Abstract
M.Ed. (Educational Psychology)
The picturesque landscape of South Africa is woven by the rich threads of diverse languages,
cultures and beliefs. Despite the inherent differences of the different groups in this country, a
common golden thread lies in the imparting of knowledge, values and beliefs from one
generation to the next. This is often done formally, in the school environment, where children
learn, develop and grow. The potential of children should be harnessed in an environment that is
conducive to maximising the strengths and minimising their weaknesses. One domain is in the
psychological testing arena where children’s cognitive abilities are assessed. The Junior South
African Individual Scale (JSAIS) as an assessment measure is currently being used by
psychologists despite the last norming having taken place in 1984 and the norms not being
representative of the South African population. In an endeavour to transition from the past
monolingual and monocultural practices, the JSAIS has been translated into isiZulu to eliminate
any possible language barriers to participants. However, it is evident that translating tests is a
risky procedure and every effort should be made to adhere to criteria proposed by the
International Test Commission (ITC) to minimise cultural bias. One such way, is to thoroughly
investigate the psychometric qualities of translated tests. The participants of the research are
children in the Grade 1 isiZulu class at school in Soweto where research is being conducted. This
longitudinal research project is a collaborative effort between the University of Basel,
Switzerland and the Centre for Education Practice Research (CEPR) at the University of
Johannesburg. The aim of this study was to apply a Rasch Analysis to the Numerical Intelligence
Quotient Eight (NUM Scale) of an isiZulu translation of the Junior South African Individual
Scales (JSAIS) to determine the quality of the items in relation to the ability of participants in the
sample group. As an exploratory, quantitative study, the findings suggest that the JSAIS has
some strong psychometric properties and that adaptation, translation and renorming of the test
should be conducted in future studies. The findings of this study show that the data generated
varied in its fit of the Rasch Model. While findings confirm that the numerical subtests of the
Number and Quantity are valid measures of the construct for which it was designed, it is
recommended that further studies be conducted into the equivalence of the isiZulu translated
JSAIS, the further testing of larger samples of participants and the reduction of cultural loading
on items. The information gleaned from this study regarding the misfitting of items, item difficulty and overlapping should also be considered to enhance the quality of this test and
thereby maintaining the tenets of our new, modern democracy in South Africa.