Naïve Rating of Big Five Personality Traits Using Text Data

• In Event: Personality and Politics

Abstract

In recent years, researchers have trained predictive models of personality traits using text data, allowing for the personality traits of elected officials to be estimated with speech data, which is often widely available. This study aims to study whether naïve raters (that is, the voting public, not expert observers) are able to evaluate the psychological characteristics of the authors of text based on their text extracts alone. Using a sample of 500 adults in the United States, we provide respondents with pairwise comparisons of short text extracts written by authors with measured Big Five personality traits, and examine the ability of naïve raters to accurately evaluate author personality trait differences.

Authors

• Jonathan D. Klingler, University of Mississippi

• Gary E. Hollibaugh, University of Pittsburgh

• Adam Ramey, New York University Abu Dhabi