Many children are in school for years before they gain the foundational reading skills they need to keep learning.
How long that takes — and for which children — was the central question behind a UNICEF education data project I co-led. The analysis covers roughly 45 countries and territories surveyed through MICS6, the sixth round of the Multiple Indicator Cluster Surveys. The survey directly assesses whether children can read and understand a short story — a clearer measure of learning than enrolment alone.
We called it the Foundational Learning Gradient: a reproducible analysis that tracks the share of children with foundational reading skills as they progress through school. The goal was to show where learning starts, where it accelerates, and where inequality persists.
What the gradient shows
The first pattern is stark. In MICS6 Africa countries, foundational reading starts near zero in the early grades. It improves as children move through school, but the gains come late. Children in MICS6 countries outside Africa start higher and stay ahead across all grades.
This matters because Grade 3 is widely treated as a turning point. By that stage, children are expected to have the reading skills that let them engage with the rest of the curriculum. When most children have not reached that point, the problem is not just low learning. It is delayed learning.
That delay has consequences. Children who do not gain reading skills early are less prepared for what follows. They may stay enrolled, but they are asked to progress through school without the basic tool that unlocks the rest of learning. Enrolment, on its own, cannot show this. A learning gradient can.
Delay and inequality together
The second pattern: learning is not only delayed. It is unequal.
Across school stages, children from richer households are more likely to have foundational reading skills than children from poorer households. This wealth gap appears in both MICS6 Africa countries and countries outside Africa, but the levels differ sharply. In MICS6 Africa countries, children from the poorest households remain at very low reading levels even by the end of primary school. In other regions, richer children are already far ahead in the early grades.
Two things are true at once. Many children do gain reading skills as they move through school. But poorer children start lower and gain less ground. The result: where and when you are born still shapes whether school teaches you to read.
Behind the charts
I co-led the analysis and built the workflow behind it. The GitHub repository reproduces the full pipeline from harmonised MICS indicators stored in the UNICEF Global Data Warehouse. It pulls the data, reshapes them into a consistent structure, applies analytical rules, and exports the tables and charts used in the story.
The final charts are designed to be simple, but the process behind them is reproducible. The repository makes every analytical choice visible: which indicators were used, how grade-level and stage-level outputs were structured, how disaggregations were handled, and how figures were generated. That transparency matters. It means the reader can check both the message and the machinery.
The message is clear: many children gain foundational reading skills too late, and children from poorer households face the steepest climb.
Where this points
Schooling is not enough if learning comes too late. The gradient shows that access alone does not close the gap. Timing and equity do.
Read the interactive story on Flourish: The learning divide: Which children are gaining foundational skills?
Explore the GitHub repository: Foundational Learning Gradient
About the data
This analysis uses the Foundational Learning Skills module from MICS6, covering around 45 countries and territories. The reading measure identifies children who can read and understand a short story. Results are shown across grades and school stages, with comparisons between MICS6 Africa countries and other MICS6 countries. The workflow uses survey-weighted estimates, applies minimum sample-size rules and produces reproducible output tables and chart files from the UNICEF Global Data Warehouse. The charts on this page load assets/data/030101_output_grade.csv and assets/data/030102_output_grade_band.csv (the same tables produced by the LearningGradient R pipeline) and apply the same filters as 02_scripts/0203_produce_charts.R before drawing.
Acknowledgements
I co-led this work with Sakshi Mishra, whose guidance shaped it throughout, under the leadership of JP Azevedo, within the Education team in the Data and Analytics Section at UNICEF Headquarters.