Your data

A Truman export is a zip of spreadsheets with what every participant saw and did, plus a codebook that explains each column. This page covers how to get it and how to read it. It also covers how to delete one person's data when they ask.

Download it

1. On the Studies page, press Monitor and download for the study.
2. Find Download the data and press Download.

You get a zip of CSV files and codebook.md. Two options sit above the button:

• Shortest view to count (ms) and Longest view to count (ms) set which views of a post count toward the per-person summary. Every view stays in the full data, so you can apply other limits later. Blank counts them all.
• Include recruitment IDs adds a file linking study data to people, for things like paying bonuses. Only a lab's owner sees this option. Don't share that file.

Owners and editors can download. Viewers can't. Each download is recorded, and Who has downloaded it lists them.

Photos that participants uploaded are a separate download, Download photos. It includes photos.csv, which ties each photo to the data.

Test participants are never in a download.

What's in the zip

File	What it has
participants.csv	Each participant: their ID in this study, their version, their condition, how they were assigned, and how far they got
summary.csv	One row per person with totals: likes, flags, votes, follows, time active, views
events.csv	The event log: everything participants did
impressions.csv	What each person was shown, in what order, and whether a post was removed or labeled then
content.csv	The study's own posts and comments, and what participants wrote
beat_runs.csv	Every decision about a beat, and how each line was worded
assignments.csv	Every assignment in order, when a decision point moved anyone
removals.csv, labels.csv, and the other moderation sheets	What the version scripted, when a study has some
recruitment_ids.csv	Only when asked for
codebook.md	The guide to all of it

participant_id is specific to this study. It isn't the person's Prolific ID. If you passed it to your survey as in Consent, surveys, and debriefing, it's the column you join on.

The codebook

Open codebook.md first. It's written for this export, so it describes exactly what you have. It lists every file and column, each version of the study with its settings and the features on in each condition, when each version was open, and the event types that appear. It also says how many people's data was deleted, split into declined consent and withdrew.

Whatever your study used is frozen with its version, so you can write your methods section from here.

Events

events.csv has one row for each thing a participant did: a type, who did it, what it was done to, a payload with the details, and when. Some types you'll use often:

Type	Means
content.created	They wrote a post, comment, or message
content.liked, content.voted, content.reposted	They reacted to something
content.flagged, actor.reported, actor.blocked	They flagged a post, or reported or blocked an account
content.viewed	A post was on their screen, with dwell_ms for how long
actor.followed	They followed someone
flow.step_completed	They finished a step such as consent
decor.pressed	They pressed something that's only for show, such as Marketplace. key and label say what, and page where

Three things are worth knowing when you read it:

• Scripted content appearing on schedule isn't an event. The version already says when each post appears. To see what someone was shown, use impressions.csv.
• Being in impressions.csv doesn't mean it was seen. It means the post was in the list sent to their browser. content.viewed records time actually on screen.
• There are two clocks. occurred_at is the participant's browser clock, which can be wrong. recorded_at is when the server received it.

Withdrawal and deletion

When a participant asks to withdraw, you delete their data:

1. Open Monitor and download for the study.
2. Under Withdraw a participant, type Their recruitment ID and a Reason.
3. Press Withdraw, then Delete their data to confirm.

This can't be undone

This permanently deletes everything about them in this study, including their events and any photos they uploaded. The page tells you how many recorded events were deleted.

What's kept is a record that a deletion happened, with its reason, and a scrambled form of their ID that stops it entering this study again. They can still join other studies. Declining consent does the same thing automatically.

Deleted participants don't appear in any later download. Truman can't reach a zip you already downloaded, though, so if you exported before a withdrawal, delete that person from your own copies too.

In a shared world, what a withdrawn person wrote is blanked and marked withdrawn, so other participants' events still make sense. See Several real people together.

Synthetic data

You can have data shaped like your study's before a single participant arrives. That lets you write and debug your analysis ahead of time.

1. On the Studies page, press Simulate beside a saved version.
2. Choose At random or Personas, and how many.
3. Tick Keep them as synthetic data, and press Simulate.
4. When it's done, press Download synthetic data on the same page.

Its codebook says at the top that nobody real took part.

Synthetic participants are never in the study's real data, its counts, or the Monitor page. Their timestamps are in the past, starting one study length ago. If you want real and synthetic rows together, combine the two downloads yourself.

Before you share data

• Leave recruitment_ids.csv out.
• Your survey data probably contains Prolific IDs now. Remove that column.
• Participants' own posts and comments are in content.csv. Read them before you publish them.

The export can also be written from the command line. See Studies as files.