Dating site claims "iPhone users have more sex" than other smartphone users

OkCupid, an American dating site, claims to have statistical information that shows iPhone users have twice as much sex as Android users. It claims this after correlating information from over half a million profile images uploaded on the site, and the 11.4 million responses generated by them. Confused as to the link between a particular smartphone’s user’s sex life and the camera used to take the user’s profile picture? Join the club, and make insightful remarks about correlation not being synonymous with causation, in the below comments section.

Instead of focussing on the context of the profile photo, OkCupid has used flash, focus, aperture, and time of day as parameters of measurement. Here are the steps of their ‘experiment’, paraphrased by us:

1. Collected 552,000 example user pictures.

2. Paired them up (randomly?) and asked users to make snap judgments, like so:3. Collated millions of judgments with EXIF data of the photograph, including flash, focus, aperture, and time of day

4. Made rather uncontrolled statistical statements, without ensuring other variables remain the same, or if causation is actually equal to correlation in this case.

These are their findings:

Panasonic > Canon > Nikon

(type and brand of camera you use have an effect on how attractive you look)

Okay, so as OkCupid claims is evident by the graph, better cameras make you look better! Or maybe, if you are better looking you sailed through life with a smile and so could afford a better camera? Or if you are better looking, you are able to snag a partner with enough money to buy a better camera? Seem absolutely ludicrous? That’s the problem with such stats. They don’t prove anything except the ingenuity of the human mind. Using three pictures of one user taken by three different classes does NOT eliminate all other variables. EXIF data, the entire basis of their statistical avalanche, is by itself considered error-ridden!

How did OkCupid hit upon this amazing statistic? Let’s just see shall we? Using the same age group for the data of all three major phone brands, of 30 years, OkCupid used the above photo- attractiveness data, and then “crossed all kinds of user behaviours with the camera models, and found [they] had data on the number of sexual partners for 9,785 people with smartphones. We dropped what we found into Excel, and voila.” Yes, that explanation really helps us understand how the site went from photo-attractiveness of the profile images clicked by each camera phone to the number of sexual partners per smartphone user. In fact, if one looks at the photo-attractiveness by camera brand and class graph again, we see that amongst the three major smartphone brands, BlackBerry phones take the least attractive photos, iPhones take the most attractive, while Android phones fall somewhere in between. The number of sexual partners per smartphone user graph however (above), shows Android below BlackBerry in numbers. Puzzling, and again begging the question: exactly what other data they used to collate the results?

Perhaps the most telling mistake about the post and the statistics is: even if we accept iPhone users have more sexual partners than Android or BlackBerry users, this does not mean they have “more sex”. Without being overly elaborate and risk offending our more sensitive readers, we can safely say promiscuity is not directly related to frequency. In fact, the title of the post, the wording of the inference, and even the axes of the graph, are misnamed: ‘Sexual Activity’ is once again not directly proportional to ‘avg. number of sexual partners’. Yes, even if you ARE aged 30.

The rest of the post was more information correlating different camera parameters and the photo- attractiveness, from time of day to flash and aperture range. The post ends by advising you to buy a better camera, take pictures in soft light - pictures which focus on you more than the background - without flash, preferably in the afternoon (though if the charts are anything to go by, then take pictures at either 6AM or 5PM), and then post those as your profile pictures. And then, inexplicably but predictably, it asks you to go buy an iPhone.

Apparently OkCupid has a daughter site called OkTrends, which compiles such data regularly, and presumably ( if the above claim is anything to go by) every so often announces some rather arousing inferences. Maybe they should stick to making dates, not data. You say the world doesn’t need another statistician? We ask - just how many is too many?