Abstract
M. A.
The inequalities regarding opportunities for the various cultural groups in South Africa to obtain
access to tertiary education, could largely be attributed to the insufficient schooling that was available
to black scholars until now. A new political dispensation and the coupled drastic changes to the
educational system, including tertiary institutions, demands that the admission policies regulating the
access tot tertiary education be reviewed and changed to provide for the educationally disadvantaged
student. The Rand Afrikaans University created an alternative admission policy to accommodate
educationally disadvantaged students and or students who do not qualify in terms of the usual
admission criteria. Apart from certain minimum requirements, the policy requires that the learning and
developmental potential of the prospective students be established. Language proficiency,
intelligence, learning potential and other non-cognitive components are elements of the alternative
evaluation.
RAU found that the admission policy creates the necessity to research the predictive validity of the
cognitive part of the test battery, with regard to the criterium academic performance. This research
project attempted to assess whether the psychometric test battery which the University uses, within
the framework of an academic support programme called Project 100, can be viewed as valid in
predicting future academic performance of these students.
The battery includes the measurement of intelligence by means of the LSAT; learning potential as
assessed by the APIL; as well as language proficiency that is measured by using the
Leesbegripstoets and the Reading Comprehension Test. M-score was also included into the
statistical procedures, as to see whether the predictive validity of this factor is really as low as it is
generally assumed.
The statistical procedures that were employed to establish the predictive validity, included correlation
coefficients for determining the validity of the individual components. Multivariate regression analysis
was performed to determine the predictive validity of the components of the battery in conjunction
with each other. Using the technique of discriminant analysis, the success of grouping through the
test battery was assessed - looking at the proportion of the cases correctly classified into the
categories pass, fail, and no admission to exam; and the categories successful and unsuccessful.
The individual correlation coefficients revealed that each component correlated significantly with
academic performance. M-score also correlated with academic performance, but this coefficient was
the lowest of all the factors. The component which correlated highest with academic performance,
was language proficiency. The GSAT and the APIL correlated to the same extent with performance,
and the relationship between the two instruments was also relatively high. This implies that these two
instruments probably measures the same construct.
The results from the regression analysis proved M-score, APIL and the GSAT in combination to
explain more of the variance in academic performance than any of these factors individually. M-score
seems to explain other aspects of performance than intelligence and learning potential, and should for
this reason be reckoned with when assessing a prospective student's potential success.
The discriminant analysis showed that the tests successfully distinguish between the categories
successful and unsuccessful, but not between the three categories pass, fail and no admission to
exam.
The factors should however never be seen in isolation. No single factor can predict academic
performance to the full. The person should be seen as a whole, and the influence of non-cognitive
factors should always be kept in mind.