The Evidence on Teaching

Meta-analysis of the reduced relative feedback frequency effect on motor learning and performance

McKay, B., Hussien, J., Vinh, M.-A., Mir-Orefice, A., Brooks, H., & Ste-Marie, D. M. · 2022

grade Bmeta-analysisindependentunreplicated
Population
Mostly healthy young adults (university samples) on laboratory motor tasks
Design
Meta-analysis of reduced vs 100% terminal-feedback-frequency experiments, explicitly testing the guidance hypothesis' key prediction that the effect REVERSES from acquisition to delayed retention. Includes z-curve power analysis and prediction intervals. Aggregates small lab RCTs on researcher-designed tasks; pooled point estimates behind paywall (unverified).
Key findings
The textbook claim that frequent feedback degrades learning is NOT supported meta-analytically. Reduced relative feedback frequency showed no significant effect at acquisition, immediate retention, or delayed retention, and — fatally for the guidance hypothesis — no acquisition-to-retention crossover. Prediction intervals are wide enough that a >1 SD advantage in EITHER direction fits the data. Underlying studies are severely underpowered.
Genetic confound
Not addressed; randomized between-subject designs. Heterogeneity source 'remains unaccounted for' — plausibly individual differences in trainability/feedback dependence.
Replication notes
Contradicts the Winstein & Schmidt (1990) / Salmoni (1984) consensus; consistent with the same group's bias findings in self-controlled and attentional-focus literatures.

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Acquisition performancereduced vs 100% feedback frequencyNo significant effect (point estimate unverified)researcher-designedacquisition; during practice; labactive-alternativeend-of-treatmentdomain-skill
Immediate and delayed retentionreduced vs 100% feedback frequencyNo significant effect at either; 95% prediction intervals span >1 SD favoring reduced to >1 SD favoring 100%researcher-designedretention; immediate and delayed; labactive-alternativeend-of-treatmentdomain-skill
Acquisition-to-retention CHANGE (the guidance-hypothesis test)phase x condition contrastNo significant change from acquisition to retention — the predicted crossover is absentresearcher-designedacquisition vs retention contrast; labactive-alternativeend-of-treatmentdomain-skill
Evidential value of the literaturez-curve / power analysisStudies 'severely underpowered' (~3 of 4 per a citing paper); substantial unexplained heterogeneityunknownn/a; meta-levelnonenot-applicabledomain-skill

Cited by