Abstract
Research suggests that men and women differ on mean levels of Dark Triad personality constructs such as Machiavellianism, but few studies have investigated whether or not these differences are due to actual latent trait differences or bias in measurement. Further, recent research suggests important challenges associated with existing measures of MACH in terms of overlap with psychopathy and matching expert descriptions. The present study took a recently developed measure of Machiavellianism (the Five Factor Machiavellianism Inventory; FFMI), based on the five-factor model, and examined its invariance across gender. Strong (or scalar) factorial invariance was established, indicating that latent factor means can be compared between men and women using this measure. Mean-level differences showed that men had higher levels of latent factors related to antagonism and social dominance. In terms of total score, men reported significantly higher mean levels of Machiavellianism. The findings of the present study lend support to the notion that mean level differences in Machiavellianism across gender are not artifacts of measurement bias.
Open Scholarship
This article has earned the Center for Open Science badges for Open Data through Open Practices Disclosure. The data are openly accessible at osf.io/ruvwp/?view_only=929e019c5b524ad9b3acceb1055b7be0. To obtain the author's disclosure form, please contact the Editor.
Notes
1 Indicators were treated as continuous variables because all skewness and kurtosis statistics were between -2 and +2, which is considered an acceptable range in order to prove normal univariate distribution (George & Mallery, Citation2010).
2 There is a fourth step that can be included when assessing MI, which is strict invariance. For strict invariance, factor loadings, intercepts, and residual variances are all constrained between groups. If invariance holds at this level, both means and variances in FFMI can be explained by gender. Because the goal of the study was to assess whether Machiavellianism scores could be appropriately compared across men and women at the mean level, this fourth step was omitted.
3 The determinacy coefficients for the factor scores ranged from .89 to .95. Correlations between estimated factor scores and latent factors ranged from .80 - .90. Estimated F1 was correlated with estimated F2 and estimated F3 at -.17 and .22, respectively. Estimated F2 and estimated F3 were correlated at -.32. Latent F1 was correlated with latent F2 and latent F3 at -.16 and .15, respectively. Latent F2 and latent F3 were correlated at -.26.