Brechas de observabilidad en la vigilancia del riesgo psicosocial laboral en América Latina: fundamentos para una arquitectura dinámica con evidencia desde Chile

Periodic psychosocial risk assessment systems share a limitation that is architectural rather than psychometric: the aggregation function that produces the organizational indicator irreversibly destroys spatial information about the state of the internal units composing it, regardless of instrument quality or measurement frequency. This paper reports a simulation study with empirical calibration that tests four confirmatory hypotheses, declared a priori, using an agent-based model formalized under the ODD 2020 protocol and calibrated with CEAL-SM 2023 data (N=768,505 Chilean workers). The model demonstrates that concentrated deterioration in a minority of organizational units produces a Spatial Dilution Ratio of up to 4.11×, rendering it either fully invisible (existential gap) or detectable but unlocalizable (granularity gap) to periodic aggregate assessment, while uniformly distributed deterioration produces convergence between architectures — confirming by inversion that the gap is spatial, not temporal. This paper proposes DYNOR (Dynamic Organizational Risk Surveillance Framework) to occupy the follow-up step of the standard five-step preventive cycle in Latin America, operating continuously between periodic assessment anchors rather than replacing them. The framework is directly relevant for normative systems in Chile, Colombia, Mexico, Peru, Argentina, Uruguay, and Brazil.