Heard Shots – Call the Police? An Examination of Citizen Responses To Gunfire

Abstract

Many crimes go unreported, making the true scope of crime unknown, and criminal justice reform based on potentially limited data. An acoustic gunfire detection system (AGDS) broadens the data available and provides a unique picture of gun use and violence in communities, separate from crime reported by victims. Using data from an AGDS in the City of St. Louis, this study models variation in community rates of calls to the police for gunshots detected. The results provide new insights into the prevalence of gunfire in a high crime community. We find that community residents are more likely to call the police when the incident was a homicide, and communities with a higher proportion of Black residents are less likely to call 911, net of community disadvantage and violent crime. Policies that encourage community building and improved access to the police and technology are discussed.

Keywords:

• Gun crime
• geography
• acoustic gunfire detection systems
• third party reporting

Acknowledgements

Thank you to the St. Louis Metropolitan Police Department for their partnership on this project and for providing data for the analysis.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 More than 86% of incidents involved only one gunshot detection recorded in the dispatch system. This may or may not reflect the number of shots fired, as the gunshot alert is sent from the AGDS company to SLMPD dispatch. SLMPD does not record the number of shots fired in the CAD system.

2 Although we could potentially measure the dependent variable as a count outcome, we choose to use the binary measure, given that police need only one 911 call to be alerted of a crime and the SLMPD call center protocol classified ‘shots fired’ calls as high priority for dispatch and response (i.e., additional calls to 911 did not elevate the police response to a report of gunfire). Further, more than 85 percent of incidents generated either ‘0’ or ‘1’ 911 call, with the remaining 14.4 percent of incidents generating more than one 911 call and only less than one percent generating more than six 911 calls.

3 Records in the computer-aided dispatch data that are associated with an incident report have an incident number that we used to link with officially reported crime data. Only 12 incidents could be linked to a homicide.

4 The month June and year 2018 were selected because the percentage of gunfire incidents that generated 911 calls was lowest for these categories. It is not clear why calls declined in 2018, but it could be associated with a parallel drop in violent crime in these communities.

5 Data on 311 calls are publicly available at https://www.stlouis-mo.gov/data/service-requests.cfm. The total calls for the study period is 579,482.

6 We also estimated random coefficients to test if the community effects differ across block groups, but this did not change the substantive conclusions and we therefore report the simpler results.

7 A map showing all shootings detected 2014-2018 in this area was too cluttered for visualizing spatial clustering and heterogeneity in 911 reporting. We chose instead to display a map of only those shootings occurring during 2017, the year with the highest number of shootings in this area, to visualize spatial clustering.

8 In additional analyses not shown, we included dummy variables for each month and year (excluding June and 2018) and found no substantive differences in the results. We chose to present the simpler models, including only the month with the lowest reporting rate. Controlling for temporal differences (i.e., time, month, year) accounts for potential differences in the number of people that may have been present to hear gunfire.

9 Additional analyses suggests that percent Black population accounts for the relationship between the dependent variable and median age, disadvantage, and population density, attenuating these effects below the threshold for statistical significance.

10 We estimated a random slopes model to allow the community effects to vary across block groups; substantive conclusions did not change so we elected to present the more parsimonious models.

11 We identified some instances in the data when a citizen called 911 to report a shooting that could not be matched to an AGDS alert. It is possible that an AGDS does not detect all incidents of gunfire within the zone boundaries, resulting in missing data, or individuals may have called 911 outside of our identified 30 minute call window. We tested if including 911 calls for shootings that were not detected by the AGDS affected the results. No substantive differences were found, but the possibility of missing data suggests that findings should be met with some caution. The results of these analyses are available by request.

Additional information

Funding

This work was supported by the Bureau of Justice Assistance [2016-DG-BX-K001] Creating Whole Communities Collaboration at the University of Missouri-St. Louis. The opinions and conclusions expressed in this article are those of the authors and do not necessarily reflect the Department of Justice.

Notes on contributors

Beth M. Huebner

Beth M. Huebner is a Professor in the Department of Criminal Justice at the University of Missouri-St. Louis. Her work focuses on the consequences of a criminal conviction, community and institutional corrections, and public policy.

Theodore S. Lentz

Theodore S. Lentz is an Assistant Professor in the Department of Criminal Justice and Criminology at the University of Wisconsin-Milwaukee. His research examines spatial and temporal patterns of crime and the processes underlying those patterns. He is involved in several ongoing projects on topics of offender decision-making, crime networks, gun violence, and policing.

Joseph A. Schafer

Joseph A. Schafer is Professor of Criminology & Criminal Justice at Saint Louis University. His research considers issues of police organizations, leadership and organizational, policing strategies, police stress and well-being, and officer discretion.