Syntes: Chan, Davis, Yurtsever & Myers (2024) — TE independent of practice performance

Källa: Chan, J. C. K., Davis, S. D., Yurtsever, A., & Myers, S. J. (2024). The magnitude of the testing effect is independent of retrieval practice performance. Journal of Experimental Psychology: General, 153(7), 1816–1837. https://doi.org/10.1037/xge0001593

Metadata

• ID: LEARN-013
• Tier: RCT (six experiments)
• N / design: university students; weakly associated cue–target pairs; Exp 1 response-deadline manipulation of practice success (short/medium/long); Exp 2–3 encoding presentations (1× vs 2× / 4×); feedback nested; final cued recall vs no-test control (not restudy)
• Population: vuxen-livslangt
• Subject: whether testing-effect magnitude depends causally on retrieval practice performance

Finding

1. Across six experiments, experimental manipulations that strongly changed practice success did not change TE magnitude (forest-plot meta of manipulations: practice-success d large; TE d ≈ 0).
2. Exp 1a: practice success short M=0.46 vs medium 0.63 vs long 0.64; TE magnitudes .14, .12, .15; deadline on TE F(2,92)=0.87, p=.421, η²p=.02, BF01=7.10; pattern held with and without feedback (BF01 no-feedback=4.50; feedback=7.57).
3. Feedback boosted TE in Exp 1a (M=0.19 vs 0.09) but did not create a deadline × feedback interaction on TE size.
4. Aggregated correlational analysis (n=204, no-feedback): practice success correlated with TE (r=.34), which authors argue is an artifact / individual difference, not a causal experimental effect.
5. Control condition throughout is no initial test, not restudy.

Mechanism

Challenges accounts that treat practice success rate as causal driver of TE size; students can benefit from retrieval practice even under conditions that lower practice accuracy.

Limits

No restudy arm — does not speak directly to C0's retrieval-over-restudy contrast or Rowland's ≤50% restudy moderator. Word-pair lab paradigm. Immediate and 1-week finals; not classroom prose.

This claim dies if …

(For C1 kill prong 2:) equal delayed retention for restudy vs retrieval when the stricter accessibility/success test fails — Chan does not run that contrast.

Swedish implication

Do not treat low practice accuracy alone as proof that retrieval practice is useless when feedback is available; do keep restudy comparisons separate from no-test comparisons.

Links

• DOI: https://doi.org/10.1037/xge0001593
• Related: rowland-2014-testing-effect-meta, kornell-2025-accessibility-learning, roediger-karpicke-2006-test-enhanced-learning