« Take Your Relationship to the Movies | Main | Marriage vs. The Ph.D. »
Monday
Aug182014

The Ethics of OKCupid’s Dating Experiment

So let’s talk about the elephant in the room. You probably have noticed some widespread media coverage about OKCupid’s “experiment” wherein, to look for patterns in dating behavior, they manipulated aspects of the site without informing users (see OKCupid’s announcement here as well as coverage here and here). This revelation comes in the wake of Facebook’s massive experiment, which attracted similar attention and criticism. Commenters have questioned the ethics of these experiments primarily due to the fact that Facebook and OKCupid users did not know they were participating and did not consent to be in the study—nor were users directly notified about their participation after the experiment ended.

The idea that these large corporations would manipulate people’s emotions or behaviors without telling their users sounds very disturbing to some. But was this really such a big deal? Were these experiments really “unethical”? Let’s examine these issues further.

A Typical Scientific Experiment

It’s very common for relationship scientists to perform controlled experiments, like these, where some people are randomly put into conditions in which they are exposed to a particular stimulus (while others are put in a control condition in which they are not exposed to the stimulus). This allows the research team to see if exposure to the stimulus changes people’s behavior. If the experiment is done properly, participants won’t know which condition they were placed in. 

Before most experiments begin, participants are explicitly told that they are volunteering to participate in a study, and they are told what procedures will be used. Then participants can decide whether or not they want to continue with the study, and know they can quit any time without penalty if they feel uncomfortable. Participants are also told about the potential risks and benefits if they choose to participate, and participants sign a form indicating that they understand all of this information about the study. This process is called informed consent. But sometimes researchers will use deception, when necessary, which is an official scientific term to denote when researchers do not fully describe the purpose of the study before it begins. In such studies, participants are not informed about the nature of the study until afterwards.

OKCupid’s Experiment

OKCupid ordinarily uses a matching algorithm to give people a sense for how much they have in common with other users. The match % you see with other users indicates the percentage of questions that you agreed on (the higher the %, the more you agree on OKCupid’s questions; see more about how they calculate match % here). In their experiment, OKCupid randomly selected some of its users to see a very high match percentage (90%) when, in reality, their actual match was much lower (30%). Importantly, OKCupid users had no idea that the match % they saw was actually fake. Perhaps unsurprisingly, users were more likely to exchange messages with each other when they perceived greater levels of similarity (saw a higher match %), even if they were not actually very similar. This finding replicates many similar studies showing that in general, people like others more if they perceive themselves as more similar to those others (rather than different).1 Read about a classic experiment here that shows this effect. In both of these cases (OKCupid and the psychological research cited in the prior sentence), the experimenters used deception. 

However, the use of deception doesn’t necessarily mean those experiments were unethical. According to standards for experimental ethics, researchers may use deception in their research if the risk of harm to participants is low. A good rule of thumb is whether the potential harm of participating is no greater than what people would ordinarily experience in their daily lives (this is called minimal risk). So in OKCupid’s experiment, the worst that could possibly happen is that a participant goes on an unsuccessful date with someone whom they met on the site. Maybe they saw a high match percentage online but ultimately found they weren’t really that compatible after all once they met in person. This kind of thing happens all the time in the context of people’s normal dating lives.2 I wish I had a dollar for every time I went on a date with someone I met online and walked away thinking, ‘She’s definitely not the one.’

So, essentially, the research ethics in cases like OKCupid’s study can be determined by the harm principle (no harm, no foul, even if some deception is used). This also applies to the Facebook experiment, in which participants were selectively exposed to a greater number of positive or negative posts…but the posts they saw were exactly like the ones they would ordinarily see in the context of everyday Facebook use. There was nothing artificially bad or good about the posts that people saw. And ultimately, the net effect on people’s mood was extremely small; the researchers reported that people’s moods (as measured by the positive or negative content in their posts) changed by less than 1%.3  So people’s outrage may have been a bit of an overreaction.

