Effect of providing free glasses on children's educational outcomes in China - cluster randomized controlled trial

3,177 children (of 19,934 screened) in 252 primary schools, 84 schools per arm; 3,052 (96.0%) completed

Fourth and fifth graders (mean age 10.5) in two prefectures of western China with visual acuity <6/12 correctable to >6/12, 2012-13

Large, investigator-masked, three-arm cluster RCT (prescription only vs voucher vs free glasses in class). Graded B rather than A for one reason that matters under this archive's rules: the achievement outcome was "a specially designed mathematics test", i.e. researcher-designed rather than an independent standardized assessment, and such measures inflate effects roughly 2x. The trial was powered to detect 0.20 SD and the observed effect was below that.

All 3,177 eligible children would have benefited from glasses but only 15% wore any at baseline. At closeout, observed wear was 41% in the free-glasses arm vs 26% in control (self-reported 68% vs 37%) - so even free in-class provision left the majority uncorrected. Effect on the mathematics test was 0.11 SD (95% CI 0.01 to 0.21) for free glasses vs control. The adjusted effect of free glasses (0.10, 95% CI 0.002 to 0.19) exceeded that of parental education (0.03) or family wealth (0.01). The authors themselves flag that the effect was smaller than the prespecified 0.20 SD the study was powered to detect.

Low for the treatment contrast: schools were randomized, so heritable child and parental traits are balanced across arms. Notably the trial's own comparison shows the randomized glasses effect outperforming parental education and family wealth - two variables that in observational work carry heavy passive gene-environment correlation.