by Elisa A. Hurley, PhD, Executive Director
Much has been written in the past few months—pro and con—about the results of the Facebook emotional contagion study published in June in the Proceedings of the National Academy of Sciences. The study manipulated the News Feeds of 700,000 unknowing Facebook users for a week in January 2012 by adjusting Facebook’s existing algorithm to over-select for either more positive or more negative language in posts. At the end of the week, the results showed that these users were more likely to follow the trend of their manipulated feed, that is, to use more positive or negative language in their own posts, respectively, based on their study grouping. Additionally, the study revealed that lowering the emotional content of posts overall caused users with affected News Feeds to post fewer words in their own statuses.
The public reaction to the revelation of the study in June was swift, loud, and dramatic. I myself was surprised by the uproar and still am not sure what to make of it.
Those who have written about the study in scholarly and popular media have voiced differing opinions about whether adequate informed consent for the study was provided via Facebook’s Terms of Service, as well as whether informed consent was even needed. Further debate has centered on whether the study required IRB review. And still other commentary has zeroed in on the merits of the research itself. As James Grimmelmann, a law professor from the University of Maryland said (quoted in The Atlantic, June 2014):
[The Facebook study] failed [to meet certain minimum standards]…for a particularly unappealing research goal: We wanted to see if we could make you feel bad without you noticing. We succeeded.
But are these the reasons users have been so incensed? I’m not sure.
Consider that, by their own admission, Facebook routinely manipulates its users’ News Feeds, filtering 1,500 possible news items down to 300 each time a user logs in. Many Facebook users object to this filtering (wanting instead to see everything and choose the content they engage with themselves), but that’s not enough to make the majority of account holders abandon or deactivate their accounts. The algorithm is also used to deliver related advertising content to users, and words in posts are parsed to target that advertising precisely to users’ recent activity: post enough about being on the treadmill, and your ads begin to feature running gear and related products. Yet again, no hue and cry, no mass exodus from Facebook by its billion plus worldwide users.
So it would seem that commercial audience manipulation—the basis for every marketing campaign the world over—is held to a lower standard than the presumably more noble and societally beneficial work of acquiring knowledge for the larger public good. Why is that?
The outcry about the study might be due to several factors: the perceived hubris of publishing a research paper about what perhaps should have remained internal commercial research; the fact that hundreds of thousands of Facebook users are left wondering if they were part of the experiment (as of this writing, there has been no indication that Facebook debriefed the subjects whose News Feed were affected); or the realization by those users and others that Facebook is able and willing to manipulate its user population in a variety of ways, and for purposes other than product enhancement or selling goods and services. In the words of Robinson Meyer (The Atlantic, June 2014):
And consider also that from this study alone Facebook knows at least one knob to tweak to get users to post more words on Facebook. [Author's emphasis]
Perhaps we’re so accustomed to commercial manipulation that the instances that occur in our everyday lives—the placement of items on grocery store shelves, the tempo of music in shopping malls during the holidays, commercials for junk food peppered liberally through children’s television programming—don’t register as manipulative. Perhaps, too, we’re so used to them that we don’t even realize the effects they have on us. Some have suggested that the Facebook study and the public reaction to it should make us question our complacency about how our information is provided to and used by commercial entities. As Janet D. Stemwedel noted (Scientific American, June 2014):
Just because a company like Facebook may “routinely” engage in manipulation of a user’s environment, doesn’t make that kind of manipulation automatically ethical when it is done for the purposes of research. Nor does it mean that that kind of manipulation is ethical when Facebook does it for its own purposes. As it happens, peer-reviewed scientific journals, funding agencies, and other social structures tend to hold scientists building knowledge with human subjects to a higher ethical standard than (say) corporations are held to when they interact with humans. This doesn’t necessarily mean our ethical demands of scientific knowledge-builders are too high. Instead, it may mean our ethical demands of corporations are too low. [Author’s emphasis]
I think this is a point well taken. I also think there is an analogy to be drawn here to our collective attitudes about clinical care versus research. Consider the daily interaction between clinical care providers and patients. Patients trust doctors to make treatment decisions via prescriptions, referrals to specialists, and other interventions—some of which present more than minimal risk to a patient’s life or well-being. But not all doctors are equally knowledgeable, up-to-date on the current research, or without their own biases. And many of those decisions are made without any sort of consent process. It’s only when interventions—and sometimes the very same interventions, as in the case of comparative effectiveness research—are presented within the context of a research study that that the requirements for informed consent, and indeed an entire set of ethical questions and considerations, get triggered.
There are surely good reasons for this. Whether or not research is always inherently riskier to subjects than care is to patients—and I don’t believe it is—the very fact that one is participating in research, an enterprise whose goal is the creation of generalizable knowledge rather than personalized benefit, seems to me good reason for invoking a fairly robust ethical and regulatory machinery (though I acknowledge that the “machinery” we currently have in place may not be a good fit for much contemporary research). To make the parallel point to Professor Stemwedel’s, the fact that we seem to have different ethical standards or thresholds for research than for practice doesn’t, or doesn’t necessarily, mean that our standards for research are too high. Maybe it should, though, raise the question of whether our ethical standards for clinical practice are too low.
So, as with the Facebook case, I am left wondering, do we unfairly hold research to a higher ethical standard than we do clinical practice, or marketing practice? And if so, are we, as some argue, thereby hindering important scientific progress? Or does this highlight that we are we too lax about ethical considerations in other domains? What do you think?
I invite and encourage you to join PRIM&R for a webinar on lessons learned from the Facebook study Thursday, October 30, at 1:00 PM ET. The Future of Internet Research: What We Can Learn from the Facebook Emotional Contagion Study, features Elizabeth Buchanan, PhD, Mary L. Gray, PhD, and Christian Sandvig, PhD, who will discuss the study and some of the commonly raised questions pertaining to internet and social media research, including: questions about how to classify social data; the ethical principles that accompany any such classification; how consent and debriefing strategies can be redesigned to scale up beyond the lab or field setting; how minimal risk can be assessed in online settings; and how to determine what constitutes publicly available information, communication, or social interaction versus private information, communication, or social interaction.