Understanding Donor Motivation through Social Constructs, Machine Learning, and Predictive Analytics: Evidence from Natural Disasters in Chile

This study examines donor motivation in the context of natural disasters by integrating psychological constructs, hierarchical modeling, and predictive analytics. Using structural equation modeling (SEM), multigroup and multilevel SEM, and semi-supervised latent profile analysis, we evaluated the effects of solastalgia, resilience, deprivation cost, and vulnerability on donation intentions among rural communities in Chile. Results indicate that solastalgia and resilience significantly influence prosocial behavior, with egoistic motivation emerging as a strong direct predictor. Multilevel analyses confirmed contextual effects, while semi-supervised learning identified five donor profiles with distinct emotional and motivational patterns. The findings support the integration of psychosocial constructs into humanitarian logistics and offer robust analytical tools to improve resource allocation, reduce material convergence, and enhance disaster response strategies.