Researchers are also required to inform participants that they were deceived after an experiment involving deception has ended. In other words, scientists must come clean and admit the true purpose of the study when it’s done. This process is called debriefing. In a typical study, the researchers will debrief each and every participant in the study (sounds intimate; it’s actually quite mundane). It’s unclear whether OKCupid or Facebook debriefed the participants in their studies, although they did an “indirect debriefing” by making their procedures and results publicly known (and perhaps assuming that their users follow the news). While psychologists do publish their findings publicly, we don’t assume that the public reads academic journals, so we must debrief each participant directly.

Of course, a key difference between academic research (at universities and institutions like NIH/NSF) and research at private companies like Facebook or OKCupid is that academic researchers must have their research approved by an ethics board. In the U.S., we call this an Institutional Review Board (IRB). This is essentially a committee of people who oversee and approve studies on humans/animals, with a keen eye on ethical considerations. They do a qualitative “cost-benefit analysis” to see if the risk to participants is less than the beneficial knowledge we would get from the study. Some people claim that the Facebook study probably would have passed IRB approval at a university if it were held to the same standards. Facebook also has its own “internal review process” (their words), but we don’t know if this is similar to university IRBs. OKCupid, as far as I can tell, did not mention any internal review process for ethical considerations in their study. Their ethical defense has been basically to claim that (a) participants’ gave consent to be experimented on when they signed the terms of use (though most people don’t read it fully) and (b) people should assume that they are experimented on when they use the internet at all. This lack of a rigorous ethics board has caused much outcry from academic researchers who do similar studies.

However, it is also worth mentioning that there are no universal standards for ethics boards internationally, or even in America. Some countries don’t even have any IRB process for minimal risk research (they only have IRB for studies with substantial risk, like medical/drug trials). In America the federal government offers general guidelines, but each individual institution that conducts research sets its own ethical standards, and there is substantial variability in what IRBs consider “risk” to participants. Some IRBs are stricter than others; a study may be approved at one university but not allowed at another university. What this means is that ethical standards for risk of harm to participants are very subjective.

In this scientist’s opinion, we should embrace OKCupid’s type of experimentation, because the benefits outweigh the potential costs. Experiments are valuable because they help us understand social behavior in the real world, and because they help us determine causality (even if the experimental effect is small). The OKCupid study allowed them to test: (a) whether people are actually paying attention to their “match percentage” with other site users, (b) whether the manipulated match percentage predicts social attraction better than actual similarity, and (c) whether the past results from hundreds of psych experiments actually generalize to “real-world” attraction (and it does!). I’d like to commend the folks at OKCupid for actually putting their algorithm to an empirical test (something that other online dating companies are really reluctant to do, because it might show that their algorithms are invalid!).

Going forward, perhaps a good solution to people’s ethical concerns is to encourage companies like OKCupid to form research teams with relationship scientists, so that we can collaborate together on ethically sound experiments that will yield more knowledge to its users and the general public. 

If you’d like to learn more about our book, please click here (or download it here). Interested in learning more about relationships? Click here for other topics on Science of Relationships. Like us on Facebook or follow us on Twitter to get our articles delivered directly to your NewsFeed.

1Montoya, R., & Horton, R. S. (2013). A meta-analytic investigation of the processes underlying the similarity-attraction effect. Journal of Social & Personal Relationships, 30(1), 64-94. doi:10.1177/0265407512452989

2Norton, M. I., Frost, J. H., & Ariely, D. (2007). Less is more: The lure of ambiguity, or why familiarity breeds contempt. Journal of Personality and Social Psychology, 92(1), 97-105. doi:10.1037/0022-3514.92.1.97

3Kramer, A. I., Guillory, J. E., & Hancock, J. T. (2014). Experimental evidence of massive-scale emotional contagion through social networks. PNAS Proceedings of the National Academy of Sciences of the United States of America, 111(24), 8788-8790. doi:10.1073/pnas.1320040111

Dr. Dylan Selterman - Science of Relationships articles Website/CV
Dr. Selterman's research focuses on secure vs. insecure personality in relationships. He studies how people dream about their romantic partners and how nighttime dreams are associated with daytime behavior. In addition, Dylan studies issues related to morality and ethics in relationships, including infidelity, betrayal, and jealousy.

image source: onlinedatingmagazine.com Related Posts Plugin for WordPress, Blogger...

PrintView Printer Friendly Version

EmailEmail Article to Friend