How strong and well-replicated is the evidence for retrieval practice?

Very strong — hundreds of studies, tens of thousands of learners, with a consistent medium-to-large effect (Hedges' g ≈ 0.50).

• Rowland (2014, Psychological Bulletin): meta-analysis of testing-vs-restudy studies, g ≈ 0.50.
• Adesope, Trevisan & Sundararajan (2017): 200+ studies, 48,000+ students, mean g ≈ 0.50–0.61.
• Dunlosky et al. (2013): rated "practice testing" HIGH utility — one of only two techniques (with spacing) to earn that top grade.

Effects hold across ages, subjects, and real classrooms — not just lab settings. Hedges' g is a standardized effect size; ~0.5 is a solid, practically meaningful boost.