Looking for a Date? A Site Suggests You Check the Data

New York Times article Looking for a Date? A Site Suggests You Check the Data (February 12th, 2010) mistakes a correlation between profile pictures on dating websites and interest in the profile for a causation, and then advises readers to adjust their profiles to maximize readership based on these results. Whether the relationship is truly causal or correlational, the article should note that after its publication, the magnitude of the effects it reports should diminish.

If you’re a man, don’t smile in your profile picture, and don’t look into the camera. If you’re a woman, skip photos that focus on your physical assets and pick one that shows you vacationing in Brazil or strumming a guitar.

Those are some of the insights that OkCupid, a free dating site based in New York, has gleaned by using statistical tools to analyze how the mating game plays out on its site.

There are two possible worlds: one in which the relationships the article reports (the effects of pictures on interest in a profile) are causal, likely guided by a signaling model, or they are correlations and the article has falsely suggested causality. In the second case, choosing photos that don't focus on physical assets may be caused by a good education, or a modest upbringing. These factors may be what draws interest to the profile. Then, we should not expect an uneducated person to draw a larger crowd simply by covering up.

In the first case, however, profile picture choice may causally draw interest to the profile. This would be the case if profile picture choice signaled unobservable traits, such as creativity, self-confidence, etc. Those who are not creative would not realize that turning away from the camera may signal artistic flair. Similarly, those who are not self-confident will not realize that they can attract men without showing off their physical assets. Such signaling, however, is only valuable as long as it accurately predicts personality. Once the trade secrets of the artistic and confident are revealed, they become worthless. So, after the article's publication, Corrections expects the value of such signals to diminish.

Consider the example below (click to enlarge):

We see people who are creative are much more likely to look away from the camera for their picture than those who are uncreative. After the signal is made public, and those who want to fake creativity do so, the signal becomes almost meaningless.

Notably, if OkCupid were to succeed in predicting the "perfect" picture, their business would become worthless. The company's product is the ability to learn about the personality of others through their conscious (or subconscious) profile signals. When these signals become worthless, so too does the company.