Using Machine Learning to Advance Personality Assessment and Theory

Wiebke Bleidorn, Christopher James Hopwood
2018 Personality and Social Psychology Review  
Machine learning has led to important advances in society. One of the most exciting applications of machine learning in psychological science has been the development of assessment tools that can powerfully predict human behavior and personality traits. Thus far, machine learning approaches to personality assessment have been focused on the associations between social media and other digital records with established personality measures. The goal of this paper is to expand the potential of
more » ... ne learning approaches to personality assessment by embedding it in a more comprehensive construct validation framework. We review recent applications of machine learning to personality assessment, place machine learning research in the broader context of fundamental principles of construct validation and provide recommendations for how to use machine learning to advance our understanding of personality. MACHINE LEARNING AND PERSONALITY ASSESSMENT 3 Machine learning has led to remarkable advances in society including self-driving cars, speech recognition tools, and an improved understanding of the human genome. One of the most exciting applications of machine learning in psychological science has been the development of assessment tools that can predict personality traits using digital footprints such as Facebook (Youyou, Kosinski, & Stillwell, 2015) or Twitter profiles (Quercia, Kosinski, Stillwell, & Crowcroft, 2012). Machine learning approaches to personality assessment involve automated algorithms for data extraction, cross-validation, and an emphasis on prediction, as described in detail below. These methods begin by gathering a large number of digital records with little or no relation to established theory to create scales that are associated with individual differences in enduring patterns of thoughts, feelings, and behavior (e.g., Funder, 1991; Tellegen, 1991) as assessed by more traditional personality measures. To do this, machine learning approaches focus on identifying empirical associations between digital records and established personality trait measures within specific samples. This strong empirical and mostly a-theoretical focus has led to the development of potent assessment tools that can be used to reliably predict induvial differences in personality traits. However, relatively little is known how these scales can be used to advance our understanding of personality constructs and human behavior. Machine learning approaches offer an unprecedented opportunity to advance both personality assessment and theory. The purpose of this paper is to embed the principles of machine learning approaches to personality assessment in a construct validation framework that is concerned with both predicting and understanding human behavior (Cronbach & Meehl, 1955; Loevinger, 1957) . MACHINE LEARNING AND PERSONALITY ASSESSMENT 4 We first describe the basic approach to using machine learning algorithms for personality assessment and review recent research that has used this approach. We next situate the findings of these studies within the broader principles of construct validation theory. We emphasize how this theory can supplement the focus on prediction characteristic of machine learning research with attention to other aspects of measurement, such as content, structural, external, and discriminant validity, and argue that doing so would appreciably enhance the potential of machine learning to generate novel tools and insights about personality traits. We conclude with nine specific recommendations for how to integrate machine learning techniques for personality assessment within a construct validation framework.
doi:10.1177/1088868318772990 pmid:29792115 fatcat:cjq4rrt5ezaoznicwtgu7tbx7a