AI-Assisted MCQ Creation Study Data on Item Quality and Teacher Interaction

A study of 152 multiple-choice questions authored by 19 teachers under three conditions—teacher-only, AI-only, and teacher–AI collaboration—evaluated using a 19-criterion rubric. The dataset, authored by Dennis Menze and last updated in May 2026, originates from research on automation bias in AI-assisted educational resource creation. It includes logged authoring-interaction density and item quality scores.

Use Cases

• Analyzing patterns of item-writing flaws based on the 19-criterion rubric
• Studying automation bias in human-AI collaboration based on logged interaction density
• Comparing surface quality of AI-generated versus human-authored assessment items
• Investigating specific cueing biases like lexical overlap between stem and answer

Strengths

• Includes 152 items from 19 teachers, providing a substantive sample for analysis
• Evaluates items using a detailed 19-criterion item-writing-flaw rubric
• Compares three distinct authoring conditions: teacher-only, AI-only, and teacher–AI collaboration

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment
• The dataset is very small at 2.4 KB, indicating limited scope

Provenance

Source

figshare

Collection Method

Experimental study comparing MCQ quality across three authoring conditions.

Freshness

Last updated 2026-05-26 04:38:58