Processes for informed consent to research

1. Background is intended as a collective intelligence engine. Most conversations are part of one or more participatory processes, and are aggregated to produce a result. In other words, we almost always do some kind of research with Edgeryders material. This implies taking very seriously the issue of research ethics. In this topic, we document the different ways we acquire from participants to our conversations their informed consent to their content being used for research.

2. Consent funnel

To post on, people need to create an account. The first time they post from that account, the forum serves them a simple questionnaire (two questions at the time of writing), to inform them of the website’s research purposes, and make sure they understand how to use it. For example, if they reply that they would gladly share their credit card number, or intimate information about their health, or about illegal political activities, the questionnaire “stalls”, preventing the user from posting until he or she gets the response right.

Once the user provides the right answers, the website records that he or she has consented to participating in research and understands what that means. The date of completion of questionnaire is stored in the database. Upon retrieval for data processing (for example, ethnographic coding or SSNA, only the content associated to users who have completed the process is served (this is necessary because some content was created before we put this process in place.

You can see the text of the questions, answers and feedback on our GitHub repository. This is also where we can make changes to it: just edit the file, commit and ask @matthias to deploy the new version.

3. Other methods: physical events

The process used for physical events, based on the workshop on Inequalities in the age of AI, works like this:

  1. In the opening of the event, inform participants that notes will be taken, and they will be used for research. Introduce the research in question.
  2. Apply Chatham House rules, so that opinions expressed during the event are noted, but not attributed to anyone. Also, forbid taking photos or video.
  3. People are still encouraged to post on the platform; if they do, the consent funnel applies as usual.

This method is privacy-aware, but does not lead to well-formed SSNs, because the identities of individual speakers are lost.

4. Other methods: posting on Edgeryders before signup through an online form

We have created two feeder websites. elicits reflections pertaining to our projects in the form of a registration to a physical event. is a facility for reporting the results of listening triads during events. Both feeder websites contain questions pertaining to the research, and elicit one piece of structured data, the informant’s email address. Both contain a tick box next to a statement that the person filling the form understands that the content will be used for research. The statement uses the same wording as the consent funnel. If the box is not ticked, submission of content and email address is impossible.

Upon submitting the form, creates a new account associated to the email address in question; marks this new account as having successfully completed the ethical funnel; and posts the answers to the question in the form of a new topic, whose authorship is attributed to the newly created account.

5. Other methods: one-on-one interviews

For one-on-one interviews, two methods have been used. One is a more or less standard informed consent process.

TODO: Link a resource with available evidence (email threads?).

The second method re-uses the idea of questions that the interviewee must answer, and proposes the same questions that the consent funnel does, only in paper form.

TODO: Collect the signed form and link to a resource where they can be found.

6. Other methods: the interviewee posts her own interview as a topic

This process was prototyped by @hugi at Internetdagarna:

7. Other types of content: audio recordings of community calls

Some people ask us to post audio recordings of community calls. We recommend not to do this, as doing this ethically is quite tricky. Much better to post written accounts, authored by a specific individual (normally a community manager), and viewed through the lens of their subjectivity. Additionally, community calls are meant to encourage casual sociality, and we have found that people are more reluctant to socialize if they know there are permanent records of their interaction.

If you still decide to go ahead and do it, you need to make sure that people give their consent before the call, ideally in writing. This is because consent is not supposed to be sprung on people: they need to have time to consider whether to give it or not. In a group setting, there is the additional issue that an individual might feel pressured to say yes if everyone else has said yes. When you start the call, begin by confirming that everyone is still OK with the consent they have given. If someone comes on the call that has not given consent, you are then supposed to stop recording.

8. Additional resources

For cases not covered here, the association of internet research has ethical guidelines. Their 2019 report is available here.