Violence is preventable, and the United Nations' Sustainable Development Goals set out violence prevention as a global priority-calling for countries to halve their violent death rates by 2030. Despite action since then, there has been limited progress in reducing violence globally. In this essay, we argue that current violence prevention efforts are being heavily shaped by reductionism-the now-dominant research paradigm across the sciences. We make the case that this reductionist philosophy has prematurely misguided violence research away from studying populations as a whole. We further argue that the mainstream statistical methods in violence research are reinforcing this reductionist bias by oversimplifying cause-effect relationships. After revisiting foundational principles in sociology and public health, and drawing on advances in social epidemiology and complexity science, we suggest that violence-at any level-is better understood as an emergent property of a complex system. We call on the field of violence research to return to a holistic lens to maximize gains in explanatory theory and better position the evidence to directly inform effective intervention strategies for reducing violence at scale. (Am J Public Health. Published online ahead of print February 12, 2026:e1-e10. https://doi.org/10.2105/AJPH.2025.308366).
Esposti et al. (2026) studied this question.