In addition, as discussed by Smith and colleagues (2009), the unidimensionality of scales are optimal when attempting to examine the convergent and discriminant relations between psychopathic traits and relevant external criteria. However, conformity to the strict specifications of SEM (e.g., confirmatory factor analysis) is by no means the sole basis for evaluating the internal structure and construct validity of a measure. Moreover, we do not mean to imply that instruments that allow for the construction of latent variables are necessarily better or more valid than ones that rely on manifest variables. Indeed, numerous scholars have noted that some omnibus personality inventories are a poor fit for confirmatory factor models (Neumann et al., 2008), often those based on exploratory factor models, which nonetheless show some replicable EFA results across multiple samples, and demonstrate criterion validity (Church & Burke, 1994; Grucza & Goldberg, 2007; Hopwood & Donnellan, in press; McCrae et al., 1996). Still, SEM-based approaches have broad measurement and structural implications (e.g., clear item-to-factor specification); therefore, further application of these approaches in developmental studies of psychopathy is recommended as it can more precisely estimate the magnitude of stability of psychopathic traits (as measured and conceptualized for a particular instrument).
