TY  - JOUR
TI  - Predicting Psychological Characteristics from Digital Footprints
T2  - Psychology. Journal of Higher School of Economics
IS  - Psychology. Journal of Higher School of Economics
KW  - digital footprints
KW  - prediction
KW  - personality traits
KW  - emotions
KW  - depression
KW  - values
AB  - The article discusses the prediction of individual psychological characteristics (personality traits, emotional states, values, motives, etc.) based on person’s digital footprints. As studies have shown, such characteristics can be very accurately detected on the basis of various types of digital footprints: texts, images, Internet-surfing features, the nature and duration of phone calls, "likes" (I like), financial transactions, and changes in a person’s location. Most often, to perform this task, textual information is used from a variety of sources (user profiles, blogs, tweets, etc.). With vocabulary-oriented predicting of psychological characteristics, two main approaches to text analysis are used. One, the so-called fixed (closed-vocabulary), uses a limited vocabulary dictionary, and the other (open-vocabulary) uses an unlimited vocabulary dictionary. In the case of a fixed approach, a certain set of words and categories is initially set, the relationship of which with personality traits is revealed. Unlike the fixed one, in the case of using the open approach, there is no predefined list of words, and lexical predictors of personality traits are found directly in the course of text analysis. The greatest accuracy of predicting was achieved in the case of personality traits of the "Big Five". According to the degree of success in predicting, they were arranged as follows (from the most successful to the least): extraversion, openness to experience, conscientiousness, neuroticism, agreeableness. Emotional states, values, motives, and life satisfaction are predicted slightly worse. The simultaneous use of several types of digital footprints, as well as more advanced procedures for collecting and analyzing data, can significantly increase the accuracy of the prediction. Immediate and more distant prospects for research in this area are evaluated.
AU  - Vladislav Latyunov
AU  - Victoria Ovsyannikova
UR  - https://psy-journal.hse.ru/en/2020-17-1/359612461.html
PY  - 2020
SP  - 166-180
VL  - 17