ReSEP hosts its third Quantitative Applications in Education Research Conference, 28 & 29 September 2017

ReSEP held its third annual “Quantitative Applications in Education Research” conference at STIAS in Stellenbosch from 28 – 29 September 2017.  A total of 110 participants from a range of backgrounds,  including education researchers, policy-makers and PhD students attended the conference.

During her video address, Minister of Basic Education, Mrs Angie Motshekga, welcomed participants to the conference and expressed her support for the ongoing research conducted within the ReSEP group:

I want to acknowledge the invaluable role played by the Research on Socio-Economic Policy (ReSEP) team in producing such rigorous research and the important links they maintain with the Department of Basic Education (DBE). We hope that this partnership continues in years to come, and is also extended to previously disadvantaged universities in order to increase the production of high quality, policy relevant research within the sector”.

Professor Servaas van der Berg gave the first presentation of the conference titled “How we’ve progressed” where he emphasized the improvements in TIMSS between 2003 and 2015, and also the large increase in the numbers of black matriculants receiving high-level passes in mathematics and science.

Two international keynote speakers, namely David Evans from the World Bank and Yuri Belfali from the OECD, respectively presented on  “Getting the most out of our teachers: Lessons from recent quantitative research” and “International assessment for excellence and equity: Experiences from PISA for Development”.

Both presentations were well received by the audience eliciting a number of questions about the role of international assessments in education policy making and the politics of teacher reform in developing countries.

The first day of the conference ended with a panel discussion on “The practice of improvement:  Getting from here to there” which was chaired by ReSEP’s Dr Nic Spaull and included panellists Prof Jonathan Jansen (UFS), Prof Brahm Fleisch (WITS), Prof Peliwe Lolwana (WITS) and Dr Itumeleng Molale.

The second day of the conference also included two parallel sessions where PhD students were given the opportunity to present their PhD proposals and ongoing research and gain valuable feedback from the conference participants.

The full program with links to all presentations can be found below:

28 September (Day 1)

29 September (Day 2)

One of the keynote speakers at the conference, David Evans, wrote a blog post summarising the conference.  The original can be found here.

For any further information about the conference, please email

SA data shows large and unexpected increase in births around 2004

This article originally appeared on the Business Day website as a partially edited version, which can be found here. For the original article, please read below.

By Martin Gustafsson

It is clear that something remarkable happened to South Africa’s births around 2004. What led to this discovery reveals important improvements to ways social data have become available in South Africa.

Starting in 2011, education analysts started noticing sharp increases in the number of children entering primary school, after at least five years of steady declines. Between 2010 and 2013, Grade 1 enrolments rose by 13%, or just over 100 000 children. These figures refer to all schools, whether public or independent. The wave moved its way up the primary grades, reaching Grade 7, the last primary grade, in the current year, 2017. Class sizes increased, resources became a little more stretched.

The increases confounded planners. Concerns were raised about the possibility that schools and provinces were creating ‘ghost learners’ to attract more funding. But why would this happen across virtually the entire country at the same time?

Had there not been a second, and completely separate data source to confirm what was seen in the enrolment data, we may still have been debating whether the enrolment trends were real or not.

The second data source was the Home Affairs birth registrations data, which had recently become available to researchers through the DataFirst web portal hosted by the University of Cape Town.

The Home Affairs data revealed a surge in births, starting in 2004, which corresponded very closely with the subsequent rise in enrolments seen in schools. This can be seen in the following graph, which uses only publicly available data. These patterns are confirmed by the age-specific enrolment data of the Department of Basic Education, nonpublic data which analysts in the Department have been looking at. In the graph, Grade 1 enrolment in 2012 is assumed to represent births in 2005, and so on for the other birth cohorts.

births and enrolments

Importantly, births declined in 2009, but did not return to their pre-2004 levels. The fact that the Grade 1 curve in the graph is on average 16% higher than the births curve is attributable to grade repetition, which means that a proportion of children remain in Grade 1 for more than year. This proportion has remained fairly stable in recent years. The important thing is that the two curves display the same shape.

The surge in births was remarkably widespread across the country. All provinces were affected to some extent, though most affected were Limpopo and Mpumalanga. The map provides details at the district municipality level (administrative units available in the Home Affairs data).

birth registrations 2003 to 2006

The approximately 15% increase in births between 2003 and 2006 is remarkable by international standards. Only a couple of countries have seen comparable increases. For instance, Ukraine did in the early 2000s following government efforts to boost an exceptionally low birth rate.

