In just a few weeks, Chicago’s contract with ShotSpotter, the gunshot detection system, will expire. The debate about ShotSpotter has focused on whether it helps solve crime. We have new data about shooting fatality rates that might be considered as well.
Let’s start off by noting what’s not useful evidence in this debate: One-sided analyses that consider only the benefits or the costs of this technology. For example, some police chiefs wrote an open letter, published on the company’s website, touting improved response times to shootings. The Chicago inspector general’s report, while noting that more data and analyses are needed, made the (now widely repeated) point that many ShotSpotter alerts don’t lead to gun recoveries or arrests. But most cancer screenings don’t detect cancer. We screen anyway because the cancers that do get caught are so helpful compared with the costs of the screenings.
What does that sort of cost-versus-benefit comparison imply for ShotSpotter?
Most obvious is the financial cost of ShotSpotter’s contract with the city, which has averaged something like $10 million a year.
Another potential cost is slower police response times to 911 calls for non-gun-related reasons as ShotSpotter shifts police attention to following up on gunfire alerts, as has been documented in a study by researchers Michael Topper and Toshio Ferrazares. The result is about 9% fewer arrests for such calls. We say “potential cost” here because we recognize many Chicagoans are concerned that the city is overpoliced and so may view these changes as a positive rather than negative development.
A third potential cost is that “false positives” — ShotSpotter alerts triggered by something other than gunfire, such as fireworks, could lead the police to respond too aggressively in settings where there is no actual shooting. Unfortunately, there is no good data on either the false positive rate (ShotSpotter claims 0.5%, but who knows) or on how police respond to those false positives.
What do we see on the potential benefit side?
The consensus seems to be that ShotSpotter does not lead to reductions in gun violence or other gun crimes, such as robberies. Whether the impacts on crime would be larger if police departments were better run on average — there appears to be a great deal of variability in this across cities and even across districts within a city — is an open question.
Another potential benefit from ShotSpotter that has gotten less attention in the discussion is getting rapid first aid to victims. Several of us saw an example recently on a midday visit to District 11 (Garfield Park) when a ShotSpotter alert came in. On the police camera system, we saw officers arrive, discover the victim slumped in the driver’s seat of his car, apply first aid and call for an ambulance. Also relevant is that we were monitoring the police radio the whole time — at no point was there a 911 call, so who knows how long it would have taken for the victim to get medical care otherwise.
New data analysis by our research center, the University of Chicago Crime Lab, points suggestively to how big this benefit might be. We compared shooting fatality rates — the odds a victim dies of their injuries — at the boundaries of police districts that have ShotSpotter and those that don’t, the type of comparison considered to be one of the best ways of identifying the effects of a place-based policy. We can see these are otherwise similar types of places because before ShotSpotter is launched, there’s no detectable difference in shooting fatality rates.
After ShotSpotter goes live, fatality rates are about 4 percentage points lower in the areas with the technology. With an overall fatality rate of 17%, this is about a one-quarter drop in the odds the victim dies. This evidence is suggestive but not definitive because there’s about a 1-in-4 chance that the result reflects noise in the data rather than a true impact on fatality rates. (This data come from Chicago’s Violence Reduction Dashboard, which the Crime Lab helped build for the city to support public access to data — including to let others replicate these sorts of results.)
So where does this leave us?
Given the number of shootings each year in the police districts that currently have ShotSpotter, there is, roughly speaking, a 3-in-4 chance that the technology saves about 85 lives per year. That comes from multiplying a 4-percentage-point change in the fatality rate by the total number of shootings in the ShotSpotter areas, equal to 2,124 in 2023.
By any measure, 85 lives saved per year would be tremendously valuable. One way the federal government tries to quantify that value when considering new safety regulations is to use a figure of $13 million per life (in 2024 dollars). So if Chicago used the federal rule, there’s a 3-in-4 chance of a lifesaving benefit that equals $1.1 billion per year ($13 million times 85).
Against that benefit is the financial cost to the city of around $10 million, plus whatever weight we as a society put on slower response times and fewer arrests for nongun crimes. There is also the cost of possible overly aggressive police response to ShotSpotter false positives; figuring that out should be a high priority because there is currently no way to quantify that phenomenon.
Whether this trade-off with its various uncertainties is worth it is ultimately a question of values and politics — how do Chicago voters and the politicians who represent them view this mix of pros and cons? However the city ultimately answers that question, the time to answer it is now — the clock is ticking.
Henry Josephson is a research intern at the University of Chicago Crime Lab and a fourth-year college student at the University of Chicago. Javier Lopez is analytics manager at the Crime Lab. Jens Ludwig is the Pritzker director of the Crime Lab and the Edwin A. and Betty L. Bergman distinguished service professor at the University of Chicago. Alejandro Roemer is a research analyst at the Crime Lab.
Submit a letter, of no more than 400 words, to the editor here or email letters@chicagotribune.com.
Originally Published:
