Flagged Spring 2026 students scored lower on all three exams than non-flagged peers, with performance declining as flagged assignments increased¶

Claim · All claims
Evidence · 1 study (2 entries) · associational r1–r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Flagged students averaged 61.9% across the three exams versus 73.5% for non-flagged students, and means declined from 69.4% with one flagged assignment to 53.5% with six. → David Racovan 2026 q2 i? The gap was largest on Midterm 2, where flagged students averaged 59.0% versus 72.2% for non-flagged students. → David Racovan 2026 (2)

Evidence¶

David Racovan 2026¶

David Racovan, Ajay Rawat, Christopher K. May, Jeffrey A. Turkstra. (2026). Argus: Academic Integrity in the Era of Generative AI. https://arxiv.org/abs/2609.36073 doi:10.48550/arxiv.2609.36073

q2 · i? · associational · r1

Spring 2026 case study comparing manually flagged (n=267) and non-flagged (n=317) students on proctored exams; the text reports "a mean of 61.9%" versus 73.5%. Only descriptive means are printed, so no effect size is coded.

"Averaged across all three exams, flagged students achieved a mean of 61.9% (median 65.9%), compared to 73.5% (median 75.7%) for non-flagged students."

David Racovan 2026 (2)¶

David Racovan, Ajay Rawat, Christopher K. May, Jeffrey A. Turkstra. (2026). Argus: Academic Integrity in the Era of Generative AI. https://arxiv.org/abs/2609.36073

q2 · i? · associational · r2

Per-exam breakdown in the Spring 2026 case study (Figure 7); the Midterm 2 contrast names flagged versus non-flagged students with printed means of 59.0% and 72.2%. Descriptive only; no test statistic is printed for this contrast.

"The gap was largest on Midterm 2, where flagged students averaged 59.0% versus 72.2% for non-flagged students."

Discussion¶

Related Claims¶

• Students received significantly higher homework scores during GenAI-evaluation periods, but this advantage did not transfer to downstream exams — the general claim this finding bears on, with the studies behind it
• In Spring 2026, 45.7% of students in a large CS2 course were flagged as LLM-assisted following structured manual review — related
• MOSS detected far fewer suspicious cases in Spring 2026 than in prior offerings, as LLM-assisted submissions lack similarity signals — related
• Students who used LLMs on assignments scored significantly lower than non-users on the first midterm (large effect size), but this gap narrowed and was non-significant on the later midterms — related
• Average H-scores rose from near-zero in pre-LLM offerings to 11.9 in Spring 2026, with maximum scores reaching 81 — related