Commentary: A robust data-driven approach identifies four personality types across four large data sets

Abstract

What kinds of personalities do humans have? Can these personalities be classified into several discrete types? These issues have been of considerable concern as they could potentially provide deeper understanding of the nature of human individuality and mental disorders. Recently, Gerlach et al. (2018) addressed these issues by applying established machine-learning techniques to big data (more than 1.5 million respondents in total). They found four “meaningful clusters” in personality dimensions, suggesting the existence of at least four personality types. Here, we propose an alternative interpretation of their result: a skewed distribution with no cluster structures in personality space can erroneously lead to the seemingly meaningful clusters.