Academic Outcomes 2 Years After Working Memory Training for Children With Low Working Memory: A Randomized Clinical Trial

452 children randomised (226 per arm) from 1,723 screened across 44 schools; 88% retention at 2 years

First-graders aged 6-7 with low working memory, Melbourne, Australia

Population-based RCT with concealed allocation by an independent statistician, blinded screening, intention-to-treat analysis, directly assessed academic outcomes at 12 and 24 months, and a costing. This is the trial that should have settled the question favourably if the effect were real: real schools, real delivery (90.3% completed at least 20 sessions), a targeted population with room to improve, and a two-year horizon.

Of four short-term and working-memory outcomes, only visuospatial short-term memory benefited, at 6 months (ES 0.43, 95% CI 0.25 to 0.62) and 12 months (ES 0.49, 95% CI 0.28 to 0.70) - and NOT at 24 months. No benefit to any other outcome. Mathematics scores in the intervention arm were WORSE at 2 years (mean difference -3.0, 95% CI -5.4 to -0.7, P = .01), plausibly the cost of lost classroom time. Intervention cost A$1,035 per child. The authors cannot recommend population-based delivery of Cogmed.

Low - randomised with concealed allocation. Note the opposite lesson: the children were selected on low working memory, so regression to the mean would have flattered the intervention arm if anything, and it still failed.