Having analyzed 700 million words and phrases collected from 75,000 volunteer’s Facebook messages, a team of researchers at the University of Pennsylvania is confident in predicting a person’s age, gender and personality type just based on their status updates. In addition, the researchers examined social media on a deeper level for clues of what may appear as linguistic patterns.
This research is part of the World Well-Being Project, which seeks to explore unchartered territories of the psychosocial phenomena of though language analysis. The study is an interdisciplinary effort of the Science, Engineering, Psychology, and Arts departments at the UPENN.
Generating computer models specially modeled to predict individualistic traits, the scientists used this open-vocabulary technique to make connections between personality and behaviors, as well as a measurement of the effectiveness of psychological interventions. With a 92% accuracy rate, the model has proven to demonstrate a consistency in results.
To display their statistically results in a easy to comprehend way, the researchers created word clouds that summarized the types of words used that led to the prediction of a particular trait.
The word cloud visually depicts specific vocabularies used by certain cliques and age groups. For example, researchers grouped teenagers from ages 13 to 18 in one group to get a better understanding of their inner psyche. These teenagers used the words “school”, “homework”, and “tomorrow” more than any other age group, while the 19 to 22 year olds used words like “semester” and “drunk” more. Different groups commanded different vocabularies that demonstrated similarities to their personalities.
If you want to find out more about the project, the research article is can be read in full here.