A negative behavioral path through frequent resubmissions counteracted the affective benefit of layered feedback¶

Claim · All claims
Evidence · 1 study (2 entries) · causal r2 · q3
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person

Subclaims¶

q3 i? The average number of tasks with three or more submissions negatively predicted post-test scores (B = −0.33, p = .001) and carried a significant negative indirect effect (Ind = −0.19). → Jie Cao 2026 q3 i? The indirect effect through mental effort was not significant (Ind = −0.02, p = .80). → Jie Cao 2026 (2)

Evidence¶

Jie Cao 2026¶

Jie Cao, Chloe Qianhui Zhao, Christian Schunn, Elizabeth A. McLaughlin, Jionghao Lin, Kenneth R. Koedinger. (2026). Assessing the Impact and Underlying Pathways of Sequenced AI Feedback on Student Learning. arXiv. https://arxiv.org/abs/2604.07469 doi:10.48550/arxiv.2604.07469

q3 · i? · causal · r2

Bootstrap mediation model (N=199); the behavioral mediator sub_3 negatively predicted post-test (B = −0.33, SE = 0.10, p = .001), which the authors read as trial-and-error rather than constructive iteration.

"Conversely, there was a significant neg- ative indirect effect through the average number of tasks with >=3 sub- missions (Ind=−0.19,SE= 0.09,p=.03,95%CI[−0.374,−0.037])."

Jie Cao 2026 (2)¶

Jie Cao, Chloe Qianhui Zhao, Christian Schunn, Elizabeth A. McLaughlin, Jionghao Lin, Kenneth R. Koedinger. (2026). Assessing the Impact and Underlying Pathways of Sequenced AI Feedback on Student Learning. arXiv. https://arxiv.org/abs/2604.07469

q3 · i? · causal · r2

Same mediation model; mental effort did not significantly predict post-test (B = −0.03, p = .77), so the cognitive pathway was neutral despite layered feedback raising mental effort.

"The indirect effect through mental effort wasnotsignificant(Ind=−0.02,SE= 0.06,p=.80,95%CI[−0.167,0.099]), indicating that self-reported mental effort did not translate into a per- formance advantage or disadvantage in this context."

Discussion¶

Related Claims¶

• In a case analysis, probing instructor feedback with required revise-and-resubmit moved one teacher candidate toward citing student evidence, though his progress was not linear and another candidate did not improve — the general claim this finding bears on, with the studies behind it
• Layered AI feedback was perceived as more encouraging and independence-supporting but imposed higher mental effort — related
• Layered AI feedback led to significantly more tasks requiring three or more submissions, while total submissions, learning time and edit distance did not differ — a broader claim this one bears on
• The total effect of layered feedback on learning outcomes was significantly negative, with an unmediated negative direct effect — related
• Perceived independence did not mediate a significant indirect effect of layered AI feedback on learning performance — related
• Perceived risk negatively predicts use behavior and operates as a stage-sensitive deterrent, activated at submission and evaluation rather than initial exploration — related
• The number of substantive comments during discussion mediated the effect of preparatory condition on immediate learning, but not on long-term learning — related