Digital Life

License article

How online 'likes' reveal your politics, sexuality and drug use

If you thought clicking "like" on Facebook posts of cute kittens and Ryan Gosling memes said little about you, think again.

Social psychologists have studied Facebook user's "likes" and found they reveal a startlingly accurate picture of personality traits including sexual orientation, political leanings and drug use.

The findings build on previous research that has shown the seemingly innocuous information people post about themselves on social media and the internet can imply a lot about a person, all of which can be easily accessed by advertisers?.

Scientists from the University of Cambridge studied the Facebook "likes" – a mechanism for users to express their fondness for other user's photos, status updates, or pages – of more than 58,000 Americans, who, as well as giving the researchers permission to analyse their online proles, provided demographic details and sat for psychometric tests. When they entered the data into a mathematical model they found it accurately discriminated between African Americans and Caucasian Americans in 95 per cent of cases; between Democrats and Republicans with 85 per cent accuracy and between gay and straight men 88per cent of the time. It could pick a drug user in 65 per cent of cases.

While the model was better at discriminating between attributes that had only two options, such as gender, it could also predict, with varying degrees of accuracy, personality traits or features that had multiple variables such as age, number of Facebook friends and intelligence.

The model forecast a person's intelligence accurately in about 38 per cent of cases; determined whether they were an extrovert in 40 per cent of cases and estimated a person's emotional stability 30 per cent of the time.


While the ability to infer information from basic digital records could be used to improve products and services, the researchers noted there may be considerable negative implications of the technology because it could be applied to large numbers of people, without their consent, or without them even noticing.

''Commercial companies, governmental institutions, or even one's Facebook friends could use software to infer attributes such as intelligence, sexual orientation, or political views that an individual many not have intended to share,'' said the study's lead author, Michal Kosinski.

The findings are published in the Proceedings of the National Academy of Sciences journal.

'Likes' that predicted high intelligence
- The Colbert Report
- Science
- Curly fries

'Likes' that predicted low intelligence
- I love being a Mum
- Harley Davidson
- Lady Antebellum

'Likes' that predicted male homosexuality
- Mac Cosmetics
- Wicked the Musical

'Likes' that predicted male heterosexuality
- Wu-Tung Clan
- Shaq
- Being confused after waking up from naps