Illustrative AI-Generated and Student-Generated MCQs: Random Sample of 30 Paired Items with Item-Level Discrimination Indices

This dataset contains a randomly sampled subset of 30 paired multiple-choice questions (MCQs) from a participant-blinded randomized controlled trial comparing AI-generated and student-generated MCQs for formative medical education assessment.

Each pair includes one AI-generated and one student-generated MCQ aligned to the same MEDD 411 learning objective. The dataset includes question stems, answer options, answer keys, explanations, shared learning objectives, and de-identified item-level discrimination index values for transparency and reproducibility.

AI-generated items were produced using a dual-LLM workflow involving ChatGPT-4.0 for item generation and Google Gemini v1.5 Flash for independent validation. Student-generated items were authored by senior medical students without AI assistance using a structured peer-review process.

This public subset is provided to illustrate the item-development protocols and support reproducibility while preserving participant-level confidentiality.