The Upside to the Facebook and OkCupid Experiments

 

By Allan Fromen

You may have seen the recent announcement by OKCupid (the dating site), where they admitted –bragged actually – that they regularly experiment on their users. This was a well designed PR ploy, riding the coattails of the widespread and indignant criticism Facebook received for manipulating the newsfeed of its members to see if subsequent posts would be more positive or negative.

Some have denounced these experiments as unethical, noting that there was no informed consent or IRB (internal review board), both of which are common in academia. But on college campuses where informed consent is received, the basis of that consent is generally a subterfuge so the subject (student) is not tipped off to the experiment’s true objective. So while the student must provide consent, they do so under false pretenses, without knowing the ultimate purpose of the experiment.

Why the deception? Because we know that if subjects were informed of the true objective, their behavior would probably change in important ways. The psychological literature is littered with biases where subjects behaved differently after learning they were part of an experiment. Perhaps the most famous of these is the classic Hawthorne Effect, where worker productivity increased as a result of being observed.

Consumers today already regard social media companies with some cynicism, as it becomes increasingly obvious that their personal information is being used to fuel targeted advertising. But as a researcher, I am excited that Facebook and others can run experiments with consumers in their natural setting. As OKCupid’s cofounder stated, “once people know that they’re being studied along a particular axis, inevitably they’re gonna act differently.” While not exactly articulately stated, I wholeheartedly agree with the sentiment. The fact that companies can now further our understanding of human behavior, outside the confines of an artificial lab setting with creative tricks to mask the experiment’s true purpose, is a significant opportunity for scientific advancement.

Google runs over 20,000 search experiment a year, Netflix constantly tweaks its recommendation algorithm, and Twitter uses a mix of humans and analytics to improve its service. We are the subjects of Big Data analytics whether we like it or not. While I recognize the ethical concerns, they are outweighed by my enthusiasm for greater scientific insights and optimism for potential new discoveries. In that regard, I applaud Facebook and OkCupid for releasing the results of their studies, and wish more companies would do the same.