In 2025 South Africa had 13 596 869 learners in 24 766 ordinary schools, taught by 459 906 educators, and every one of those figures, with its provincial breakdown, is in a four-page Department of Basic Education publication that most education dissertations never cite. This guide lists the nine datasets an education master’s or honours student in South Africa can actually download, what each one holds, the headline figure each is known for, and the trap each one sets. Every figure carries its source and year in the row.
The nine datasets, and what each is for
| Dataset | Publisher | What it holds | Latest release | Where to get it |
|---|---|---|---|---|
| School Realities | Department of Basic Education, EMIS | Learners, educators and schools by province, sector, grade and school size; learner-educator ratios | 2025 edition, published February 2026 | DBE statistical publications page, free PDF |
| Master List of Schools | Department of Basic Education, EMIS | Every registered school with its sector, phase, quintile and district | Updated through the year | DBE EMIS downloads page, spreadsheet |
| National Senior Certificate reports | Department of Basic Education | Examination Report, School Performance Report and School Subject Report for each cohort | 2025 cohort, released 13 January 2026 | DBE NSC results page |
| PIRLS 2021 | IEA; South African study run by the University of Pretoria’s Centre for Evaluation and Assessment | Grade 4 reading achievement with learner, home, teacher and school questionnaires | 2021 cycle, results 2023 | IEA international results and data repository |
| TIMSS 2023 | IEA; South African study run by the Human Sciences Research Council | Mathematics and science achievement for Grade 5 and Grade 9 with context questionnaires | 2023 cycle, results December 2024 | IEA international results and data repository |
| General Household Survey | Statistics South Africa | School attendance, fees, feeding, transport, reasons for non-attendance, by household | 2025 wave on DataFirst | Stats SA and DataFirst, microdata |
| National Income Dynamics Study | SALDRU, University of Cape Town | Panel of the same households across five waves, 2008 to 2017, with child schooling outcomes | Wave 5, 2017 | DataFirst, microdata |
| Early Grade Reading Study I and II | Department of Basic Education; Taylor and Fleisch | Randomised trials of reading interventions with five waves of learner assessment | EGRS I 2015 to 2021; EGRS II 2017 to 2020 | DataFirst, public and secure versions |
| Thrive by Five Index | DataDrive2030 with the Department of Basic Education | Nationally representative assessment of children aged 50 to 59 months in early learning programmes | 2021 Index; 2024 Index published 2025 | thrivebyfive.co.za, reports and technical documents |
The figures education students most often need
| Figure | Value | Source and year |
|---|---|---|
| Learners in ordinary schools | 13 596 869, of whom 12 840 275 in public and 756 594 in independent schools | DBE, School Realities 2025 |
| Ordinary schools | 24 766: 22 259 public, 2 507 independent | DBE, School Realities 2025 |
| Educators | 459 906: 411 203 public, 48 703 independent | DBE, School Realities 2025 |
| Learner-educator ratio, public schools, all educators counted | 30,6 | DBE, School Realities 2025 |
| Largest provincial systems | KwaZulu-Natal 2 904 341 learners (21,4%); Gauteng 2 684 785 (19,7%) | DBE, School Realities 2025 |
| Smallest provincial system | Northern Cape 312 237 learners (2,3%) | DBE, School Realities 2025 |
| Independent schools by province | Gauteng 931, Western Cape 341 | DBE, School Realities 2025 |
| Share of learners in Grade 12 | 5,7%; Grade 10 is the largest grade at 9,2% | DBE, School Realities 2025 |
| Grade 4 reading achievement | 288, down from 320 in 2016; 81% of learners below the low international benchmark | IEA PIRLS 2021, South African report 2023 |
| Grade 9 mathematics and science | Mathematics 397 (up 8 on 2019); science 362 (down 8) | IEA TIMSS 2023 |
| Grade 5 mathematics | 362, down 12 on 2019 | IEA TIMSS 2023 |
| Children in early learning programmes on track for early learning | 45%; 28% falling far behind; weighted sample 5 139 children aged 50 to 59 months | Thrive by Five Index 2021 |
Two of those rows are the ones examiners see misquoted most. The PIRLS figure is a Grade 4 figure from 2021 fieldwork, published in 2023; writing “PIRLS 2023” signals that you read a headline rather than the report. The TIMSS figure is for Grade 9, not Grade 8: South Africa tests a year later than most countries to match its curriculum, and the international tables footnote this.
Which dataset answers which dissertation question
A study set in schools you will visit yourself. Use School Realities and the Master List for the background chapter and the sampling frame. The Master List gives you every school in a district with its quintile, which is how you justify a purposive sample of two quintile 1 schools and one quintile 5 school, and School Realities gives the provincial denominators. The permission chain for getting into those schools, from provincial department to governing body to parent consent, is in our guide to the methodology chapter of an education mini-dissertation.
A secondary analysis of learner achievement. PIRLS and TIMSS are the only nationally representative achievement datasets with public microdata, and both come with home, teacher and school questionnaires that let you model, say, the relationship between home language and reading achievement with the survey weights applied. A master’s dissertation can be built entirely on one of these files; the analysis is not trivial, because the achievement scores are plausible values that must be handled as a set, and the sampling design requires weighted estimation. Our guide to choosing a statistical test for a dissertation covers the design questions that come before the software.
A study of access, attendance or cost rather than achievement. The General Household Survey is the instrument. It asks every household about each child’s attendance, the fees paid, whether the school provides meals and transport, and why any child of school age is not attending. Stats SA publishes a schooling-focused report from it, and the microdata on DataFirst lets you cut it by province or income. How to register on DataFirst and pull a Stats SA file is in our step-by-step guide to getting South African data for a dissertation.
