In a study published this week, and one which is sure to re-ignite the Facebook-privacy fire, researchers have shown that Facebook users are unwittingly revealing all about themselves through their likes.
Funded by Boeing and Microsoft Research, the study shows that Facebook can be used to accurately predict a wide range of personal attributes including: sexual orientation, ethnicity, religion, political views, personality traits, use of addictive substances and age and gender. It can even tell researchers whether or not your parents were separated by the time you were 21.
The average person has 170 likes on their profile and it demonstrates how, through looking at the very basics of information a person provides online, an automatic and accurate estimation of what that person is like unfolds.
Although one could say the same could be said by looking at a person’s Google history, Web browsing or Twitter subscriptions, the only difference here is Facebook likes are currently publicly available by default.
58,466 volunteers took part in the study offering up their Facebook like data obtained through a Facebook application.
The study was able to correctly predict if:
- 95% of the cases were African American or Caucasian American
- 93% of the cases were female or male
- 82% of the cases were Christian or Muslim
- 67% of the cases were single or in a relationship
- 73% of the cases were smokers or not
- 65% of the cases were drug users or not
- 60% of the cases were the users parents were separated by the time they were 21 or not
It goes on to show how personality traits were a little more difficult to predict accurately with them only being able to predict correctly:
- The emotional stability of users in 30% of the cases
- The openness of users in 43% of the cases
- The satisfaction with life of users in 17% of the cases
To explain the lower accuracy for the ‘satisfaction with life’ trait, researchers say, it’s difficult separating long-term happiness with varying mood swings on Facebook.
People post less when they’re happy but give them something to complain about and off they go like a rocket.
It also shows that people who like ‘Hello Kitty’ tend to be high on Openness and low on Emotional Stability.
They’ve set up a website where users can test out the personality traits aspect for themselves at youarewhatyoulike.com. This was my result taken from my 120 likes:
Helpfully, it told me this:
I don’t ever remember liking Jeff Healey. I do like him but I don’t remember liking him. The others I’ll cop to that yes, I do like them. Crowdrise is awesome. You should all check it out when you‘re done reading this.
With a Facebook ‘like’ button adorning almost every website (and growing) it can be hard to avoid the pull of clicking like to express your feelings.
Intriguingly, the study demonstrates how individual likes and traits correlate. For example, they found the best predictors of high intelligence include ‘thunderstorms’, ‘curly fries’ and The Colbert Report’. Good predictors of male homosexuality were ‘No H8 Campaign’, ‘Mac Cosmetics’ and ‘Wicked the Musical.’ Liking ‘Britney Spears’ and ‘Desperate Housewives’ were considered moderately indicative of being gay whereas people who liked ‘Being Confused After Waking Up From Naps’ were most likely to be straight men.
It’ll be interesting to see, moving forward with this study, if businesses start using it to tailor their advertisements and services more to the individual as oppose to the whole and how the publication of this study will serve to push the debate on privacy in the Digital Age.
If anyone wants to check out the study, it’s called: ‘Private traits and attributes are predictable from digital records of human behavior (2013)‘ – Michal Kosinski, David Stillwell, Thore Graepel published by Proceedings of the National Academy of Sciences (PNAS) journal