By Dr Junaid Mubeen, Director of Education at Whizz Education
Learning loss has floated into the lexicon of educators, as we reckon with the consequences of COVID-19 on students’ academic progress. It can be defined in multiple ways. At Whizz Education, we simply take it to mean the erosion of previously acquired knowledge. It is not a new phenomenon: the so-called ‘summer slide’ is an established annual occurrence, with students losing around 2-3 months’ worth of maths knowledge over the extended break. Estimates from the World Bank place the global learning loss as a result of COVID-19 at 6-12 months. Not all students are affected equally, however. Disadvantaged students from low-resource settings bear the brunt of prolonged breaks from schooling as our own new research from rural Kenya shows.
To date, there have been relatively few studies focused on low-resource settings. Fewer still have probed learning loss at the level of individual topics. To address this gap Whizz Education undertook an analysis of learning levels in rural Kenya, where schools were closed between March and September 2020.
Our analysis of almost 1000 students in Kenya showed that 53% of them experienced a decline in Maths Age (analogous to Reading Age: a nine-year-old student is expected to have a Maths Age of nine and so on)corresponding to learning loss. The average loss among students experiencing knowledge declines was equivalent to 1.1 years, or 13 months, with losses greatest in topics involving formal calculation methods. This confirms the widespread and emphatic nature of learning loss. It reinforces the urgent need for educators and policymakers to direct attention and resources towards recovery efforts.
Figure 1: Learning loss by topic
Our analysis found that grade level is a predictor of learning loss, with a higher proportion of students in lower grade levels affected. We speculate that this is because students with a higher Maths Age have a stronger foundation of knowledge to fall back on during periods of disruption.
We found no significant difference in the proportion of boys and girls who experienced learning loss, although losses were slightly higher for affected girls: an additional seven weeks of lost learning compared to boys. A greater proportion of students from ‘hardship’ areas experienced learning loss.
Figure 2: Learning loss by grade level
Figure 3: Learning loss by school setting
Sample and methodology
Our analysis is based on a sample of 965 students across four counties in rural Kenya who participated in Project iMlango, an FCDO-backed eLearning Programme aimed at improving education outcomes in mathematics, literacy and life skills. Mathematics provision took the form of AI-enabled virtual tutoring via Whizz Education’s Maths-Whizz platform along with teacher training, capacity building and ongoing implementation support. As students interact with Maths-Whizz, the virtual tutor computes their knowledge levels in terms of a criterion-referenced international Maths Age metric, which is automatically updated in real-time.
The sample consists of students who enjoyed consistent access to Maths-Whizz prior to school closures and thus had a reliable Maths Age, who were then reassessed upon returning to school in the 2020/21 academic year. The difference in Maths Age between these two points in time can therefore be taken as a measure of the change in students’ learning levels in mathematics during the first wave of COVID-19, with a decline in Maths Age corresponding to learning loss.
Familiar trends amplified and an urgent call to action
Relative to Whizz’s previous findings from the United Kingdom and the United States, a greater proportion of students in rural Kenya experienced learning loss, and to a greater extent on average. The findings are largely explained by the limited learning provision for students in rural Kenya during COVID-19.
Since the sample comprises students who had sustained access to virtual tutoring prior to COVID-19, the students in this analysis may represent relatively high-attaining, highly motivated students who were able to secure stronger knowledge foundations via Maths-Whizz prior to COVID-19. The 53% figure should therefore be considered an underestimate.
An open question for research, which Whizz continues to explore, is how the remaining 47% of students managed to sustain or improve their knowledge levels in spite of limited resources. What is known for certain is that students in rural Kenya who were already lagging several years behind their affluent peers in Kenya and worldwide are now at a further disadvantage, with knowledge gaps exacerbated across the mathematics curriculum. Our most recent assessment data suggests that these students are now more than five years behind their expected levels and the levels of their peers in more developed contexts.
Figure 4: Difference between a student’s ‘maths age’ and actual age, by topic
Recovery is another term pursed on the lips of every educator right now. In the context of these findings, communities in low-resource settings must be supported with tools to accelerate their students’ learning. The good news about lost learning is that it can be recovered, provided that every student is supported with a learning plan that identifies, and then targets, their specific knowledge gaps.
The Global Education Evidence Advisory Panel recently highlighted adaptive learning software as a ‘Good Buy’, with the usual caveats of accessibility and infrastructure, and training and support for teachers. When implemented with fidelity, and access is available even in the face of disruptions to schooling, virtual tutoring is a proven mechanism for accelerated learning which, in light of the losses observed in places like rural Kenya, has become imperative for achieving SDG4.
Download the full learning loss report here.
For more like this, sign up to Whizz Education’s monthly newsletter, Whizz Insights.