Category: Criminal Justice / Forensics
Traditional cognitive-behavioral approaches that focus on distorted thought patterns common to internalizing (INT) symptoms, such as overestimation of danger or responsibility, may not apply across all contexts, individuals, or problems. For instance, incarcerated populations often underestimate danger and display irresponsible attitudes and externalizing (EXT) behavior. Such criminal thinking (CT) patterns have emerged as a key predictor of criminal behavior and recidivism (Walters, 2012) and recently extended to community populations, with evidence for distinct reactive (RCT; e.g., rashness, problem avoidance) and proactive (PCT; e.g., externalizing blame, entitlement) dimensions (Mitchell et al., 2017). Yet the role of CT dimensions in predicting different types of psychopathology is unknown, including INT and EXT symptoms, which prevail across contexts and often co-occur (Kessler et al., 2005). While forensic research frames CT as predicting EXT behaviors, we hypothesized that this link primarily reflects PCT, whereas RCT may also promote INT symptoms, distinctions that could have important implications for CBT interventions.
We tested these predictions in an SEM latent-variable framework with a large sample of college students (N=529, 74% female, age M = 19.28, SD = 1.48) using the Psychological Inventory of Criminal Thinking Styles-Layperson (PICTS-L; Walters et al., 2009). Indices of Antisocial Behavior (Visser et al., 2010) and Proactive and Reactive aggression (Reactive-Proactive Aggression Questionnaire; Raine et al., 2006) loaded on the EXT latent construct, whereas Patient Health Questionnaire (PHQ; Spitzer et. al., 1999) Depression and Anxiety, and Trait Anxiety (State-Trait Inventory for Cognitive and Somatic Anxiety; Ree et al., 2000) loaded onto INT.
Outliers, missing data, and model assumptions were addressed following standard conventions. Cronbach’s alphas for all measures were > .80. To address unreliability in single indices of PCT and RCT, their reliability scores were used to estimate error terms for corresponding latent variables.