Feedback comprehension showed the strongest bivariate association with revision quality, and indirect associations through actionability and comprehension remained positive after adjustment¶

Claim · All claims
Evidence · 2 studies · 2 causal r2 · q2 · i3 large
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i3 Comprehension correlated with revision quality at r = 0.501, stronger than diagnostic accuracy (r = 0.445) and actionability (r = 0.480). → Yao X-C and Fan L 2026 q2 i? Exploratory indirect effects from diagnostic accuracy to revision quality through actionability (0.043) and comprehension (0.024) remained positive after adjustment for condition and cycle using student-cluster bootstrap confidence intervals. → Yao X-C and Fan L 2026 (2)

Evidence¶

Yao X-C and Fan L 2026¶

Yao X-C and Fan L (2026) From automated scoring to learning diagnosis: a mechanism study of AI-supported formative assessment in English writing. Front. Psychol. 17:1905455. https://doi.org/10.3389/fpsyg.2026.1905455

q2 · i3 · causal · r2

Bivariate correlations among process variables in the cycle-level dataset. "comprehension showed the strongest correlation (r = 0.501, p < 0.001)" with revision quality, exceeding actionability (r = 0.480) and diagnostic accuracy (r = 0.445).

"Diagnostic accuracycorrelatedwithrevisionquality( r =0.445, p < 0.001); actionability correlated with revision quality (r = 0.480, p < 0.001); and comprehension showed the strongest correlation (r = 0.501, p < 0.001)."

Yao X-C and Fan L 2026 (2)¶

Yao X-C and Fan L (2026) From automated scoring to learning diagnosis: a mechanism study of AI-supported formative assessment in English writing. Front. Psychol. 17:1905455. https://doi.org/10.3389/fpsyg.2026.1905455

q2 · i? · causal · r2

Exploratory indirect-effect analysis with diagnostic accuracy as predictor, actionability and comprehension as intermediates, and revision quality as outcome, using 5,000 student-cluster bootstrap samples. All adjusted indirect effects remained positive.

"After adjustment for assessment condition with 5,000 student-cluster bootstrap samples, the indirect effects were 0.043 through actionability (95% CI [0.012, 0.078]), 0.024 through comprehension(95%CI[0.005,0.050]),and0.067intotal(95%CI [0.024,0.113])."

Discussion¶

Related Claims¶

• Diagnostic Review revisions contained larger shares of meaning-change, organization, and content/argument work, while Control revisions remained more dependent on surface-level editing — the general claim this finding bears on, with the studies behind it
• Group-by-cycle interactions were significant for diagnostic accuracy, comprehension, revision quality, and self-assessment accuracy, but not for actionability — related
• The Diagnostic Review class scored higher on diagnostic accuracy, actionability, comprehension, revision quality, and self-assessment accuracy across all three cycles — related
• Qualitative analysis found staged disclosure sometimes withheld inaccurate diagnostic details, but some errors still shaped student revisions — related
• Perceived usefulness of response-level LLM feedback is most strongly associated with actionability and perceived accuracy — related