What you Like on Facebook could reveal your race, age, IQ, sexuality and other personal data, even if you've set that information to "private".
A study by Cambridge University in collaboration with Microsoft found that by using the Like data, which is available publicly by default, they could make accurate predictions about personal attributes -- the most surprising being an apparent link between Liking "Curly Fries" and having a high IQ, or Liking "That Spider is More Scared Than U Are" and being a non-smoker.
The research shows that although you might choose not to share particular information about yourself it could still be inferred from traces left on social media, such as the TV shows you watch or the music you listen to or the spiders that are afraid of you.
Working with a dataset taken from 58,000 US Facebook users via the myPersonality app, the psychometrics research team used algorithms to analyse the Likes. The result was a set of statistical models able to infer personal details with a high degree of accuracy.
Obviously some Likes will give more of an indication about their users than others -- Liking a political candidate to reveal their political affiliation, for example -- but the majority of predictions used far more innocuous activity to produce profiles of the users.
The models had a 95 percent success rate in separating African-Americans from Caucasian Americans, 88 percent accuracy for male sexuality, 82 percent for determining Muslims and Christians, while relationship status and substance abuse came in at between 65 and 73 percent.
The fact that rich demographic information can be inferred from publicly available data would be very much of interest to marketing departments, but also to privacy campaigners because of the implications about the difficulty of controlling the data you reveal online. "Similar predictions could be made from all manner of digital data, with this kind of secondary 'inference' made with remarkable accuracy -- statistically predicting sensitive information people might not want revealed," said Michal Kosinski of the Cambridge Psychometric Centre.
He went on to say, "I am a great fan and active user of new amazing technologies, including Facebook. I appreciate automated book recommendations, or Facebook selecting the most relevant stories for my newsfeed. However, I can imagine situations in which the same data and technology is used to predict political views or sexual orientation, posing threats to freedom or even life."
The research feeds into the ongoing discussion of user privacy online, particularly in regard to social networks. Microsoft Research's Thore Graepel noted: "Consumers rightly expect strong privacy protection to be built into the products and services they use and this research may well serve as a reminder for consumers to take a careful approach to sharing information online, utilising privacy controls and never sharing content with unfamiliar parties."
This article was originally published by WIRED UK