Contextual interference: a meta-analytic study

61 studies, 139 effect sizes

Mixed; laboratory motor-task learners and applied sport-skill learners, children through adults

139 effect size estimates from 61 studies, coded for basic (laboratory) vs applied research, age group, and outcome phase (retention vs transfer). Pre-modern methods: no dependency correction, no preregistration, no formal publication-bias analysis. Author computed the mean statistical power of the reviewed studies (.43), i.e. the corpus was chronically underpowered.

The canonical demonstration that CI is largely a laboratory phenomenon: basic-research d=.57 vs applied d=.19, a 3x attenuation, established two decades before the 2024 metas found the identical pattern (0.92 vs 0.23 retention; 0.75 vs 0.34 transfer). Also the earliest quantitative signal of the age reversal — adults .50 vs younger learners .10. Mean power of .43 means the underlying literature was chronically underpowered, which inflates published effects. Brady's framing question, whether CI is a laboratory artifact or a genuine sport-skill phenomenon, is still unresolved 20 years later.

Not addressed; no individual-difference or aptitude-treatment analysis.

Both the lab/applied split and the age moderation were reproduced with modern multilevel methods by Czyż et al. (2024) on a largely non-overlapping study set — an unusually clean 20-year replication of a moderator pattern.