Beware: Your Phone Data Says a Lot About Your Personality

Psychologists are getting exceedingly good at predicting someone’s personality from their online behavior. In a 2015 study, researchers from the University of Cambridge and Stanford University designed an algorithm to predict personality traits using no more than a record of someone’s Facebook likes. With 10 likes, the algorithm was about as accurate as a coworker. With 70 likes, it was as accurate as a friend. With 300 likes, it was more accurate than a spouse.

New research published in the Proceedings of the National Academy of Sciences pushes this idea even further. A team of data-savvy psychologists found that three of the Big Five personality dimensions — extraversion, conscientiousness, and openness to experiences — could be accurately inferred from people’s smartphone use. The other two dimensions of the Big Five — agreeableness and emotional stability — remain hidden from view, at least for now.

“We assessed personality in terms of the Big Five dimensions, the most widely used and well-established system in psychological science for organizing personality traits,” said the researchers, led by Clemens Stachl of Stanford University. “This taxonomy describes human personality in terms of five broad and relatively stable dimensions: openness, conscientiousness, extraversion, agreeableness, and emotional stability, with each dimension subsuming a larger number of more specific facets.”

Using the Big Five as a starting point, the scientists examined whether people’s smartphone usage patterns — such as their app usage, communication and social behavior, music consumption, mobility behaviors, overall phone activity, and daytime and nighttime phone activity — could be used to accurately predict their personality traits. They found that they were. To be exact, the researchers accurately predicted 57% of personality traits from behavioral patterns derived from people’s cellphone data.

Some personality traits were easier to predict than others. The researchers found "sociableness" to be the most predictable trait and "good naturedness" to be the least predictable trait. Other personality traits that were highly predictable from the cellphone data were "love of order," "assertiveness," "sense of duty," and "self-consciousness."

Which behavioral patterns were most effective at predicting the various dimensions of personality? Among the three predictable dimensions of the Big Five personality taxonomy, the number of nightly calls was the best predictor of extraversion, the use of the weather app in the evening was the best predictor of conscientiousness, and the length of one’s text messages was the best predictor of openness.

The researchers are quick to point out the privacy concerns raised by this research. They state, “We should not underestimate the potential negative consequences of the routine collection, modeling, and uncontrolled trade of personal smartphone data. For example, organizations and companies can obtain information about individuals’ private traits (e.g., the Big Five personality traits), without the personality information ever being deliberately provided or explicitly requested. Mounting evidence suggests that these data can and are being used for psychological targeting to influence people’s actions, including purchasing decisions and potentially voting behaviors, which are related to personality traits.”

Privacy concerns notwithstanding, the researchers also note the benefits of their approach: "On the positive side, obtaining behavior-based estimates of personality stands to open additional avenues of research on the causes and consequences of personality traits, as well as permitting consequential decisions (e.g., in personnel selection) to draw on behavioral data rather than estimates derived from self-report questionnaires, which are subject to a range of biases."