and school-leaving results, in predicting future academic performance.

South African Journal of Higher Education Vol. 20(4) 2006: pp.547-562