1. AIM AND OVERVIEW
The aim of the research was to determine with which variables and
what measure of success the academic achievement of black pupils
can be predicted in different school subjects.
To achieve this aim a theoretical study was firstly made of the
phenomenon, academic achievement prediction as it is manifested
in the educational context. It was indicated that the action to
predict is of supporting value to education, and that it assumes
a central position in the teaching-learning situation. In
addition it was indicated that academic achievement prediction is
a difficult undertaking, that it demands exceptional knowledge of
the teacher, and that it is essential that the teacher should be
able to predict as accurately as possible.
Secondly a study was made of the methodology of academic
achievement prediction. It was indicated that there are
different procedures for the initial selection of predictor
variables, procedures which are mainly variations of factor
analysis. Possible usages of moderating variables for the
grouping of pupils in homogeneous groups were also discussed and
procedures for the identification of the best predictors of
academic achievement were described.
In the third instance the academic achievement of black pupils
was investigated together with the factors by which it is
influenced. The indications are that the percentages of standard
ten passes are relatively low, and that there are, inter alia,
sex differences, language group differences and also urban/rural
differences in the achievement of pupils. Then the causes of
poor academic achievement by black pupils in the school and
family environment were considered, and also the extent of
educational and cultural deprivation. It was concluded that the
family environment does not always promote academic achievement
at school.
With regard to the variables which are specifically related to
the achievement of black pupils, a limited number of
research results are available. In spite of this, there are
indications that the pupil and milieu variables are related in
the same measure to the academic achievement of black pupils as
indicated by the general tendencies in research results. This
conclusion was arrived at with some reservation, because there
are certain indications of differences in the relative weighting
of variables pertaining to achievement. In addition it was
discovered that there are non-test factors which also
co-determine the achievement of black pupils in psychometric
tests. It transpired too, that further research is necessary as
regards the sources of the unexplained variance in the academic
achievement of black pupils.
In the fourth instance a study was mad0 of the relationship
between a specific selection of variables and the academic
achievement of black pupils. An image was obtained of the
relationship between ability, previous achievement, interest and
personality, and the academic achievement of black pupils, and
then some problem questions based on the relevant literature,
were formulated. The empirical research which followed,
attempted to find some of the answers to these questions.
2. AIM, PROCEDURE AND RESULTS OF THE EMPIRICAL INVESTIGATION.
The aim of the empirical investigation was to determine the
extent to which the standard ten achievement of black pupils can
be predicted with the aid of cognitive, non-cognitive and
biographical variables.
Four groups of hypotheses were drafted. The group-A hypotheses
investigated the contribution of the subsets of the best
predictors from all the independent variables to R2. The groups
B, C, and D hypotheses studied respectively, the contributions of
the cognitive, non-cognitive, and biographical variables to R2 in
the dependent variables.
In the research, the data was used of 871 subjects of the 19179
students who comprised the 1980 pupil population which took part
in the Guidance Project of the Department of Education and
Development Aid. Data was compiled of three psychometric tests,
the Academic Aptitude Test, the Vocational Interest
Questionnaire, and the High School Personality Questionnaire.
This data was subjected to factor analysis and nine factors were
identified.
To test the different groups of hypotheses, 81e technique of
multiple regression analysis was applied in two ways. Firstly,
the all possible subsets multiple regression method was used so
as to identify the subset of best predictors per dependent
variable, on the basis of the Cp criterion. The group-A
hypotheses were tested on this basis.
Secondly, the remaining hypotheses were tested by calculating
differences in R2 and testing their significance. In the case of
each dependent variable R2 1 was calculated for a basic set of
control variables which varied per group hypotheses. To this set
of control variables the cognitive predictors were added and R2 2
calculated. The difference between R2 2 and R2 1 then gives an
indication of the contribution which was made by the cognitive
variables to the prediction. The same procedure was followed to
calculate the contribution of the non-cognitive and of the
biographical variables to the prediction. The sets of control
variables varied in the different analyses. In each case the
educational significance of the contributions to R2 was
calculated with the aid of the formula for the calculation of
effect sizes as drafted by Cohen (1977). With the formula of
Cohen, a more precise indication was obtained of the educational
significance of the contributions to R2. The effect sizes
indicated whether contributions to R2 were of small, medium or
large educational significance.
