The Computational Democracy Project



Aug 2020: Data export is temporarily disabled in the main instance of as we reconfigure how emails are sent to accommodate third party deployments.

The data export for a given conversation is in CSV form.


  • Summary
    • Basic summary statistics about the conversation, including number of voters, commenters, conversation topic, etc
  • Stats History
    • A timeline of n-votes, n-comments, n-visitors, n-voters & n-commenters
  • participants votes
    • a table of participants (as rows), with data such as participant (numeric id), xid (if export was generated with the -X flag at command line; not currently supported in UI export), group-id (the K-means clustering assignment), n-comments, n-votes, n-agree, and n-disagree
    • additionally, a column corresponding to each comment in the conversation, collectively serving as a vote-matrix suitable for doing downstream analyses
  • comments
    • complete set of comments, with comment-id, author-id, comment-body, agrees, disagrees
  • votes
    • complete vote history, with voter-id, comment-id, timestamp and vote columns

Note that the data export schema is still considered to be in BETA, and may change in the future.


  • This is a very high-level summary of the data about the conversation, including:
    • Topic As specified during conversation creation.
    • URL The url of the conversation.
    • Views The number of unique visitors to the conversation. (Note: this won't necessarily be accurate if people view the conversation with multiple devices but don't log in. For more accurate assessment of this metric, we recommend using Google Analytics on the page in which you embed Polis).
    • Voters Number of participants who voted on at least one comment in the conversation.
    • Voters (in conv) Number of participants who voted enough to be represented in the conversation view. [*]
    • Commenters Number of participants who wrote comments.
    • Groups Number of opinion groups in the conversation.
    • Conversation Description As specified during conversation creation.
  • [*] Only voters who vote on all the comments present at the time of voting, or at least 7 comments, whichever is less, are represented in the visualization.
  • This table has a separate row for each field.

Stats History

  • This table is meant to give you more detailed information about the growth of the conversation over time. The 5 columns in this table are:
    • Votes Total number of votes in the conversation.
    • Comments Number of comments in the conversation.
    • Visitors Number of visitors to the conversation
    • Voters Number of voters in the [[conversation]][XXX is this (in conv) or not?]
    • Commenters Number of participants that wrote comments.
  • Each row in this table corresponds to a change in one of these values due to an interaction with some participant (a vote, comment, or visit). In the future, we may also include a timestamp indicating when that vote occurred.

Participants Votes

  • This table provides information about individual participants' voting patterns and opinion group assignment. Participants remain anonymous in this output, and are referred to by an integer ID unique within the conversation.
  • Each row of this table corresponds to a particular participant. The following columns are included:
    • Participant The participant identifier; an integer identifier unique within this conversation.
    • Group ID Group ID of the group the participant belongs to.
    • Comments Number of comments written by the participant.
    • Votes Number of votes by the participant.
    • Agrees Number of agree votes by the participant.
    • Disagrees Number of disagree votes by the participant.
    • Additionally, for every comment in the conversation, there is a column with that comment's ID as the column name. For a given participant row and comment column, the entry in this table represents the vote that participant gave on that column. A vote of 1 represents agree, a vote of -1 represents disagree, and a vote of 0 represents a pass. If a participant did not vote or pass on the given comment, the corresponding entry is blank.
  • In the future, we may add a column indicating whether the participant was included as "In conversation" or not (as described above).


  • This table provides summary data for each comment in the conversation. Each row corresponds to a particular comment, with the following columns:
    • Comment ID The comment identifier; an integer identifier unique within this conversation.
    • Author Participant ID of the comment's author.
    • Agrees Number of agrees on the comment.
    • Disagrees Number of disagrees on the comment.
    • Moderated Moderation status; 1 for allowed, -1 for moderated out, 0 for unmoderated.
    • Comment Body The actual body of the comment.
  • Note that Comment ID and Author Participant ID correspond with the values in the Participants Votes table.

The Computational Democracy Project

© 2021 The Computational Democracy Project, a 501c3 nonprofit.


© 2021 The Computational Democracy Project, a 501c3 nonprofit.