A day after the shooting, the road gets deadlier

On 4,142 days between 2014 and 2023, at least one person was killed in a mass shooting in the United States. The day after each of those days, the country's roads were measurably more lethal. Fatal car crashes rose by roughly 0.7 per cent on the day following a major incident, according to a working paper published in late July 2026 by a team led by researchers at the University of Pennsylvania and the University of Colorado. The mechanism the authors propose is as mundane as it is unsettling: a nation that has absorbed the shock of a public massacre drives the next day with slower reflexes, dimmer attention, and a slightly heavier foot.

That a mass shooting should leave a trace on traffic safety rather than on some adjacent policy domain is, on the face of it, strange. The link runs through attention, the authors argue. The first 24 hours after a major shooting are saturated with cable-news replays, social-media clips, and workplace conversation. Sleep is shorter. Cognitive bandwidth the morning after is poorer. The result, if the working paper is right, is a small but real bump in road deaths across a country of 330 million people, layered on top of a baseline that already runs near 40,000 fatalities a year.

The numbers

The study pulls its event data from the Gun Violence Archive, the most widely cited crowd-sourced tally of U.S. shootings, and pairs it with the National Highway Traffic Safety Administration's Fatality Analysis Reporting System, which records every on-road fatality on American soil. Between 1 January 2014 and 31 December 2023, the researchers identified 4,142 days on which four or more people were shot in a single incident, the archive's standard threshold for classifying an event as a mass shooting. The day after each of those days carried an estimated 0.7 per cent more fatal crashes than the baseline expectation for that calendar date, day of week, and region. The bump is small in relative terms but compounds into roughly 60 to 70 additional road deaths a year on the authors' headline estimate.

The effect is not evenly distributed. It concentrates in the 12-to-24-hour window after the shooting, and it is sharper for daytime crashes than for nighttime ones, consistent with a story about distracted driving rather than, say, alcohol consumption. The authors are careful to note that a 0.7 per cent shift is within the range of statistical noise for a single event; what makes the finding credible is its consistency across more than 4,000 incident-days.

A pattern hiding in plain sight

What makes the paper uncomfortable to read is how ordinary the proposed mechanism sounds. The authors do not claim that gun laws, mental-health policy, or the cultural politics of firearms explain the bump. They claim distraction does. A driver who has spent the previous evening doomscrolling a livestream from Uvallei or Buffalo is, on average, fractionally worse at the job of piloting two tonnes of steel at 70 mph the next morning.

This is not the framing American public-health debates tend to use. Discussions of mass shootings usually pivot to firearm access, to red-flag laws, to the political coalitions for and against assault-weapon bans. Traffic safety sits in a different silo, overseen by a different bureaucracy, written about by a different press corps. The working paper's contribution is to point out that the silos are leaky. A country that cannot insulate its drivers from the psychic shock of a mass killing is a country in which the consequences of gun violence travel further than the ballistic range of the weapons involved.

What the data cannot tell us

The paper is honest about what it cannot show. It does not identify the individuals involved in the post-shooting crashes; it cannot say whether any of them had read coverage of the shooting, whether they knew someone involved, or whether their state of mind matched the model the authors have in mind. The 0.7 per cent effect is an aggregate signal lifted out of millions of anonymous trips. There is no plausible causal experiment here. Mass shootings cannot be randomly assigned, and no institutional review board would permit one.

Two reasonable objections sit on the table. The first is that the Gun Violence Archive uses a broader definition of mass shooting than the FBI's four-or-more-killed threshold, which means the dataset includes drive-bys and domestic incidents alongside the cinematic "active shooter" events that dominate news cycles. The mechanism the authors propose should apply most cleanly to the latter, but the working paper does not separate them. The second is that fatal-crash counts have drifted upward since 2020 for reasons that have nothing to do with mass shootings: faster driving on emptier roads, a heavier mix of SUVs, and a long-running decline in traffic-enforcement staffing. The authors control for time trends, but a careful reader will note that their 0.7 per cent figure is the residual after those controls, not the raw difference.

What it would take to disprove

The cleanest test would be to look at non-U.S. comparators. Mass shootings of comparable scale occur in other wealthy democracies, albeit rarely. If the same bump appears in, say, Germany or Canada, the case for a generic attention-shock mechanism strengthens. If it does not, the U.S. signal may reflect something specific to American news consumption, social-media saturation, or driving culture. The authors flag this comparison as future work; the working paper itself stops at the U.S. border.

A second test would be to exploit variation in media exposure. Not every mass shooting is wall-to-wall news: the Gun Violence Archive records hundreds of four-or-more-shot events a year, of which perhaps a dozen a year become national stories. The paper's effect is estimated across all of them. If the bump concentrates in the nationally broadcast incidents and disappears for the locally buried ones, the case for an attention-driven mechanism tightens considerably. The authors say they plan to run that disaggregation in a later draft.

The bigger frame

Read against the grain, the working paper is less a finding about traffic safety than a finding about bandwidth. A country that spends a measurable share of its days processing the aftermath of mass gun violence has, in aggregate, less attention left over for everything else. The price is paid in lives on the road rather than in lives lost to bullets, which is why the connection has gone unobserved for so long. Public-health epidemiology is good at linking proximate causes to proximate outcomes. It is less good at linking distant causes, working through diffuse channels like sleep and attention, to outcomes that look unrelated.

The policy take-away, if there is one, is unromantic: after a mass shooting, expect the next day to be a slightly worse day to drive. Employers who can shift the morning commute later may want to. Insurers who price risk by the hour may want to. And the editors who decide what to put in front of a national audience on the evening of a shooting may want to think, briefly, about the drivers who will be reading their coverage on the way to work the next morning.

Desk note: Monexus reports the working paper's headline figure and its methodological caveats as published. The 0.7 per cent estimate and the 60-to-70-extra-deaths-a-year extrapolation are the authors' own; readers should treat them as a single research team's reading of an aggregate dataset, not as a settled epidemiological consensus.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://www.nhtsa.gov/crash-data-systems/fatality-analysis-reporting-system
• https://www.cdc.gov/transportation-safety/index.html