Via The Register, I learned about people getting upset with FaceBook manipulating their news feed for the purpose of a scientific study.
I understand why people are upset. I do not use FaceBook, but I will be upset if FaceBook manipulated my feed if I were on FaceBook, even if it is just one feed. It is an emotion thing. In truth, the scale of manipulation, as experienced by an individual FaceBook, is very small, i.e. average out at most 5. It is not going to change your enjoyment of FaceBook. In fact, bugs and glitches in FaceBook system probably accounts for more feed lost than the experiment.
Most people pick up on the fact that there are several group of people that might have to answer for how they allow the study to proceed to publications. The obvious candidates are FaceBook, the authors themselves and the publishers. Less obvious is the role of the University and Funding Body. We do not know who the funding body is although it might be FaceBook, so that leaves the two universities involved. Both Universities are represented in the form of Ethics Committee. For practical reasons, one of the ethics committee probably refers to the other. Nonetheless, the act of referring means the committee in question had considered and approved the study.
As such, the researcher in me ask this question: Will I, if I am in the Ethics committee considering the research, approve the study?
Here is what I will consider:
First, I will completely underestimate the reaction to the fact that I am approving a project which does not require explicit consent.
Second, the fact that explicit consent is not given will mean I will think very hard about whether it should be approved. Explicit consent is an important pillar for participation in a study. However, it is recognized that explicit consent is not always possible or appropriate and there are specific rules and guidelines for it.
Third, here are the factors that I need to consider to waive explicit consent:
(1) What is the level of harm to the participants? This is the most crucial question. No explicit consent means participating in the experiment is compulsory and participant therefore do not have a choice. Researcher must satisfy me that the harm level is low. In this case, I tend to agree the harm level is very low.
(2) Is the experiment worth waiving explicit consent? I am not sure here. Let’s make it crystal clear, I will not grant waiver simply because it is more convenient for the researcher. I need to see explicit evidence that one cannot formulate a plan that incorporate explicit consent. Getting 155000 participants is one big task, so is managing them and manipulating theirs and only their news feed. However, pit that against the skills and involvement of FaceBook means it is not difficult. The question is therefore whether can the applicant write consent form that the participants can give them informed consent? I think so.
Where I am wavering is will getting explicit consent itself will bias the outcome of the study. It is not unusual in psychological studies to be “economical with the truth”. A good example is the Milgrim experiment where participants had to press a button that will cause pain to another person while in truth, the other person is an actor who is unharmed by the experiment. In this case, I think we can formulate one.
(3) Is there an alternative to this experiment that is acceptable? For this question, I do not know because it requires knowledge about the topic being studied. The big issue here is can the manipulation of users’ news feed be avoided? Any researchers worth their salt will consider studying the news feeds as they are instead of manipulating them. In this case, whether one can achieve the same aim by simply studying the emotional content of individual feeds. I think it can be done. I will much prefers this approach because the experiment step of omitting one feed can bias the result.
In all, I am not sure whether will I reject the study.
As an aside, if you look at Google News Search, it is interesting to see how a story turned sour (from the FaceBook and the paper authors’ point of view) over a few days.