Research ethics: Facebook Study

[futureapppsy2]

This brief report (http://www.pnas.org/content/111/24/8788.full.pdf) has got a lot of attention for it's arguable lack of ethics. Basically, researchers randomly selected 680,000+ FB users and then manipulated the content in their newsfeed so that it would more positive or more negative. The outcome variable was the percentage of positive or negative words in people's posts. The researchers have claimed it was existing data use covered by FB's TOS and so they didn't need to acquire informed consent or debrief people (http://www.theatlantic.com/technolo...ooks-mood-study-thought-it-was-creepy/373649/). I don't see how that held up with Cornell's IRB, as they were engaging in active manipulation of participants' experiences without IRB approval. Also, it seems that they really overstate their results. The effect sizes are tiny (toping out at d=.02), and the p values are most likely an artifact of the huge n.

 

[MCParent]

Pretty clearly violates modern concepts of informed consent as a process and not just checking a box on a form once.

 

[WisNeuro]

I wholeheartedly agree about the ethics violation. But, unless people talk with their feet and punish Facebook by quitting the service and hitting their pocketbooks, this will continue to happen. And, I don't see that happening. I imagine Facebook will offer some half-hearted apology to save some face in teh next couple weeks, but that's about it.

This thing happens on a smaller scale with other advertisers. Places like Amazon run tests about things like ad placement, wording, color schemes, etc to see if there are things that make people more likely to buy. Now, those are somewhat different things in severity as Facebook sought to directly alter mood states, but on the same continuum.

 

[deadmau5]

They had to pay facebook for this implementation or for the data, right?