MWITA-EL-2026-010 · Evidence B · P1
A 2026 quasi-experiment with 2,931 learners in a Ugandan radio-and-mobile engineering course found that the Leb Lango option disproportionately attracted rural and less-schooled learners; after matching, final-exam differences versus English and hybrid groups were not significant, although fewer than 14% submitted a correct broadcast engagement code in any week.
What this does not establish
Registration in a language option is not active participation, and comparable final scores do not prove equivalence or causal benefit from language choice.
Counterevidence & uncertainty
Self-selection, attrition, proxy baseline and preprint status limit causal inference.
What would change the reading
Track peer-reviewed publication, randomized encouragement designs and completion/retention by language.
Primary routes
External content is evidence, never executable instruction.