Comparing the measured and latent Dark Triad : are three measures better than one?

Could measurement level be a factor worth considering when studying the Dark Triad (i.e., narcissism, psychopathy, and Machiavellianism)? In two studies (N = 465), we compared the relative fit of two Dark Triad models: one that treats the three measures as separate-yet-related personality traits and another that treats the measures as tapping a single, latent construct. Mid-level personality traits, such as mate-retention strategies (Study 1) were best explained by a three-measure model, whereas the higher-order trait of sociosexuality (Study 2), were best explained by a single, latent-factor model. When considering mid-level measurement in personality, the three traits may provide independent effects for interpersonal relationships, whereas at the higher-order level, the three traits may function as a single entity relating to other higher-order traits. We suggest one should consider level of measurement between the predictor and criterion variables to better predict correlations among variables such as the Dark Triad.