Prompt-count distributions differ significantly across assignments, with moderate prompting most common and extended prompting (10+) rare except in Algorithms (18.0%)¶

Claim · All claims
Evidence · 1 study (2 entries) · associational r1 · q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? A Kruskal–Wallis test confirmed prompt count distributions differed significantly across assignments (H = 71.18, df = 6, p<0.001). → Nizamani 2026

Evidence¶

Nizamani 2026¶

Nizamani, S. B., Lee, Y., Donekal Chandrashekar, N., Ellis, M., & Ramakrishnan, N. (2026). Understanding Student Use of Large Language Models Across Computer Science Subfields. https://arxiv.org/abs/2610.01158 doi:10.48550/arxiv.2610.01158

q3 · i? · associational · r1

Kruskal–Wallis test on ordinal prompt-count ranges (0, 1–2, 3–5, 6–10, 10+) reported by 211 students across seven assignments. The article reports "H = 71.18, df = 6, p<0.001"; no effect size is printed.

"A Kruskal–Wallis test confirmed that prompt count distributions differed significantly across assignments (H = 71.18, df = 6, p<0.001)."

Nizamani 2026 (2)¶

Nizamani, S. B., Lee, Y., Donekal Chandrashekar, N., Ellis, M., & Ramakrishnan, N. (2026). Understanding Student Use of Large Language Models Across Computer Science Subfields. https://arxiv.org/abs/2610.01158. doi:10.48550/arxiv.2610.01158

q3 · i? · associational · r1

Descriptive distribution of prompt counts among LLM users across the seven assignments. The article reports that "3–5 prompts accounted for approximately 29-46% of users" and that 10+ prompts was "rare except in Algorithms (18.0%)"; minimal 1–2 prompt use was more common in Machine Learning and Software Engineering.

"Across assignments, 3–5 prompts accounted for approximately 29-46% of users, while 6-10 prompts accounted for 13-34%. Extended prompting of 10+ prompts was rare except in Algorithms (18.0%)."

Discussion¶

Related Claims¶

• Students predominantly conceptualize the LLM as an Assistant rather than Peer or Expert, and role distributions differ significantly across assignments — related
• Verification strategy type varies by assignment context: testing is more common in structured tasks (Algorithms 87.8%) while web search is more common in open-ended tasks (HCI 56.0%) — related
• LLM adoption varies substantially across CS subfield assignments under a shared instructional intervention, with highest use in Algorithms (89.6%) and Web Development (80.6%) and lowest in Software Engineering (15.2%) — related
• LLM scaffolding level varies systematically across student request types and across assignments, but high-assistance responses remain dominant — related
• Frequency of generative AI use for homework differs significantly by subject, with natural sciences most resistant and economics most adopted — related
• In AQUALAB, estimated β is ordered across episode completion status (means 1.181, 1.506, 2.343; Kruskal–Wallis p = 8.62×10−29) and higher-β quintiles complete more content more efficiently — related
• Perceived usefulness of AI and intent to continue using it differ significantly across challenge domains, with academic tasks rated most useful — related
• Students issue fewer prompts on team assignments, a medium-large within-student difference — related