Are our increases in the number of children reflected in the official mid-year population estimates of Statistics South Africa? Yes and no. Historical figures published with the mid-year estimates some years ago, for instance 2013, offered no hint that there had been a surge in births around 2004. However, the most recent Stats SA estimates do point to increases, but of only around half the size seen in the enrolments and birth registrations data. Stats SA’s 2016 estimates of past fertility rates reflect, for the first time, a rise and fall in births, with a peak in 2008 (as in the above graph).

One could say there has been a lag of as long as a decade between a major demographic phenomenon and its registration in the official population data, and even then it seems only partially reflected. By developing country standards, this is not unusual. This situation could be improved, in part if demographers and education planners compared notes more frequently. This kind of comparison hardly ever occurs, despite the fact that school systems tend to collect data on child populations with much greater frequency than national statistical agencies. A part of the problem is that pressure to standardise population estimation methods across countries, in line with the prescripts of UN bodies, can detract from the monitoring of local peculiarities and shocks.

The reasons behind the increase in births is not completely clear, but is being investigated. The patterns in the data suggest that the child support grant was not a major contributing factor. Nor does immigration into the country explain the phenomenon. Instead, it seems as if greater access to anti-retroviral treatment, and the associated reduction in the risk of mother-to-child transmission, brought about a sharp increase in planned pregnancies. However, conclusive evidence on causes is still to come.

Analysis of the kind just described has become increasingly possible, for researchers and students, in part thanks to the greater accessibility of data. In particular, the availability of multiple sources allows for important comparisons and verifications to occur. The United Nations Data Revolution Group refers to data as ‘the lifeblood of decision-making and the raw material for accountability’. Whilst data on its own cannot resolve the range of policy conundrums we face in South Africa, it can be hugely beneficial in bringing about a more informed debate. However, capacity to use data, in particular microdata (or ‘raw data’) in government, our think-tanks and university departments is weaker than it should be, given the complexity of the policy challenges we face, and the data that are available, and will increasingly become available.

RESEP launches its synthesis report entitled “A Society Divided: How Unequal Education Quality Limits Social Mobility in South Africa”

On Friday, 24 March 2017, RESEP launched its synthesis report entitled “A Society Divided: How Unequal Education Quality Limits Social Mobility in South Africa”. The research project was headed up by Prof Servaas van der Berg and the report incorporates the research of about 20 authors. The central focus of the report is the role of education in promoting social mobility for the poor in the highly unequal South African economic landscape. The investigation is of particular relevance in a country where the rapid expansion of educational attainment since the 1970s has not produced the desired labour market outcomes for many South Africans, for the most part perpetuating patterns of poverty and inequality along the apartheid dimensions of race and geography.

Given the deep structural nature of inequality in South Africa, this report employs a conceptual framework (shown Figure E1) to illustrate how differences in education quality offered to South African learners are at the root of income inequality that persists two decades into democracy. The grim labour market prospects facing South Africa’s young adults are in large part attributable to an education system that still manages to produce vastly different education outcomes that favour a small elite in the wealthy part of that system and disadvantage mainly black and coloured learners in the less affluent part of the system.

A small minority of learners attend functional, high quality (mostly former white) schools, staffed by qualified teachers and characterised by good management, assessment and parental involvement. Learners graduating from these schools have relatively good chances of entering the upper end of the labour market, often (but not always) first acquiring some form of tertiary education. The high productivity jobs in this part of the labour market offer high returns. Traditionally this part of the labour market has been dominated by whites, but the removal of apartheid era restrictions, government interventions (such as black economic empowerment and affirmative action) and improved access to better quality education for some black children have allowed a relatively small black minority to achieve upward social mobility through the labour market.

In contrast, the majority of South Africa’s (mostly black) learners attend formerly black schools. These schools, that often also suffer from poor management, little parental participation and poor assessment, produce poor cognitive outcomes, which are poorly rewarded in the labour market, resulting in low employment probabilities and low wages from unskilled occupations. While the transition from low quality schools to low productivity jobs is relatively deterministic, it is possible for individuals from this part of the education system to access the high productivity part of the labour market through vocational training, affirmative action or other forms of labour market mobility.

This conceptual framework is used throughout the report to discuss how education, and particularly education quality, are critical inputs in advancing social mobility for South Africa’s economically vulnerable citizens.

