Model-based analysis shows the relative efficacy of testing is reduced for highly related word pairs, with the dual-memory model overpredicting test performance only in the high-relatedness group¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r2 · q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? In an ANOVA comparing observed vs. model-predicted test performance, the Data Type × Relatedness interaction was significant (F(1, 229) = 16.2, p < 0.0001, ηp2 = 0.07), driven by a robust observed-vs-predicted difference in the high relatedness group but not the low group. → Gupta 2026 q3 i? The efficacy of testing was reduced by about 26% in the high relatedness group relative to the model prediction. → Gupta 2026 (2)

Evidence¶

Gupta 2026¶

Gupta, M.W., Pan, S.C., & Rickard, T.C. (2026). Semantic relatedness and the efficacy of retrieval practice. npj Science of Learning. https://doi.org/10.1038/s41539-026-00416-8

q3 · i? · causal · r?

Model-based ANOVA on observed vs. predicted test performance (Data Type × Relatedness) in the cued-recall experiment; the interaction was significant, F(1, 229) = 16.2, p < 0.0001. Pairwise tests showed "a robust difference in the high relatedness group" but none in the low group.

"There was a non -significant difference in the low relatedness group, t(117) = 1.42, p = .16, BF01 = 3.65, but a robust difference in the high relatedness group, t(112) = -4.74, p < .0001, BF10 = 2411. In summary, there wa s evidence from the model fits that the efficacy of testing is reduced in the high relatedness group compared to the low relatedness group."

Gupta 2026 (2)¶

Gupta, M.W., Pan, S.C., & Rickard, T.C. (2026). Semantic relatedness and the efficacy of retrieval practice. npj Science of Learning. https://doi.org/10.1038/s41539-026-00416-8

q3 · i? · causal · r2

Authors' summary of the mean-level model misfit in the high relatedness group: "the efficacy of testing was reduced by about 26%" relative to the dual-memory model's predicted mean testing effect. No standardized effect size is printed for this reduction.

"Nevertheless, the efficacy of testing was reduced by about 26% in the high relatedness group (based on the difference between mea n TE in that group and the model’s predicted mean)"

Discussion¶

Related Claims¶

• A correlated-strengths extension of the dual-memory model with correlation parameter ρ = .334 fits the high-relatedness mean and distribution data — related
• Cumulative distribution analysis shows the dual-memory model systematically overpredicts test performance at all restudy levels for high-relatedness pairs, ruling out a participant-level ceiling effect — related
• Item-level generalized linear mixed model finds no Testing Condition × Semantic Relatedness interaction, and item-level proportions correct show no pervasive ceiling effect — reports the opposite
• Retrieval practice with feedback improves 24-hour cued recall more than restudy, and high semantic relatedness raises overall accuracy, but the Training Task × Relatedness ANOVA interaction is non-significant — related