RESEP’s Martin Gustafsson recently analysed historical trends in performance in three international evaluations, PIRLS, PISA and LLECE, for Unesco’s Institute of Statistics (UIS) in a paper titled “How fast can levels of proficiency improve? Examining historical trends to inform SDG 4.1.1 scenarios“. He finds that gaps between countries have been narrowing, with a ‘Business as Usual’ trend of around 0.04 standard deviations improvement per year for the least developed countries, and between zero and 0.01 for developed countries. Moreover, the data suggests a ‘speed limit’ (an upper estimate of what is realistically possible to sustain) of around 0.08 standard deviations per year for the least developed countries.


Gustafsson finds that the trend for the percentage of children who are proficient in reading or maths is an annual improvement of 2.0 percentage points among the worst performing countries, “though for the best performing countries, the norm is zero. The latter is not surprising as well performing countries are left with very few non-proficient children.”


In a second report, “Evidence-based projections and benchmarks for SDG Indicator 4.1.1“, Gustafsson draws from the first report in producing projections for the world, and world regions, with respect to the percentage of children who are proficient. This modelling involves bringing together an existing dataset on country-level proficiency levels, with corrections for a few countries, including India and South Africa, UN population projections, and statistics on children who are out of school. This report confirms what is widely accepted, namely that SDG targets with respect to proficiency, are largely aspirational, and not driven by empirical research of what has been possible in recent history. But more importantly, it provides possible targets which are realistic, and which could be helpful in gauging whether progress seen in future years is acceptable or not. To illustrate, moving from a current situation, where 57% of lower primary children worldwide are proficient, to a level of 65% by 2030, would, in fact, be an optimistic scenario.


The two papers are introduced by Gustafsson in this blog post.