On the basis of the results for multiple regression analyses,
selections were made of the biographical variables, and the
potential role of each as moderators of the prediction of
dependent variables were then investigated.
In the last instance a cross validation of results was made.
The results of the statistical analyses indicate in the first
instance that the subsets of best predictors of standard ten
achievement, consist of cognitive, non-cognitive and biographical
variables. The combinations of predictors vary from one school
subject to another, although certain trends were observed, like
for instance that the cognitive variables generally made the
largest contribution to R2 of all the predictors. It was also
found that the best subsets of predictors provide better
predictions for standard ten total, English, Afrikaans and
Biology, and weaker predictions for History and the Vernacular.
In the second instance the results indicate that the cognitive
variables, like aptitude and previous achievement separately and
together, made a statistical- and educational-significant
contribution to the prediction of achievement in the separate
standard ten subjects. These contributions were made in addition
to the non-cognitive and biographical variables which served as
control measures. The contributions of the cognitive variables
are larger in the case of standard ten total, Afrikaans and
English, and smaller in the case of the Vernacular, History and
Biology. It was also found that previous achievement always
rendered a larger contribution to the prediction, except in the
case of English and History.
In the third instance it was found that the non-cognitive
variables, interest and personality separately and together.
produce in isolated instances, significant contributions to the
prediction of the dependent variables. These contributions to
the prediction of dependent variables are produced in addition to
the control variables (cognitive and biographical variables), and
they are small and also (in terms of effect sizes), of small
educational significance.
In the fourth instance it was found that, sex in four cases,
living area (urban/rural) in four, language group 2
(Zulu/non-Zulu) in four, and language group 1 (Sotho/non-Sotho)
in six cases, made significant contributions in addition to the
cognitive and non-cognitive control variables) to the prediction
of standard ten achievement. These contributions vary from a
small to a large measure of educational significance. Together,
the variables contribute significantly to all the dependent
variables, and it was found that the contributions or the
biographical variables are largest with regard to History and
Biology. Finally, the results of the research on the effects of
moderator variables indicate chat sex has a modifying effect only
in the case of the Vernacular as regards the prediction. Living
area has no modifying effect and the effect of language group 1
is limited to History. Language group 2 is the variable which,
with regard to most dependent variables, has a moderating effect
on the prediction, as regards all the dependent variables except
the Vernacular.
In the validity study it was found that the degree of validity
was satisfactory, with the exception of the results relating to
History.
The most important conclusion made on the basis of the empirical
research is that the academic achievement of black pupils and the
inclusive variables can only be predicted with limited success.
school subjects vary a great deal in their predictability, and in
each case there is a large amount of unexplained variance of
which the source probably lies outside the scope of the variables
provided for in the research. A further conclusion is that the
variables, which together with others (and also individually)
make the greatest contribution to the prediction, are the
cognitive variables. The contribution of the non-cognitive
variables, interest and personality, is quite small, even smaller
than the indications in present literature. As a whole this is
in accordance with the results of other investigations in
developing countries. In conclusion, the inference was made as a
result of the moderator study that biographical variables are
indeed potentially serviceable as moderating variables and that
language group 2 (Zulu/non-Zulu) probably has the greatest
potential.
The results of this research have implications for teaching
practice as well as for further research. Practical
recommendations have been made with regard to the use of the
results for guidance purposes, and with a view to the
manipulation of variables for the improvement of academic
achievement. Recommendations have also been made for further
research on alternative variables as well as the non-test factors
which influence the academic achievement of black pupils. / Proefskrif (DEd)--PU vir CHO, 1987
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:nwu/oai:dspace.nwu.ac.za:10394/9406 |
| Date | January 1987 |
| Creators | Van der Westhuizen, Gert Johannes |
| Publisher | Potchefstroom University for Christian Higher Education |
| Source Sets | South African National ETD Portal |
| Language | other |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0045 seconds