Facebook Knows If You're Gay, Use Drugs, Or Are a Republican

Hugh Pickens writes writes "Not that there's anything wrong with that — as the Guardian reports that Facebook users are unwittingly revealing their sexual orientation, drug use and political beliefs– using only public 'like' updates. A study of 58,000 Facebook users in the US found that sensitive personal characteristics about people can be accurately inferred from information in the public domain. Researchers were able to accurately infer a Facebook user's race, IQ, sexuality, substance use, personality or political views (PDF) using only a record of the subjects and items they had 'liked' on Facebook – even if users had chosen not to reveal that information. 'It is good that people's behavior is predictable because it means Facebook can suggest very good stories on your news feed,' says Michal Kosinski, 'But what is shocking is that you can use the same data to predict your political views or your sexual orientation. This is something most people don't realize you can do.' For example, researchers were able to predict whether men were homosexual with 88% accuracy by their likes of Facebook pages such as 'Human Rights Campaign' and 'Wicked the Musical' – even if those users had not explicitly shared their sexuality on the site. According to the study other personality traits linked to predictive likes include for High IQ — 'The Godfather,' 'Lord of the Rings,' 'The Daily Show'; for Low IQ — 'Harley Davidson,' 'I Love Being A Mom,' 'Tyler Perry'; and for male heterosexuality — 'Wu Tang Clan,' 'Shaq,' and 'Being Confused after Waking Up from Naps.' Facebook's default privacy settings mean that your 'likes' are public to anyone and Facebook's own algorithms already use these likes to dictate what stories end up in users' news feeds, while advertisers can access them to determine which are the most effective ads to show you as you browse."

Comments

Also

[mozumder]

You can tell that just by talking to people.

Re:Also

[mozumder]

and i'm pretty sure your credit card companies already knows these things just from your purchase habits.

But

[Master Moose]

The highest intelligence indicator were the users who ignored everything, revealed very little and never "liked" anything - knowing that anything they did on facebook would be mined and used for metrics and marketing.

Re:But

[dwywit]

Wouldn't the "highest intelligence indicator" be applied to those who don't "do" facebook, twitter, etc?
 
Or did I just miss something flying over my head?

Re:But

[Anonymous Coward]

Now not having a Facebook account is treated by HR departments as suspicious behavior.

Frankly, any HR department that really thinks this can go fuck itself with an iron stick.

I think this is a myth the media is pushing to try to get more people on Facebook.

Re:But

[DerekLyons]

The highest intelligence indicator were the people who never joined facebook and want nothing to do with it.

Three quarters of the Mensa members I know use Facebook. I know a lot of *very* smart and intelligent people, and the vast majority of them are on Facebook too.
 
Seriously Slashdot, get the fuck over yourself - this ignorant bias against anyone who uses Facebook doesn't make you intelligent, it makes you look like a jackass.

Re:Knows and Presumes are not the same thing

[UneducatedSixpack]

One day you will "like" something wrong...

Base rates

[jon.willits]

I can do better than 88% accuracy at guessing if people are homosexual by guessing "no" every time.

Re:Knows and Presumes are not the same thing

[ShanghaiBill]

Your analysis of their analysis proves that you can't analyze analyses. Being right 88% of the time means being wrong 12% of the time. It's not special that they aren't right for you.

Also, being "right" 88% of the time is meaningless unless you break out false positives and false negatives. It is estimated that 5-10% of the population is gay. So I could just predict that everyone is straight, and I would be "right" more than 90% of the time.

Re:Knows and Presumes are not the same thing

[pclminion]

The more they mine data, the more they are polishing my turds.

You're just an outlier in the data. Easily identified, easily filtered out...

Coming up with a profile that is completely incorrect and undetectably so is far more difficult than just being random and contradictory.

You should reread that article.

[denzacar]

From TFA:

The Williams Institute found that, overall, an estimated 8.2 percent of the population had engaged in some form same-sex sexual activity. Put another way, 4.7 percent of the population had wandered across the line without coming to think of themselves as either gay or bisexual.

That same study found out that (from the same FA):

just 1.7 percent of Americans between 18 and 44 identify as gay or lesbian, while another 1.8 percent -- predominantly women -- identify as bisexual.

Basically, that "less than 2%" number is the people who think of themselves as being homosexual or bisexual.
8.2% apparently just like having sex with people of the same sex. Clearly, they're not gay.
Cause they don't identify with being gay.
Cause it's all about identifying.

That's why I always identify myself with Superman.

Re:You should reread that article.

[Zumbs]

8.2% apparently just like having sex with people of the same sex. Clearly, they're not gay. Cause they don't identify with being gay. Cause it's all about identifying.

That is not what your quote is saying. It is saying that 8.2% of the population has at some point in their life experimented with same-sex sex. It does not say if they liked it nor if it was more than once. It is like going to Church a few times is not the same as being a Christian.

Re:Knows and Presumes are not the same thing

[thesandtiger]

Yep.

Any well designed social psych/sociology research project will have tons of ways to check for validity and consistency of data, and the more clever ones will even have ways of identifying the particular ways people will fuck with data and developing a partial profile there, too.

The vast majority of the data will be a fairly accurate representation - the user base is so large that a few "clever" people trying to piss in the well won't have any effect - they aren't even a blip - while the rest of the userbase doesn't see much point in liking random things or going against the established function of the systems.

As to the study itself - I think it will be interesting to see how the profile for any given demographic shifts over time as various things become more or less mainstream and more or less strongly associated with various demographic buckets.