The day after: study links US mass shootings to a measurable rise in fatal car crashes

On 17 July 2026, a peer-reviewed study published in JAMA Network Open surfaced a finding that cuts across two policy silos in the United States: the day after a major mass shooting, fatal motor-vehicle crashes rise by an estimated 0.7 per cent nationwide, with the effect concentrated on the day immediately following the incident and dissipating within forty-eight hours. The authors describe the signal as small, but consistent, and statistically robust across more than a decade of incident data.

The research team, led by experts at the University of Vermont and Columbia University, used a case-crossover design that compares the same day of the week in weeks adjacent to each shooting. That method strips out the seasonal and weekly rhythms of American driving, isolating the contribution of the shooting itself. The result is a rare quantitative handle on a phenomenon that has long been suspected: that high-profile acts of mass violence leave a measurable imprint on population behaviour far beyond the victims and the crime scene.

What the data actually show

The study draws on the Gun Violence Archive's records of 469 mass shootings that killed four or more people between 2014 and 2023, then matches each to federal traffic-fatality data for the surrounding weeks. The headline estimate of 0.7 per cent translates, in absolute terms, to roughly one additional fatal crash nationwide on each post-shooting day, with that excess distributed across the country rather than clustered in any one city or region. The investigators report the strongest signal on the day immediately after the event; by the second day after, the effect is essentially zero.

Two features of the methodology matter for interpretation. First, the case-crossover design means each shooting is compared to itself, not to a national average, controlling for local driving patterns, weather, and seasonality. Second, the team tested, and largely ruled out, alternative explanations: alcohol consumption, daylight hours, and even concurrent news cycles around politically charged events did not, in their robustness checks, reproduce the same day-after spike.

A population-level stress signature

The authors are careful not to claim a single mechanism. They point instead to a body of work on acute stress responses, on the attentional cost of processing violent news, and on the documented rise in cortisol and blood pressure that follows collective trauma. The framing matters. This is not a study of gun owners, of would-be shooters, or of any specific demographic. It is a study of what happens to a country's driving population on the day that a high-profile killing saturates the news.

The implication is uncomfortable for both sides of the gun debate. For those who treat mass shootings as primarily a question of firearms policy, the result suggests that the harm radius extends well beyond the immediate victims, into the mundane category of traffic fatalities. For those inclined to dismiss the broader social cost of these events, the data argue against it: even a small per-incident excess, multiplied across hundreds of events a year, accrues into a non-trivial public-health toll.

The counter-argument, and why it does not hold

The obvious objection is that 0.7 per cent is a small number, and that any study of this kind will surface noise. The authors address this head-on: the same-day effect survives corrections for multiple testing, holds across sub-periods of the decade, and is strongest after shootings with the highest news salience, which is the dose-response pattern one would expect if stress were the operative channel. A pure chance finding should not respect the dose-response curve.

A second objection is media priming: that heavy coverage of crashes the day after a shooting might bias the data. The researchers tested this by looking at the count of non-fatal crashes, which is less subject to media attention, and found a similar though weaker pattern. If anything, they argue, the true effect is plausibly understated, because the case-crossover method is conservative.

What the study does not yet settle

The data are national. They do not tell us whether the effect is concentrated among people who were personally exposed to the shooting, or whether it spreads through the ambient news environment. The authors flag this as a productive direction for follow-up, alongside the question of whether the effect has grown as American media consumption has shifted toward mobile, push-notification-driven news cycles. There is also no breakdown by driver demographics, vehicle type, or road class, which leaves the behavioural mechanism under-specified.

What the study does establish, on the evidence presented, is correlation at a scale and with a methodological discipline that makes chance an unlikely explanation. Whether that correlation is causal, and through what channel, is the question the authors are explicitly leaving to the next round of work.

The stakes, in plain numbers

Mass shootings in the United States have become a near-daily occurrence by some definitions, and roughly annual events by stricter ones. If each post-shooting day genuinely adds one fatal crash nationwide, the cumulative cost over a decade is in the low hundreds of lives, attached to incidents whose direct death toll is much larger. The framing the authors reach for is a public-health one, and the policy implication is a familiar one: the harms of mass violence are not bounded by the walls of the building where it happens, and the systems built to track those harms need to be able to count past the obvious line.

This piece sits at the intersection of two American policy debates that rarely speak to each other: gun violence and traffic safety. The Medical Xpress summary, and the underlying JAMA Network Open paper, treat them as connected through the lens of acute population stress, a framing worth taking seriously because it survives the methodological scrutiny the authors invite.