Childhood trauma and exposure to violence as predictors of aggression: an explanatory model in peruvian adolescents
Description
Background: Adolescent aggression has been linked to childhood trauma and exposure to violence. However, multivariate explanatory models that jointly examine these factors while accounting for differences by sex and developmental stage remain limited. Objective: To examine the joint explanatory contribution of childhood trauma and exposure to violence to aggression among Peruvian adolescents and to determine whether these associations differ by sex and developmental stage. Participants and Setting: The study included 513 adolescents (52% boys and 48% girls) aged 12 to 17 years (M = 15.00, SD = 1.58) recruited from four districts in Lima Norte, Perú. Methods: Participants completed three standardized self-report measures assessing childhood trauma, exposure to violence, and aggression. Data were analyzed using covariance-based structural equation modeling (CB-SEM), including multigroup analyses by sex and developmental stage. Results: The overall model demonstrated satisfactory fit and explained 76.2% of the variance in aggression (R² = .762). Childhood trauma emerged as a strong and statistically significant predictor of aggression (β = .719, p < .001), whereas exposure to violence showed a weaker but statistically significant association (β = .184, p < .001). Multigroup analyses indicated that childhood trauma had a stronger association with aggression among girls and late adolescents, whereas exposure to violence was more strongly associated with aggression among boys and early adolescents. Conclusions: Childhood trauma and exposure to violence jointly and significantly explain aggression among Peruvian adolescents, although their relative contributions vary according to sex and developmental stage. These findings underscore the importance of considering developmental and sex-specific pathways when investigating adolescent aggression and designing prevention strategies.
Files
Steps to reproduce
Statistical Analysis Following data collection, 87 of the 600 initially obtained protocols were excluded because of elevated scores on the truthfulness, distortion, and CTQ-SF minimization/denial scales. The minimum sample size required for CB-SEM was determined using G*Power version 3.1.9.7, assuming a medium effect size (f² = .15), p = .001, statistical power (1 − β) = .99 (Memon et al., 2020; Faul et al., 2009). CB-SEM was employed, with item parceling performed within each dimension to define the latent variables and simplify the measurement model (Tessler, 2022). Because the data did not satisfy multivariate normality assumptions (Mardia’s test: skewness = 1454.88, p < .05; kurtosis = −14.19, p < .05), the weighted least squares mean and variance adjusted (WLSMV) estimator was used to improve estimation precision and the reliability of model interpretation (Little et al., 2022; Hair et al., 2021). Model fit was evaluated using the following criteria: CFI ≥ .90, TLI ≥ .90, RMSEA ≤ .079, and SRMR ≤ .079, ensuring adequate correspondence between the proposed theoretical model and the observed data (Murat & Toplu, 2020; Xia & Yang, 2019). Structural regression analyses and multigroup measurement invariance testing were subsequently conducted across sex and stage of adolescence. Finally, coefficients of determination (R²) and standardized beta coefficients (β) were examined to evaluate the strength and direction of the relationships among the study variables. All analyses were performed using the open-source software RStudio version 2024.09.0+375 (R Core Team, 2024), with statistical significance established p < .05 (Vidgen & Yasseri, 2016).
Institutions
- Universidad César VallejoLima Province, Lima