Statistical modelling of the relationship between learner support intervention and matric pass rates in Letlhabile area, North-West Province of South Africa

Abstract

This study used statistical methods to determine the relationship between matric pass rates and interventions in the schools in the Letlhabile area. Nine schools were identified in this area. Five years (2007 to 2011) were looked at. Regressions methods were tried in which various forms of relationships were compared. The methods were linear, curvilinear (quadratic to polynomial of 4th power), exponential and power regressions were used in the tentative models investigated. The measures of bias and precision were used to compare the models. Multicollinearity was also investigated where it was possible. Time series analysis was used to illustrate the trend patterns of the pass rates in the various schools as well as the pattern of the numbers of interventions. In most of these schools the numbers of interventions increased over the five years and in only a few schools the number of interventions decreased over these years. . A highlight of this study is that more interventions enhanced the matric pass rates. It was also evident that in the years in which the interventions decreased, the pass rates also de.creased. The regression methods investigated were all showing to be applicable in the prediction of pass rates from the numbers of interventions. This was concluded from realising that thelneasures of bias, precision and quality all ratified them. The methods were compared in order to select the best one based on the measures. The linear regression in which the pass rates.areregressed on the number of interventions came out as the leading model in terms of all the criteria used.