They say you shouldn’t judge a book by its cover, but what’s to stop you from building a model that looks at the covers of tens of thousands of books, cross-references that data with said books’ content, and draws its own conclusions about how the two are related? Researchers from Stanford University and the University of Cambridge did something similar in 2015 with 70,000 Facebook users who took a personality test online by comparing their results to their Facebook likes and finding correlations.
The significance is that, as explained by the New York Times in a recent article, the researchers could guess how users scored on the personality test—which measures openness, conscientiousness, extroversion, agreeableness and neuroticism—based solely on what music, movies, hobbies, and TV Shows they “liked” on Facebook.
The study is in the news because Cambridge Analytica—the sketchy consulting firm that convinced Facebook to hand over data on more than 50 million users and was hired by the Trump campaign to aid them in influencing voters in 2016—based their models for deducing voters’ personalities off of the ones used by these Stanford and Cambridge researchers.
Here’s a breakdown of how their study worked: Cambridge’s Psychometric Center had a 100-question personality test that scored participants on the aforementioned traits: openness, conscientiousness, extroversion, agreeableness and neuroticism (or Ocean). Tens of thousands of people took this quiz through a Facebook app called myPersonality—and many of those people granted the app access to their personal Facebook data, as well as the data of their Facebook friends—so the researchers were able to build a model comparing Facebook likes to Ocean scores.
To test their accuracy, the team asked friends of Facebook users who had taken the myPersonality test to assess their friends on the same metrics. They found that, using 10 Facebook likes, their model was better at guessing a respondent’s personality traits than a work colleague. Using 70 likes, it was better than a friend or roommate. Using 150, it was better than a family member, and using 300, better than a spouse.
So what did they learn about people? The Times has a neat chart on this you can check out here, but here are the music takeaways: People who like Tom Waits and Bjork are some of the most open to new experiences, whereas people who like bro-country singer Luke Bryan, Cheryl Cole, and Jason Aldean are more likely to be less open. Those who like post-hardcore/emo bands Bring Me the Horizon and Escape the Fate are more likely to be less conscientious (they are also more likely to be more neurotic). Those who rage to DJ Pauly D and Waka Flocka Fame are more likely to be more extroverted! If you like Casting Crowns and Relient K, you are probably very agreeable, versus if you like Marilyn Manson, Rammstein, Placebo, and Judas Priest, then you’re probably much less agreeable.
Also, if you like the Smiths, you’re probably more neurotic than most—which is not exactly shocking, and makes you wonder if these models are just proving things we already suspected of others with big data.