Explanatory model of well-being in victims of political violence in Colombia

The present study aimed to validate an explanatory model of well-being among victims of the Colombian armed conflict, integrating variables of national identity, collective self-esteem, social support, social inclusion, and national self-stereotypes. A cross-sectional quantitative, associative, and explanatory design was implemented with 771 victims recognized by the Colombian State, residing in the Department of Atlántico. The sample predominantly consisted of women (72.2%), with a majority of adults over 40 years old, low educational attainment, and a high rate of unemployment. Internationally validated instruments adapted to the Colombian context were applied, including scales for national identification and attachment, collective self-esteem, social inclusion, social support, and subjective, social, and psychological well-being. Linear regression analyses showed limited explanatory power, whereas Random Forest machine learning models demonstrated excellent fit. Collective self-esteem and social inclusion were the most robust predictors of social and psychological well-being, while social support was the main predictor of subjective well-being. Identification with the nation displayed limited explanatory weight. Quantile regression indicated differential effects of the predictors depending on the levels of well-being, revealing, for instance, an ambivalent role of national attachment. The findings suggest that the well-being of victims of political violence is primarily mediated by collective and community resources rather than national identification. Implications are discussed for designing psychosocial interventions aimed at strengthening social cohesion, collective self-esteem, and support networks in vulnerable populations. • Social inclusion serves as a transversal predictor of social, psychological, and subjective well-being. • Collective self-esteem is the most robust predictor of both social and psychological well-being. • Social support primarily predicts subjective well-being among victims of violence. • Random Forest models demonstrate superior explanatory power compared to traditional linear regression. • Quantile regression reveals differential effects of predictors across the distribution of well-being levels.