Date of Submission

5-2025

Document Type

Thesis

Degree Name

Master of Science in National Security

Department

National Security

Advisor

Charles A. Morgan III M.D., M.A.

Keywords

Zero-Acquaintance Assessments, Short Dark Triad (SD3) Personality Test, Professional Raters, Non-Professional Raters, Undercorrection

LCSH

Psychological tests, Impression formation (Psychology), Machiavellianism (Psychology), Narcissism

Abstract

This thesis addresses the efficacy of Zero-acquaintance assessments of Dark Triad personality features (i.e., Machiavellianism, Narcissism, and Psychopathy as measured by the SD3 instrument). The methodology involved two phases. In Phase One, 1,345 participants completed online personality assessments, including measures of the SD3. In Phase Two, 386 participants engaged in three videotaped laboratory tasks: a rapport-building interview, a mapping task, and a decision-making role-playing scenario. 60 of these videos were randomly selected for analysis. Each was rated by four operational psychologists and four undergraduate students who were not studying psychology. The accuracy of their ratings was determined by comparing their assessments to the ground truth scores derived from the participants’ SD3 selfreport questionnaires. Both groups of raters performed slightly better than chance at assessing Machiavellianism and Psychopathy, but not Narcissism. Non-professional raters were slightly more accurate than professional raters in assessing Psychopathy and Machiavellianism, with nearly identical accuracy for Narcissism. Across all traits, the most common error was that of undercorrection (rating a trait as low when the ground truth for a trait was high). The thesis concludes that human raters with limited specific training can achieve some level of accuracy in identifying Dark Triad traits under Zero-acquaintance conditions. Nonprofessional raters may outperform trained professionals in certain areas. This thesis suggests that assessments of SD3 traits may be successfully performed by analysts who are not psychologists. Future research is recommended to further explore these findings with more targeted tasks, diverse samples, and comparisons with artificial intelligence models.

Available for download on Tuesday, May 07, 2030

Share

COinS