Main findings

  • Performance on international and national standardised tests show that while educational attainment has converged dramatically over time between races, poor schools still lag far behind their affluent counterparts in learning outcomes. By Grade 9, learners in poor (mostly black) schools, have a backlog of about 3½ years relative to their rich school counterparts.
  • Substantial learning gaps between learners in different schools can be observed as early as the middle primary school years, making a strong case for decisive early interventions. As early as Grade 4, fewer than 30% of learners in the poorest 40 percent of schools are performing above international low learning benchmarks.
  • Education is an important predictor of labour market outcomes, with education beyond matric improving prospects both for employment and higher salaries. In 2007 the wage per hour of someone who had achieved a degree was three times as large as for someone who had achieved only a matriculation.
  • New evidence suggests that school education quality is also strongly positively associated with future earnings. Therefore, learners who attend poor quality schools generally earn substantially less than those who attend good quality schools, even when they have the same education levels.
  • The consequences of unequal education opportunities are particularly dire for many of South Africa’s black youth, who despite having more education than previous generations and no longer facing discriminatory labour market legislation, have no better employment probabilities than older labour market participants. Thus, despite more education, young black South Africans are less optimistic about their future.

Thus far, the report has featured in the Mail & Guardian as well as in News24.

The report can be downloaded in PDF format, here.

ReSEP report on flows from matric to, and through, university enriching debate around student fees

Recent research looking at “higher education access and outcomes for the 2008 national matric cohort”, by Hendrik van Broekhuizen, Servaas van der Berg and Heleen Hofmeyr of Stellenbosch University are colouring the debate around student fees in South Africa with in-depth analysis that are attracting welcome interest.

The Mercury recently reported on the findings of the research outputs of the ReSEP authors, and highlighted the much more pervasive extent of the challenges in resolving the student fees crisis.

Looking at matric exam data from 2008 to 2013, data from South African universities from 2009 to 2014, data from the Educational Management Information System master-list and data from the 2011 national census, the authors conducted an investigation of university access in South Africa.

Their findings included that, for those who started school, only:

  • 60% wrote matric.
  • 37% passed matric.
  • 12% gained access to university within six years of finishing school.
  • 6% completed an undergraduate qualification within six years of finishing school.
  • 4% completed a degree within six years of finishing school.

To read the full story and what underlying factors are contributing to the challenges of basic education in South Africa to contribute to the resolution of the student fees crisis, please click here to go to the IOL website, and read the original article as published in The Mercury.

New data of challenges in tertiary education presented at LMIP conference

Hendrik van Broekhuizen, Servaas van der Berg and Heleen Hofmeyr recently published a report on higher education access and outcomes for the 2008 national matric cohort. The study makes use of a unique dataset that combines data at the school and university level in order to track the transition of the 2008 matric cohort from school to university. This enables analysis of the relationship between school-level factors such as matric performance and university outcomes such as access, completion and dropout. Servaas van der Berg presented the main findings of the report at the Research Colloquium on Skills Planning on the 29th of September in Johannesburg. The colloquium was hosted by the Labour Market Intelligence Partnership and the Department of Higher Education and Training.

The main findings of the report are as follows:

• Approximately one-third of learners who write the matric examinations and obtain Bachelor’s passes never go to university.

• Amongst matriculants that obtained Bachelor’s passes, university access is not skewed against black students or students from poorer schools.

• A large proportion of matriculants who do go to university do not enter in the year following matric, but only one or more years later.

• Matric marks are a good indicator of university access, but only weakly related to eventual university completion rates.

• There are large differences across universities in the average matric performance of students who attend these universities.

• It takes a long time for many students to successfully obtain university qualifications.

• Dropout rates at university, although high, are not as high as often reported, because many students that are considered “drop-outs” from university in official statistics did not leave the university system, but changed their degree programme, switched from a degree to a diploma or certificate programme, or enrolled in a different university.

Prof Van der Berg’s presentation highlighted the extremely small proportion (14%) of learners who start school who end up with a Bachelor’s matric pass. 12% achieve Diploma passes. This implies only 26% of learners who start school will be eligible for enrolment in university. Delayed entry to university and low completion rates result in only 6% of the original group who started school obtaining some kind of undergraduate qualification within six years of completing matric. Prof Van der Berg also pointed out the importance of matric performance in ‘gateway subjects’ such as Mathematics, Physical Sciences and Mathematical Literacy in university completion rates.

These findings suggest that many patterns of university access and success are influenced strongly by school results. The weak school system has a major influence on who reaches matric and how they perform in matric. This, and particularly the achievement of Bachelor’s passes, explains much of the racial differences in access to university.