Examining the Effects of Controlling for Shared Variance among the Dark Triad Using Meta-analytic Structural Equation Modelling

Abstract

Multivariate procedures (e.g. structural equation modelling) are essential to personality psychology, but interpretive difficulties can arise when examining the relations between residualized variables (i.e. the residual content of a variable after its overlap with other variables has been statistically controlled for) and outcomes of interest. These issues have been the focus of recent debate within the research literature on the Dark Triad, which is a collection of interrelated but theoretically distinct personality constructs made up of narcissism, Machiavellianism and psychopathy. The present paper highlights previous work on the issue of partialling and also makes use of recent developments surrounding meta-analytic structural equation modelling to reliably assess the impact of partialling on the empirical profiles of the Dark Triad components. The results show that numerous interpretive difficulties arise after partialling the overlap among the Dark Triad components, most notably for narcissism and Machiavellianism. The results are discussed in the context of contemporary Dark Triad research in addition to discussing the implications for structural equation modelling methods in personality psychology more generally. Recommendations are made for how future research can mitigate the interpretive difficulties that may arise from partialling. Copyright © 2018 European Association of Personality Psychology.