A Likely Story

July 8, 2013

Did you know that each time you hit the “like” button on Facebook you are building a predictive data set that can reveal sensitive personal information about you?

Issues of data privacy have been dominating the news lately. Edward Snowden sits holed up in the Moscow airport for revealing the U.S. government’s routine collection of our citizens’ cell phone records. Doesn’t it seem doubtful that the government stopped at cell phone records?

Sociologists often say Facebook is the world’s greatest social science laboratory. No other environment offers as much behavioral data, ripe for the picking and waiting to be analyzed. So, with the help of Google Scholar, I came across a study conducted by a group of Cambridge University sociologists and published earlier this year in the “Proceedings of the National Academy of Sciences.”

In a nutshell, the study tracked 58,000 volunteers who allowed the sociologists to record each time they “liked” something (anything) on Facebook. They used a logistic/linear regression model (whatever that is) to prove that liking data accurately predicts sexual orientation, ethnicity, political affiliation, religion, substance abuse, intelligence, age, parental separation and gender.

Does it work? Kind of. The chart below shows the prediction accuracy expressed as a percentage. For example, the model predicts with 95% accuracy whether a person is Caucasian or African American.

Cambridge study

Here’s the parlor game part of the study — the best predictors of a high IQ were liking such things as “The Colbert Report,” thunderstorms, science, and curly fries. On the flip side, the best predictors of below average intelligence included liking: “I Love Being A Mom,” Harley Davidson and Lady Antebellum.

The study’s authors believe Facebook data offers advantages over traditional psychological inventories because users can’t game the system by trying to out-smart the questions. The downside — privacy. We all agreed to the terms and conditions on Facebook, which means they own the data and when the government comes asking for it, apparently the Patriot Act says “here you go, take all the data you want.” So think twice the next time you hit the ”like” button on Facebook, you never know what it might reveal about you — and to whom.