Datasets offer us knowledge about the world: they are things we know or assume about a phenomenon that allow us to make reasoned conclusions. In the absence of accurate data, the ability to make evidence-based decisions becomes tricky — whether in business, politics or education.

This logic gets to the heart of the Zero Dropout Campaign’s call for the basic education sector to improve learner-level data collection and monitoring. Why? Because high-quality administrative data can be used to track the overall health and functioning of our schooling system, learner progress, grade repetition and dropout.

But this is just the start: capturing and monitoring the right type of data can also help schools to flag learners who are at risk of dropping out based on indicators of disengagement. When used in this way, datasets form part of an early warning system (EWS) used to anticipate and prevent dropout. Crucially, datasets do not simply present us with a story but can help us to change the way the story ends — potentially for the better.

Understanding individual drivers of dropout

Eight-year-old Luyanda Khofu lives with his parents in an informal settlement in Kya Sand, north of Johannesburg. By October 2021, this little boy had been out of school for two years — not by choice but due to circumstances beyond his control. His parent’s dire financial situation compounded by the unfortunate loss of his birth certificate meant that Luyanda could not be registered at a no-fee public school. Instead of spending his days in the classroom, Luyanda often accompanied his mother, Elizabeth, to a busy intersection where she would ask passing motorists for help to care for her family.

Like Luyanda, many learners from under-resourced communities face myriad challenges as they move from childhood to adulthood. In South Africa, more than 60% of children are poor, or “multi-dimensionally” poor. This term is used by economists and researchers to describe a child that is deprived of most of the dimensions that are used to determine “wellbeing”. The cumulative stress of growing up in poverty can have deep and lasting effects on children’s capacity to learn, especially if they are not properly supported to deal with the challenges in their lives.

A 2011 study by Cornell University demonstrates the lifelong impact of growing up in poverty, using the “Risk-Stress” model to account for low achievement and income. Although the findings may seem obvious, this is a ground-breaking study that brings the inner workings of a child’s mind, and their state of mental wellness, front and centre when it comes to thinking about learning achievement at scale.

Economic exclusion, exposure to violence and poor nutrition are a few of the potential contextual challenges of growing up in poverty. To make matters worse, the pandemic has deepened inequality and financial hardships for struggling households. In March 2021, 400,000 children were living in homes that reported daily hunger, and three million had reported that their homes had no food in the seven days prior to being surveyed.

Learners with poorer access to resources, support networks and opportunities are less likely to withstand disruptions to their education This is evident in pandemic-era dropout figures: the highest rates are among those from the poorest households in rural areas. It is in understanding these contexts that we can begin to develop a body of evidence to provide the right type of support.

Often, the decision to drop out comes at the end of a long process of disengagement in which a learner is pushed or pulled away from school due to factors at home, at school and in their communities. Disengagement — or gradually disconnecting from schooling and learning — takes place within a complex web of factors linked to psychosocial issues and academic performance. High levels of grade repetition in lower quintile schools (situated in the most under-resourced communities) compound — and are symptoms of — learner disengagement.

A new report published by Resep (Research on Socio-Economic Policy) shows that learners are getting stuck in cycles of repetition that peak in Grade 10 — and only a fraction make it to matric. Consequently, around 20% of learners in Grades 10-12 are three years or more over-age. The report concludes that most learners in lower quintile schools are not acquiring the knowledge and skills they need to progress through school in the allotted time frame — reflecting an “inefficient” and costly system. The cost of having repeaters in the system could be as high as R29-billion annually.

These figures tell us that our schooling system is not functioning optimally. Even though the South African government is spending a proportionally high percentage of GDP on basic education, similar to that of Norway (and well above the OECD average), financial investment has not resulted in improved academic outcomes.

To better understand why there hasn’t been a drastic improvement in learning outcomes despite significant spending, we need to consider learners from the individual perspective. Learner-level data can help educators to better understand why they are getting stuck and what can be done to steady them along.

Using data to build early warning systems

In South Africa, the New Leaders Foundation (NLF) is pioneering the use of education data to improve learner outcomes with its Data-Driven Districts (DDD) programme in roughly 40 school districts. Given the foundation’s expertise in data collection and monitoring, the Zero Dropout Campaign enlisted the NLF’s help to develop data tools for schools as part of a pilot project a few years ago. The learnings surfaced during the pilot have been condensed into best practice toolkits available to all public schools in the country. The toolkits (which include an early warning system) were launched at the inaugural Zero Dropout Action Summit in September 2021.

The early warning system aims to help educators identify the red flags of disengagement, allowing them to call in the right type of support at the right time. The information collected by the EWS tool is based on key measurable observations that children in distress frequently display. The tool includes six key dimensions: they are like a “menu” of what one might think are the key problems faced by a classroom of learners. The first three, known as the ABCs — Academic Performance, Behavioural Issues and Chronic Absenteeism — are essential, as global research has shown them to be key indicators of disengagement and dropout.

If educators are on the lookout for key behaviours, and identify what forces are at work which are reducing the child’s ability to focus and feel motivated in school, then they can intervene with much-needed support and care. Perhaps the learner requires attention from a social worker, counsellor, or referral to a clinic. Perhaps they require a focused learning plan. This tool helps educators to plan, design and deliver interventions suited to the individual needs of their learners.

Big picture data

Learner-level data can help educators to take pre-emptive measures to prevent dropout, and “big picture” data at a country level can inform better policy decisions. But, as things stand, there are gaps and inaccuracies in existing datasets that make it difficult to track learners’ pathways through school.

Discrepancies in learner-level data are due to systems and human resource challenges at schools — complicated by the administrative burden of using different electronic and paper-based data monitoring systems. For instance, it’s not unheard-of for a school’s paper-based attendance register to differ from the figures that are later captured in an electronic data monitoring system used by a specific province or nationally. This tells us that capturing and generating high-quality data is largely dependent on the type of systems, processes, structures and human capacity in place at school level.

More than 18 months after Covid-19 reached South Africa’s shores, we still do not have an accurate account of the pandemic’s impact on dropout in the country. This would require accurate and complete datasets that track learners’ pathways through school, as well as a standardised definition of dropout that is universally applied. We do not have either of these requirements.

Before the pandemic, a longitudinal study found that 40% of learners who started Grade 1 would drop out before completing matric. But lockdowns and intermittent school closures have disrupted learners’ lives in different ways, amplifying the factors that typically lead to disengagement and dropout.

In July 2021, Nids-Cram researchers released a finding that shocked the basic education sector, reporting that 500,000 learners had not returned to school compared to pre-pandemic figures. However, a new report by the Department of Basic Education that analyses school administration data paints a more conservative picture, reflecting that 19,000 learners were no longer in the system and that 2021 saw 27,000 fewer first-time school-goers enrolled compared to previous years.

Despite discrepancies in figures (partly due to differences in datasets and methodology), these reports show that school attendance and learning are seriously affected by the pandemic and that the only way the sector can adequately respond is to have access to timeous and accurate data. Current indications are that the ripple effect of disrupted learning and poor attendance will be felt over the next 10 years when children reach the FET band of schooling (i.e. Grades 10-12).

We need to take urgent steps now to steady us into the future, starting with collecting the right type of data about our learners. There are no silver bullets or simple solutions to complex problems, but by empowering as many role-players with the right tools and information to call on the state for more of the right support, we can begin to create the kind of critical mass needed to meet the real needs of South Africa’s children. DM

Rahima Essop is Head of Communications and Advocacy for the Zero Dropout Campaign (ZDC); Merle Mansfield is Programme Director of ZDC; and Angela Biden is an independent researcher and development specialist.