Modeling the association between and predictors of two constructs of resilience.

ABSTRACT

PURPOSE: Resilience serves as a protective factor against adverse outcomes following exposure to traumatic events. The extant literature focuses on psychiatric resilience in the context of internalizing symptoms, though resilience is also important in relation to externalizing symptoms. Research is needed to clarify the predictors of resilience across contexts. The aims of the current study are twofold 1. Determine the association between psychiatric resilience (PR) and alcohol resistance (AR) and 2. Test for differential prediction of each form of resilience by exogenous predictors. METHODS: The sample (n = 7585) was drawn from the Virginia Adult Twin Studies of Psychiatric and Substance Use Disorders (VATSPSUD). Participants completed measures of internalizing symptoms, exposure to stressful life events, DSM alcohol abuse and dependence symptoms, maximum alcohol consumption, personality variables, and social support. All cross-sectional, structural equation modeling (SEM) analyses were conducted using Mplus software version 8.2. RESULTS: A single common factor model provided adequate fits for both PR and AR. In the full measurement model the correlation between the two resilience factors was estimated (r = 0.28, SE = 0.018, p < 0.001). Neuroticism and mastery predicted AR and PR, but differentially, with a stronger effect size for PR (neuroticism B = 0.35, p < 0.001; mastery B = - 0.36, p < 0.001). The positive social support factor did not predict either resilience variable, while interpersonal conflict was associated with both (AR = 0.09, p < 0.001; PR = 0.07, p < 0.001). CONCLUSIONS: Findings extend the current literature on resilience in two ways. First, rigorous measurement model based definitions of two resilience variables are specified. Second, external validation of the AR and PR constructs is carried out using latent variable modeling techniques. The modest correlation suggests resilience may not be well-characterized by a single general attribute. Findings provide further evidence for predictors of resilience by way of displaying differential patterns of prediction effect sizes of PR and AR.