Why Deer-Crash Maps Lie

ITHE MAP THAT LOOKS RIGHT

Why the 'follow the collisions' shortcut keeps catching on

This approach costs little and makes immediate sense. Police record deer collisions, and many state agencies release county-level tables. Those tables offer a convenient stand-in for deer locations without surveying hunters. Each year, usually during September, at least one major insurer ranks states by the odds of hitting an animal. Large animals covered by that ranking include moose, elk and deer. But a collision record does not simply document a deer sighting. It documents a deer encountering a vehicle. Alter the vehicle count and the map shifts even if the deer count stays unchanged. That happens throughout a state. It explains why this stand-in breaks down on its toughest assignment: separating suburban counties from rural deer counties.

The weakness can be measured. We totaled registered deer harvest in Iowa alongside crashes involving animals for each county during 2015 to 2024. Correlating those totals produces r = 0.306, with less than ten percent of variance shared between them. A close relationship between collisions and deer would yield a much larger value. Since harvest itself only approximates deer numbers, this test offers direction rather than a precise answer. Dividing collision totals by county acreage weakens the correlation slightly. We neither adjusted for traffic nor demonstrated that traffic caused the result.

IITWO COUNTIES, TWO TRUTHS

Clayton supplies deer; Polk supplies traffic

Compare Iowa's leading deer county with the county containing its capital. In the northeastern driftless region, Clayton tops the state's harvest. Its woods and farms along river breaks produce deer and keep them there. Polk County contains Des Moines, with metro highways and commuting traffic. Hunters in Clayton registered 42,354 deer during the decade, compared with 10,415 in Polk, a ratio exceeding four to one. Animal collisions went the other way: Polk recorded 2,405, versus 1,401 in Clayton. Mapping wreck totals puts Des Moines ahead of Clayton even though Clayton has Iowa's highest registered harvest.

Both a deer and a vehicle must be present for a collision. Vehicles differ much more widely across counties than deer do. Mapping unadjusted wreck totals largely shows where driving takes place.

IIITHE INVERTED LEADERBOARD

Collision rankings push deer country out of view

This skew extends through the entire ranking, beyond the comparison of Clayton with Polk. Sort Iowa counties using unadjusted animal collision totals, and the leaders trace the state's traffic hubs: Cedar Rapids, the river corridor in the southeast, Des Moines and Dubuque, then the Quad Cities and Ames. Sort those counties using registered deer take instead, and Iowa looks completely different. The northeastern driftless region and woods along the southern border come forward. Give a hunter the collision map and it sends them toward interstates, pulling them away from land that actually yields deer.

IVTHE FIX IS A RATE

Account for travel and be clear about which animals count

The fix is to put back the travel denominator missing from a raw crash total. A crash rate expresses collisions per unit of travel, measured per vehicle-mile. That accounts for the amount of driving in a county, but it does not separate differences in deer abundance. Only rates can be ranked without further adjustment. The deer-crash series that qualify in our data are those states publish with travel already accounted for. Illinois reports a deer-specific rate covering three decades. We use that series as an indicator of deer-collision risk per vehicle-mile. We retain raw totals, identify them as counts, flag them, and do not rank the counts themselves.

Species labels require another distinction. Iowa offers no code specifically for deer collisions; it reports only 'animal-involved' crashes. That category extends somewhat beyond deer. Ohio's State Highway Patrol, by contrast, reports county collision counts that identify deer. The Iowa totals discussed here come from its animal category. They accurately represent that category and are never mixed without notice with deer-only counts from another state. A shared leaderboard made from series covering different animals and using different denominators creates a comparison their sources do not justify. It repeats the collision map's error at the next level.

The field rule is simple. Unadjusted collision totals tell you 'where do deer and cars collide most,' a question largely driven by cars. For the question 'where are the deer,' collision mapping is unsuitable unless converted to a rate adjusted for exposure. After that adjustment, still confirm whether the records cover deer specifically or all road animals. If a state with harvest registration supplies county data, measured harvest reveals where hunters take their deer. That measure does not account for hunting effort.