@techreport{ai26-1646,
  title = {{The Effects of Standalone Digital Learning Programs on PreK–12 Mathematics and Science Achievement: A Meta-Analysis of Randomized Controlled Trials}},
  author = {Qi Zhang and Laura Michaelson and Kamal Middlebrook and Sarah Sahni and Natalie Bills},
  institution = {Annenberg Institute at Brown University},
  type = {EdWorkingPaper},
  number = {26-1646},
  year = {2026},
  month = {November},
  doi = {10.26300/x38n-mh30},
  url = {https://edworkingpapers.com/ai26-1646},
  abstract = {Most reviews of educational technology pool programs that supplement teachers with programs that teach on their own. We synthesized 68 randomized controlled trials (147 effect sizes) of standalone digital learning programs in PreK–12 mathematics and science, compared with traditional in-person classroom instruction. Using correlated and hierarchical effects models with robust variance estimation, we tested moderators in blocks, adjusting for study characteristics. Students who used standalone programs outperformed comparison students by an average of g = 0.47 (95\% CI [0.35, 0.60]), an estimate that held across sensitivity analyses with no evidence of selective reporting. Effects varied widely, with programs that gave personalized feedback or paired practice with direct instruction producing larger effects than programs without personalization or with practice alone. Effects were also larger in low- and middle-income countries than in high-income countries, and larger in studies with a conflict of interest. Because most programs were used at school for part of students’ instructional time, the findings describe what well-designed programs can contribute within regular schooling.},
}