Scientists at the University of Pennsylvania found that they can pinpoint with 92 percent accuracy personality traits and age based on a user's language on Facebook.
The study published by PLOS ONE, had researchers looking at the language, or an "open-vocabulary approach" in 75,000 profiles and found that it could make predictions about each user's profile and be 92 percent accurate predicting gender and age within three years, according to the MIT Technology Review.
Essentially, according to the Review, "they let the data drive which words or phrases were considered most important." Researchers looked at trait markers they already knew.
Each study participant filled out a questionnaire and scored themselves on five personality traits: extroversion, agreeableness, conscientiousness, neuroticism and openness. The scientists then looked at profile updates for language that matched with the test scores. Some of the words were consistent with extroverts, such as "party", while introverts used "anime" or "manga." Those who were more neurotic used "depressed" more often.