Abstract
D.Litt. et Phil. (Psychology)
The generalizability of scores obtained on the General Scholastic
Aptitude Test (GSAT) was investigated for a number of subpopulations.
It was shown that acceptable procedures for limiting bias against certain
subpopulations had been implemented in the course of test development.
In order to facilitate an empirical investigation of the GSAT,
random samples of Afrikaans-speaking whites, English-speaking whites,
Afrikaans-speaking coloureds, English-speaking coloureds and English Speaking
Indians were tested. The nature and intensity of cognitive
stimulation as well as the direction of intellectual development differed
from group to group and had definite implications for achievement
in a test such as the GSAT. All the groups contained persons who could
be regarded as fully integrated into an advanced technological society,
but their average level of development, measured against this criterion,
differed considerably. The test displayed a one-factor structure with
respect to all the groups. On the basis of the content of the subtests
this factor can be equated with Spearman's g factor. The ecological validity
of test scores was studied on the basis of Berry's model for the
generalization of behaviour aeross various environmental contexts. The
nature of the differences between the groups with respect to both the
internal and external structure of the test scores was investigated by
means of conceptual as well as statistical analyses. The extent to
which test scores could be generalized was indicated for different combinations
of groups. On the basis of regression models it was shown
that could to a large extent be used as a substitute for group membership
as regards the explanation of GSAT scores. Differences in functioning
found between the groups could often be attributed to
differences in levels of achievement. Within Berry's experimental context
the test was found to be almost equally valid for all the groups
and it measured developed general scholastic ability equally well. With
respect to most of the pupils in the white groups, the Indian group and
the English-speaking coloured group, the test results could be generalized
similarly in the behaviour context. The test measured general reasoning
ability with scholastic symbol material almost equally well for these four groups, so that cross-cultural comparisons witp respect to
this type of reasoning ability could be made reasonably reliably. However,
most of the Afrikaans-speaking coloured subjects were probably
handicapped by a lack of relevant experience. There are substantial
differences between the groups as regards the generalizability of scores
to intellectual achievements in the ecological context. In respect of
the majority of white pupils such generalization is probably acceptable,
as they could be regarded as fully integrated into an advanced technological
society.