Best and worst performing public schools in relation to poverty

The March map of the month presents the distribution of the best and worst performing public schools (based on the 2014 matric results) across Gauteng. The best performing schools are split into the top 30 plus schools overall (44 schools that achieved 100% pass rate) and the top 30 schools in township areas. Similarly, the worst performing schools are split into the bottom 30 schools overall and the bottom 30 schools in township areas. It is important to note that in some cases township schools in the overall categories (top and bottom) overlap with schools in the township categories (top and bottom).

The schools are overlaid on the South African Multidimensional Poverty Index (SAMPI) produced by Statistics South Africa (StatsSA) using census 2011 data. SAMPI (which is a measure of multidimensional poverty) is computed on the basis of a set of eleven indicators spanning across four equally weighted dimensions, as shown in Table 1 below. A household that falls short of the deprivation cut off is considered poor and, if the weighted average of indicators for the household totals 33.3% or more, the household is deemed multidimensional poor (SAMPI Poor).


Table 1

The SAMPI score as illustrated in the base map layer, is given as the product of the Headcount (H), which is the proportion of households defined as multidimensionally poor (i.e. breadth of poverty), and the Intensity (I), which is the average weighted score of multidimensionally poor households (i.e. depth of poverty) – SAMPI = H x I.

In this instance the unit of measurement is the ward and the higher the index, the poorer the ward. The SAMPI scores were split into 5 categories using natural breaks in the data, with the lowest category (0.000048 - 0.009174) representing the lowest rates of poverty and highest category (0.097677 - 0.219619) representing the highest rates of poverty per ward. Wards that have a SAMPI score of 0 were categorised as ‘not SAMPI poor’ (meaning they are found not to be multidimensionally poverty stricken wards). In addition, ward number 74804019 (in Merafong City local municipality), was excluded from the analysis due to missing data.

The results of the spatial overlay shows that top schools overall (the best schools in terms of pass rate) are located in wards with a lower SAMPI score (i.e. are less poor) than the bottom schools. As illustrated in Table 2 below, 7% of the best performing schools are located in wards that are not SAMPI poor and 73% of such schools are found in wards with SAMPI scores ranging between 0.000048 and 0.009174 (lowest). It should also be highlighted that 47% of the top performing schools in townships as well as 60% of the bottom schools in townships (worst performing schools in terms of pass rate) are located in wards with higher SAMPI scores ranging between 0.009175 and 0.097676.


Table 2

Interestingly enough, the mean SAMPI score of the top 30 schools in the townships is worse than the bottom 30 schools (see Table 3).


Table 3

Although the best township schools are located in wards that are multidimensionally poor, their academic performance was commendable, with pass rates ranging between 96.27% and 100% (Table 4). Furthermore, out of the 30 plus schools overall (44 schools that achieved 100% pass rate), 8 of them are located in townships. This may indicate that Gauteng’s education programmes in the township areas are having a significant impact. To better investigate this in future, the number of distinctions per learner per school may be a better measure than pass rate. However, the overall analysis shows there is no automatic or deterministic link between poverty and performance – but there clearly is a correlation in many instances, both for top and bottom schools.


Table 4


From the results of the spatial overlay analysis alone, there is no clear relationship between a school’s performance and the multidimensional poverty situation of the wards in which such schools are located. For example, a school may be located in a multidimensional poverty stricken ward while some of the learners may be residing in less poor areas. Hence, a study of the socio-economic profiles of learners, the catchment area of the school, as well as patterns of movement across wards to schools, may provide a better understanding of a school’s performance. This calls for deeper analysis to identify the exact factors that determine performance. Suggestions in terms of how the analysis needs to be reframed for a fair and objective comparison are as follows:

  1. The entire school population (770 schools) needs to be included in the analysis instead of only sampling the best and worst 30 schools.
  2. The classification of schools into quintiles determines the amount resources assigned to each school and this should be compared to SAMPI scores.
  3. The number of learners varies greatly by school and should be taken into account.
    • Schools are classified into districts, and analysis should be performed on a district level as well.
    • Distinctions per learner per school may be a better tool than pass rate, to assess the quality of the education provided.
    • The residential address of learners should be compared with the location of the school – are good schools attracting leaners from a wide area, and are under-performing schools failing to do so, or is the pattern more complex?



The GCRO sends out regular news to update subscribers on our research and events.