A Meta-Analysis of Extensive Reading Research

34 studies (32 articles, 2 dissertations), 43 effect sizes, 3,942 participants

Second- and foreign-language learners across school and university settings

Pools two structurally different contrasts and reports them separately, which is the right thing to do and rarely done: GROUP contrasts (extensive-reading condition vs control) and PRE-POST contrasts (experimental group only, no control). The gap between the two is the size of the design artefact.

Extensive reading produced a medium effect of d = 0.46 on GROUP contrasts - the estimate with a control group - and a larger d = 0.71 on PRE-POST contrasts with no control. The 54% difference between those two numbers is a clean demonstration of what dropping the control group buys an author. The controlled estimate, 0.46, is the one that counts.

Low-to-medium; most included studies compare classes assigned to extensive reading against classes that were not, within the same institution.

Superseded in coverage but confirmed in direction by Sangers et al. (2025), a larger and more recent meta-analysis that finds positive small-to-medium effects across every language domain and identifies the moderators.