CommentCrate ENTRESFR

Using exported social media comments in research

A comment export can be a data set. How you collect, store and describe it matters as much as the analysis.

What the file gives you

One row per comment: the username, the text, the likes, the time when the platform shows it, whether the comment is a reply (replies need Pro) and the address of the post. The sample export file shows the exact columns. The file is a snapshot of what the page displayed to you at the moment you scanned it.

Document how you collected it

  • The post address, and the date and time of the scan.
  • The plan you used: the free plan reads up to 200 top-level comments per post, Pro has no limit and includes replies.
  • How many comments the file holds and how many the platform said the post has.
  • That hidden or filtered comments, and comments deleted before the scan, are not in the file.
  • Whether you were signed in, since what a page shows can depend on the account, and which version of the extension you used.

It is not a random sample

The comments a page shows are not a random draw from everyone who saw the post. The platform decides the order and what is hidden. Say so among your limitations and avoid claims about all users.

Ethics and permission

Comments are personal data in many places, even when they are public. Before you collect, check the platform's terms, the data protection rules that apply to you and the rules of your institution or ethics board. Consider replacing usernames before you analyze or share the data, and store the file as carefully as any other data set. CommentCrate reads the comments in your browser and does not upload them, but what you do with the file afterwards is your responsibility.

Preparing the data

  • Keep the original export untouched and work on a copy.
  • To anonymize, replace each username in the copy with a number. Keep the key apart from the data, or do not keep it at all.
  • Keep id and parentId if you need to rebuild the threads later.

Where to go from here

Open the file in Google Sheets or Excel, use comment analysis for first counts, and see fixing garbled characters if accented letters look wrong.

More guides

See pricing