Eduar Antonio Rodríguez Flores
Research and Innovation Department, Universidad Autónoma del Perú, Lima, Peru
Marianella Alicia Suárez Pizzarello
Research and Innovation Department, Universidad Autónoma del Perú, Lima, Peru
Ciro Ernesto Redondo Mendoza
Corporación Universitaria Adventista. Medellín, Colombia
Cristian Fabian Villanueva Bonilla
Corporación Universitaria Empresarial Alexander Von Humboldt. Armenia, Colombia
Luisa Fernanda Ochoa Villegas
Corporación Universitaria Adventista. Medellín, Colombia
Joan Sebastián Arbeláez Caro
Corporación Universitaria Empresarial Alexander Von Humboldt. Armenia, Colombia
Alejandro Hernando Álvarez García
Politécnico Colombiano Jaime Isaza Cadavid, Medellín, Colombia
This cross-sectional retrospective study analyzed secondary data from the PESCOMSE project, comprising 2,961 Colombian students (children: n = 1,228; adolescents: n = 1,733). Social-emotional competencies (SEC) and empathy were measured with the COMSE-N (44 items) and COMSE-A (81 items) questionnaires. Multifactorial ANOVAs—using Welch or Tukey/Games-Howell corrections as appropriate—examined the impact of SEC levels, gender and grade on empathy. Structural Equation Modelling (JASP 0.18.3) assessed latent relationships. Among children, self-efficacy and optimism were the strongest predictors, and the model explained 66 % of empathy variance (? self-efficacy = 0.52; ? optimism = 0.34). Among adolescents, trust, self-efficacy, and prosocial behavior dominated, accounting for 75 % of variance (? trust = 0.57). Girls consistently showed higher empathy. Findings reveal developmental shifts in the predictive architecture of empathy and underscore the need for stage-specific interventions: boosting self-efficacy/optimism in childhood and fostering trust/prosocial in adolescence. Results inform evidence-based socio-emotional curricula and violence-prevention policies in Latin American schools.