I recently attended REBWest 2017, a small regional conference focused on research ethics. I got the opportunity to sit on a panel with Peter Chow-White, and we talked about how social media research introduces new challenges for research ethics. This post is a primer on my talk, which was entitled “Ephemerality and Context: Ethical Dilemmas in Online Research.”
Research in Canada that is funded by the major public funding bodies (CIHR, NSERC, SSHRC, collectively known as the “tri-council”) needs to abide by a set of ethical standards set out by the TCPS 2. This document aims to protect participants from harm by ensuring that research carried out in Canada is effectively bounded by a certain set of rules. The reason this document exists is not because researchers are unethical; rather, it is because sometimes, even when we as researchers are trying our best to protect participants, we have blind spots to the harms that may occur.
Some kinds of these guidelines are unambiguous and obvious with respect to harm: we can’t put people through pain if there is no meaningful benefit (to the participant, to society) to doing so. But some others are a little more subtle – for instance, if we are interested in understanding soldiers’ experiences in the field and interview them to find out, these soldiers could experience emotional distress, and maybe even PTSD! Other kinds of research could put situations where participants feel embarrassed, ashamed, or even where they might suffer consequences at work – for instance, if they are asked to participate in a focus group where co-workers are asked about performance of fellow co-workers.
Research Ethics Boards (REBs) at Canadian universities are tasked with evaluating researchers’ plans for research with the participants’ interests at the forefront (i.e. tasked with protecting participants). They provide suggestions to help improve research plans in ways to protect participants’ rights.
The TCPS 2 outlines three core principles that form its ethical framework. The TCPS 2 guidelines are all developed from this ethical framework:
Respect for persons. Researchers and research need to respect the participants as humans with dignity. We need to respect their capacity to make choices for themselves, and not unnecessarily withhold information from them in order for them to make these choices. The main thing that comes out of this is the principle of free, informed, ongoing consent. Each of those words, “free”, “informed” and “ongoing” are important, but we’ll come back to this.
Concern for welfare. This one is really about saying, “Hey, we should care about what happens to people that participate in our studies,” trying to make sure that they stay free from harm, whether the harm is physical or emotional, social or economic.
Justice. Finally, the justice principle is about treating everyone fairly – in terms of people’s participation (who gets to participate), to how they are treated during the study, to who gets the benefit from the outcomes of the research.
In my talk, I focused on the ethical challenges presented by observational online research. To set the context here, “observational online research” refers to research where the researchers have not compelled participants to do anything they wouldn’t have done otherwise; rather, the researchers are merely “observers”. The problem I discussed in my talk was that many of the rules of engagement laid out by the TCPS 2 were/are somewhat inadequate when thinking about online interactions that people have with one another.
The primary challenge is that online interactions are frequently archived and made publicly searchable by various entities (e.g. the social media provider, like Twitter, or by Google). This means that what would otherwise be considered as an ephemeral engagement/interaction with someone could be put on the permanent public record (i.e. it is publicly accessible!). Here, the TCPS 2 is unfortunately far too open to a loose interpretation: in such a situation, it could be considered as “archival research,” in which case, since the data is “publicly available” (in terms of access), then using this data would not require ethical review from an REB (i.e. would not even need to gain informed consent for the use of this data created by people).
This seems ludicrous, because there have been several instances of people getting burned by things that had posted/written in the past. It would seem that if a researcher chose to do research on online social media posts and subsequently publishes it, it would contravene the “concern for welfare” principle: surfacing this information (e.g. by publishing it) could put people in a situation where their social or economiy status is harmed. This seems like a big deal, particularly because it doesn’t even touch on the issue that such people aren’t necessarily in a situation where they are given an opportunity to give/withdraw informed consent.
One of the crazy problems is that even if researchers do not collect “identifying information” such as a real name or a username, when we publish quotations, these exact quotations can be searched on to find the username of the person who posted it. Combined with the fact that people frequently re-use usernames across different sites, this leads to the possibility of being doxxed, where people can easily discover lots of information about a person. The challenge this presents is that publishing these quotations (even without usernames) means that we cannot effectively protect the confidentiality of participants’ participation in our studies.
This sounds alarmist, of course… particularly since it comes from a computer scientist. But, I do not think all is lost.
I think we need to revisit TCPS 2, and how it defines internet-based research.
We need to do this revisiting with members of the community (i.e. the public) from a wide variety of age groups to understand their perspective on this kind of work. Someone in the question period called this approach an “indigenous approach” to conducting online research. I think this is not an unreasonable strategy: different communities definitely do have different kinds of conventions, and “rules of engagement” in one context may not apply in another online context.
It may ultimately be the case that we need to consider everything to do with the internet in a case-by-case basis–just as we do with “real life” (IRL) observational research today. And, perhaps it shouldn’t be a surprise that we would need to do so.