Background
The United States experiences more mass shootings than any other nation in the world. Various entities have sought to collect data on this phenomenon, but there is no scholarly consensus regarding how best to define mass shootings. As a result, existing datasets include different incidents, limiting our understanding of the impact of mass gun violence in the U.S.
Methods
We compared five datasets of mass shootings for each year included in five databases (2013-2020) and identified overlaps between each database's incidents. These overlaps and divergences between datasets persisted after applying the strictest fatality threshold (four or more) in mass shootings scholarship and policy. Findings: The datasets collectively include 3155 incidents, but the number of incidents included in each individual dataset varies from 57 to 2955 incidents. Only 25 incidents (0.008% of all incidents) are included in all five datasets. This finding persists even when applying the strictest criteria for mass shootings (four or more fatalities). Interpretation: Data discrepancies prevent us from understanding the public health impact of mass gun violence. These discrepancies result from a lack of scholarly consensus on how to define mass shootings, likely the downstream consequence of the politicization of gun violence research. We argue for a broad definition of a mass shooting and a government-supported data collection program to remedy these discrepancies. Such steps can improve the quality of research and support policy-making and journalism on the subject. Funding: This research was supported by the Pahl Initiative on the Study of Critical Social Issues, University of California, Santa Barbara.