Examining the Relationships Between the Triarchic Psychopathy Constructs and Behavioral Deviance in a Community Sample

Few studies have examined the extent to which psychopathic traits relate to the commission of mild to moderate acts of deviance, such as vandalism and minor traffic violations. Given that psychopathy is now studied in community populations, the relationship between psychopathic traits and less severe deviant behaviors, which are more normative among noninstitutionalized samples, warrants investigation. The current study examined the relationships between the triarchic model of psychopathy (Patrick, Fowles & Krueger, 2009) and seven forms of deviant behavior (drug use, alcohol use, theft, vandalism, school misconduct, assault, and general deviance) in a nationally representative sample. Triarchic disinhibition positively predicted each form of normative deviance. Boldness positively predicted drug and alcohol use as well as general deviance, while meanness negatively predicted school misconduct. Boldness and disinhibition also positively predicted overall lifetime engagement in deviant behavior. Implications are discussed, including support of the role of boldness within the psychopathy construct.

Escaping the snare of chronological growth and launching a free curve alternative: General deviance as latent growth model

Researchers studying longitudinal relationships among multiple problem behaviors sometimes characterize autoregressive relationships across constructs as indicating “protective” or “launch” factors or as “developmental snares.” These terms are used to indicate that initial or intermediary states of one problem behavior subsequently inhibit or promote some other problem behavior. Such models are contrasted with models of “general deviance” over time in which all problem behaviors are viewed as indicators of a common linear trajectory. When fit of the “general deviance” model is poor and fit of one or more autoregressive models is good, this is taken as support for the inhibitory or enhancing effect of one construct on another. In this paper, we argue that researchers consider competing models of growth before comparing deviance and time-bound models. Specifically, we propose use of the free curve slope intercept (FCSI) growth model (Meredith & Tisak, 1990) as a general model to typify change in a construct over time. The FCSI model includes, as nested special cases, several statistical models often used for prospective data, such as linear slope intercept models, repeated measures multivariate analysis of variance, various one-factor models, and hierarchical linear models. When considering models involving multiple constructs, we argue the construct of “general deviance” can be expressed as a single-trait multimethod model, permitting a characterization of the deviance construct over time without requiring restrictive assumptions about the form of growth over time. As an example, prospective assessments of problem behaviors from the Dunedin Multidisciplinary Health and Development Study (Silva & Stanton, 1996) are considered and contrasted with earlier analyses of Hussong, Curran, Moffitt, and Caspi (2008), which supported launch and snare hypotheses. For antisocial behavior, the FCSI model fit better than other models, including the linear chronometric growth curve model used by Hussong et al. For models including multiple constructs, a general deviance model involving a single trait and multimethod factors (or a corresponding hierarchical factor model) fit the data better than either the “snares” alternatives or the general deviance model previously considered by Hussong et al. Taken together, the analyses support the view that linkages and turning points cannot be contrasted with general deviance models absent additional experimental intervention or control.