Research infrastructure

Research data and registered methods

Public protocols, definitions and evidence records used by The Quiz Review. Raw benchmark files will appear here only after collection and quality review.

Last updated: Updated 6 August 2026
Editorial detail: No fabricated datasets
Editorial detail: Versioned methods

Available now

Published method

The first protocol defines four job-specific tests and the evidence required before a result is reported.

Method live

How We Test Quiz Makers: Four Scenarios, Evidence and Scoring Rules

The public test protocol used to separate AI generation, personality scoring, live participation and formal assessment instead of forcing them into one universal score.

Open the method

Registered next

Benchmark queue

A registered protocol is not a completed result. These assets will not receive findings until the stated tests are run.

Protocol planned

PDF-to-quiz source-grounding benchmark

Factual correctness, coverage, traceability, explanations and edit time across a controlled source set.

Protocol planned

AI distractor-quality benchmark

Plausibility, uniqueness, misconception quality and multiple-correct-answer risk.

Protocol planned

Quiz-format decision model

A reusable mapping from purpose, stakes and delivery mode to question, scoring and reporting requirements.

Data-release ruleEvery dataset must include collection dates, sample and exclusions, field definitions, method version, uncertainty, limitations, license and correction history. A chart or summary is not a substitute for the underlying codebook.