A question about change over time in the same children. Only the National Income Dynamics Study follows the same households, and it is the source for any claim that a child’s circumstance at one wave predicts an outcome at a later one. The last wave is 2017, which limits it for recent questions but not for the structural ones.
A question about whether an intervention works. The Early Grade Reading Study files are the rare South African case of randomised trials with learner-level assessment data released for reuse. If your dissertation evaluates a reading programme, the EGRS instruments and the published effect sizes are your benchmark.
A question about the years before Grade R. The Thrive by Five Index is the only national measurement of preschool outcomes; its 2021 round used a multistage cluster sample across all nine provinces and the 2024 round repeats it, which makes it the first source for early childhood development dissertations that need a national figure rather than a single centre.
Four traps that cost education students marks
Mixing levels. School Realities counts learners at the national and provincial level; it cannot tell you anything about one school. The Master List describes schools; it holds no learner outcomes. PIRLS describes learners, but only in the sampled schools, and it is not designed to produce a district estimate. State the level of every dataset you use and do not draw a conclusion at a level the data does not reach.
Confusing the snapshot date with the publication date. School Realities 2025 describes the 2025 school year and was published in February 2026; cite it as the Department of Basic Education, 2026, School Realities 2025. The same discipline applies to PIRLS (fieldwork 2021, report 2023) and to every NSC report, which describes the previous year’s cohort.
Reading the pass rate as the finding. The National Senior Certificate pass rate is the least useful number in the three reports the Department publishes each January. The School Subject Report gives subject-level distributions, and the School Performance Report gives school-level results, and a dissertation on mathematics outcomes in one district should be built on those, not on the headline.
Citing a dataset that has moved. The regional SACMEQ reading and mathematics surveys that older South African dissertations cite are still legitimate data, but the sacmeq.org domain no longer resolves to the consortium. Cite the archived reports through the IIEP or the Department of Basic Education, and check every data URL in your reference list the week before submission. The same habit of tracing each figure to a primary document, and saying when no dataset exists, is worked through for another field in our guide to using statistics in a background chapter.
How to cite them in Harvard
Each dataset is cited as a corporate author with the publication year, and the year of the data stays in the title. Reference list: Department of Basic Education (2026) School Realities 2025. Pretoria: Department of Basic Education. In text: (Department of Basic Education, 2026). For microdata, cite the depositor, the wave and the word dataset: Statistics South Africa (year of deposit) General Household Survey 2025 [dataset]. Cape Town: DataFirst. For PIRLS and TIMSS, cite the international report for the scores and the IEA data repository for the file you analysed, with the version number. UCT, Wits, Pretoria, Stellenbosch and Unisa each publish their own Harvard variant, so the punctuation follows your faculty guide; the rule that both the data year and the publication year appear does not change.
What no public dataset will give you
Say so rather than improvising. There is no public learner-level dataset linking a child’s Grade 4 reading score to a later NSC result; the systems are separate. There is no national teacher absenteeism dataset in the public domain. There is no national figure for the number of learners taught in their home language beyond the Foundation Phase that is more recent than the surveys above. And the Thrive by Five and PIRLS samples are representative nationally and, with caution, provincially, but not for a district or a municipality. A sentence that states a gap and cites the two datasets that come closest earns more from an examiner than a figure of unknown origin.
If your question is about post-school education rather than schooling, the sources change entirely: enrolment, graduation and staffing come from the Department of Higher Education and Training’s HEMIS system, with different definitions, and we set them out in our postgraduate enrolment statistics guide.
Getting the background chapter written from the data
The background chapter of an education dissertation is largely these figures arranged into an argument: the size of the system, the provincial variation, the achievement evidence, the gap your study addresses. Draft it with the source and year attached to each figure as you write, rather than adding references afterwards, because that is where the publication-date errors above creep in. Tesify keeps the figure, its source and the citation together as you draft, and formats the reference in your faculty’s Harvard variant, which for a chapter that cites the same Department publication fifteen times is the difference between a clean reference list and an afternoon of corrections.
Draft your education background chapter with the sources attached
Frequently asked questions
How many learners are there in South African schools?
13 596 869 in ordinary public and independent schools in 2025, according to the Department of Basic Education’s School Realities 2025, published in February 2026. Public schools held 12 840 275 of them.
What was South Africa’s PIRLS 2021 score?
288 on the PIRLS scale for Grade 4 reading, down from 320 in 2016, with 81% of learners unable to reach the low international benchmark. The data was collected in 2021 and the results published in 2023.
What were South Africa’s TIMSS 2023 results?
Grade 9 mathematics 397 and science 362; Grade 5 mathematics 362. South Africa tests Grade 5 and Grade 9 rather than the Grade 4 and Grade 8 most countries test, and the IEA tables footnote this.
Can I use PIRLS or TIMSS microdata for a master’s dissertation?
Yes. The IEA releases the learner, home, teacher and school files publicly. You need to use the sampling weights and treat the achievement scores as plausible values, and your methodology chapter must explain both.
Where do I get school-level data such as quintile and district?
From the Department of Basic Education’s Master List of Schools on its EMIS downloads page, which lists every registered school with its sector, phase, quintile and district.
Is there data on children before they start school?
The Thrive by Five Index measures children aged 50 to 59 months in early learning programmes nationally. The 2021 round found 45% on track for early learning and 28% falling far behind, on a weighted sample of 5 139 children.
Which dataset should I cite for school attendance and fees?
The General Household Survey from Statistics South Africa, which asks every household about each child’s attendance, fees, meals and transport. The microdata for 2022 to 2025 is on DataFirst.