Truck Drivers All Drink the Same. The Truck Decides Who Dies.

I cross-tabulated two FARS datasets that nobody combines: per-model death rates and per-model toxicology results, covering 490,736 driver records from fatal crashes between 2014 and 2023. Simple calculation: for each vehicle class, measure how much the death rate varies, measure how much the impairment rate varies, and divide one by the other to get what I am calling the "choice gap."

Among the 22 pickup trucks in FARS with 100+ fatalities, the impairment rate runs from 17.4% (GMC Sonoma) to 23.9% (Nissan Titan), a spread of just 1.6 times, tight enough to suggest that truck guys drink like truck guys regardless of what truck they drove to the bar.

Death rates run from 0.13 per 100 million vehicle miles traveled (Ram 2500 and Ram 1500) to 4.83 (Chevrolet S-10), a spread of 37.2 times, which means the truck you chose at the dealership determined your likelihood of dying in a fatal crash twenty-three times more than whether you decided to drive drunk.

It is not a pickup-only phenomenon, though pickups demonstrate it most cleanly because their behavioral baseline is so flat. SUVs show a choice gap of 20.3 times, with death rates spanning 65.3 times while impairment spans just 3.2 times. Vans clock in at 9 times, sports cars at 5.7. Even sedans, where driver behavior varies the most (Toyota Solara drivers test impaired at 4.1% while Buick Park Avenue drivers hit 31.7%), the vehicle still matters twice as much as behavior.

What the truck is actually measuring

Look at the pickup data and a pattern materializes instantly: every truck with a death rate above 2.0 was discontinued before 2012. S-10 production ended in 2004, Ford Ranger and Dodge Dakota stopped in 2011, GMC Sonoma in 2004. Every truck below 1.0 is either a current-production full-size (Ram 1500, Colorado) or a heavy-duty model (F-250, F-350, Ram 2500) with structural mass measured in tons rather than pounds.

What separates these trucks is not behavior, and it is not really about "the truck" in some abstract sense either. It is about three specific engineering variables: the 2012 ESC mandate that old trucks predate, the side curtain airbags they never received, and the 2,000 to 4,000 pounds of structural mass they lack compared to a modern full-size. A 2003 S-10 weighs 3,100 pounds with a vintage ladder frame, no stability control, and side airbags that exist only in the brochure for the top trim nobody bought. A 2024 Ram 2500 weighs 7,200 pounds with hydroformed high-strength steel, standard ESC, and curtain airbags across three rows. Same driver profile, same blood-alcohol distribution. One dies at 37 times the rate.^[1]

The policy arithmetic nobody runs

The United States spent an estimated $614 million on impaired-driving enforcement and education in 2023, according to the Department of Transportation's budget documents. Mothers Against Drunk Driving alone spent $43 million.^[2] The entire apparatus targets the behavioral variable that, among pickup trucks, explains a 1.6-times variation in outcome.

Meanwhile, Cash for Clunkers in 2009 pulled roughly 700,000 old vehicles off the road in a single summer at a cost of $2.85 billion. No comparable program has existed since, despite the FARS data demonstrating that the vehicle variable explains a 23-to-37-times variation in outcome. We spend half a billion a year nudging a 1.6-times lever while ignoring a 37-times lever entirely.

The strongest case against this finding

Impairment rate is one behavioral proxy, not all behavior. A 2003 S-10 owner and a 2024 Ram 2500 owner might test positive at the same rate but differ in seatbelt use, speeding habits, road type, and rural versus urban exposure. FARS does not break those variables out at the per-model level, so I cannot control for them. It is plausible that old-truck drivers speed more, buckle up less, and drive on two-lane rural highways where head-on collisions are geometrically more likely. Impairment-rate flatness proves behavioral uniformity on one axis, not all axes.

Additionally, the "choice gap" calculation divides rate spread by impairment spread, which is not a formal regression coefficient. It is a first-order signal, not a controlled causal estimate. A proper analysis would require individual-level crash data with seatbelt, speed, and road-type covariates, which FARS stores but does not publish in a vehicle-model-level aggregation.^[3]

What you should actually do

If you own a pre-2005 compact pickup (S-10, Ranger, Dakota, Sonoma), your per-mile fatality risk is three to five times the class average and up to 37 times higher than a modern full-size truck. Upgrading to any post-2015 truck with standard ESC, side curtain airbags, and a modern frame reduces that risk by 70% or more, based on the rate spread in FARS. Yes, a new truck costs more, but an S-10 was never actually cheap because it just moved the cost to a different line item.

Check your VIN at nhtsa.gov/recalls regardless, because thirty-eight percent of recalled vehicles in the U.S. are never repaired.^[4]

Limitations

FARS toxicology data covers only fatally-crashed drivers, not the general driving population. Death rates use estimated VMT and fleet sizes with approximately ±15% uncertainty for low-volume models. "Deaths" includes occupants of other vehicles and pedestrians, not exclusively the truck driver. As a ratio of spreads rather than a multivariate regression, the choice gap carries real constraints on precision, but those constraints do not undermine the directional finding: within pickup trucks, which truck you drive swamps whether you were impaired, by an order of magnitude.