Research Methods
A plain-language summary of how the project presents media, records responses, and uses participant data.

Study flow

Participants review a sequence of media items and classify each one as real or AI-generated. At the end of play, the app may ask survey questions that have not already been answered for that user.

Media selection

Collections are presented one item at a time. For a given collection, the app prioritizes images the user has not seen yet. After the collection has been exhausted, replay sessions can draw randomly from that same collection.

Recorded data

The system stores the selected collection, each response, whether the guess matched the ground truth, timestamps, per-image response duration, and passive session context such as IP-derived location, timezone, language, device type, browser family, operating system family, viewport bucket, and referrer class. Survey answers, including questions such as main area of study, are stored at the user level rather than per session.

Accounts

Guest play uses anonymous Supabase accounts. If a guest upgrades to a full account, existing linked records remain associated with that same underlying user identity.

Project-specific edits

Replace this page with your finalized methodology, recruitment criteria, consent language, and analysis notes before publishing externally.