Indian Counties Have Fewer Crashes Per 100 Population; I-90, SD 34, US 212 Add Danger?

Frequent reader Nick Nemec looked at my review of 2014 vehicular crash data in South Dakota and asked if I could rank the counties based on the ratio of crashes to county population.

Not only can I rank ’em, Nick, I can map ’em:

Statewide, the average number of crashes per 100 people is 2.033. In the map above, green shows counties below that average—i.e., with fewer crashes than their population would predict if the statewide average held everywhere. Red shows counties above that average. The darker the color, the farther from the average. Click on each county to show its crash numbers and population for 2014.

If the ratio of crashes to population indicates relative safety on the highways, then some of the safest counties in South Dakota are in Indian Country:

County Crashes (2014) Population (2014) crashes per 100 population
Todd 15 9,882 0.15
Shannon 28 14,218 0.20
Bennett 9 3,430 0.26
Ziebach 10 2,826 0.35
Dewey 21 5,662 0.37
Gregory 22 4,217 0.52
Douglas 21 2,973 0.71
Mellette 21 2,100 1.00
Charles Mix 93 9,287 1.00
Corson 44 4,182 1.05

Meanwhile, the ten seemingly most dangerous counties are the rural counties along the I-90, SD 34, and US 212 corridors:

County Crashes (2014) Population (2014) crashes per 100 population
Jones 73 975 7.49
Sanborn 99 2,336 4.24
Moody 247 6,367 3.88
Aurora 106 2,745 3.86
Lyman 143 3,877 3.69
Faulk 86 2,357 3.65
Jerauld 69 2,007 3.44
Miner 78 2,316 3.37
McCook 190 5,649 3.36
Spink 202 6,598 3.06

Those numbers don’t mean that folks in Faulk and Jones counties are uniquely bad drivers; it’s more likely that those counties simply see a lot more through traffic compared to their relatively small populations, and that traffic brings those counties more than their otherwise fair share of accidents. But those numbers also suggest that first responders in those areas may have to work harder and see more roadside nastiness than there counterparts in other, less-traveled or more policed counties.

Five of our counties with populations over 10,000, including Minnehaha and Pennington, are above the average crash-to-population ratio; the other thirteen counties with five-figure populations, including burgeoning Lincoln, are below the statewide crash-pop ratio.


13 Responses to Indian Counties Have Fewer Crashes Per 100 Population; I-90, SD 34, US 212 Add Danger?

  1. Nick Nemec

    Thanks for the work with this data Cory. The high number of miles driven in I-90 counties by people other than residents, thus driving up per capita numbers, are the flip side of the low numbers in Indian Country. Most of the reservations are in very remote areas with few outside visitors, and by all indications the SD native population has fewer vehicles than the white population. Fewer vehicles means fewer miles driven, fewer miles driven means fewer accidents.

  2. I wonder about the data reporting from Indian Country, it wouldn’t surprise me if the reservation did not report accidents that occurred on their own lands and were handled by their own agencies. Data reporting varies greatly on reservations, especially when it comes to reporting to state agencies. Correct me if I am wrong though.
    The other thing I wonder about is fatality per capita rates. Car crashes only tell part of the story. Are those people that are getting in crashes dying more or less in certain counties? This could help tell us if these reported crashes are fender benders or high speed head-on collisions with real adverse outcomes.
    One of the best public health interventions we have done over the past 30+ years is the improvement of vehicle safety. Honestly when I worked as a paramedic, I worked on few car accidents, and fewer that had poor outcomes/fatalities. Our roads are getting safer, even if we aren’t necessarily crashing less.

  3. What constitutes a “crash?”

  4. CK, according to the 2013 South Dakota Motor Vehicle Crash Summary, “Current state law requires an accident report to be filed for each motor vehicle traffic accident resulting in the death or injury of a person, or property damage to an apparent extent of one thousand dollars or more to any one person’s property or two thousand dollars accumulated damage per accident” [p. 1].

  5. MD, I share your curiosity about reporting rates. We don’t know what we don’t know!

  6. Nick Nemec

    Does data include information on which law enforcement agency filed each accident report? I have no idea what level of cooperation exists between tribal law enforcement and the state on the simple filing of accident reports.

    I suspect county by county auto accident death data would be an unreliable metric to use to determine highway safety of any given county. Since SD is a state with many low population counties one fatal accident in a place like Hyde County would really skew the data for that county. Years of zero fatal accidents would lead you to think they are safer than they are and suddenly when they have one fatality they would look like a place where you take your life in your hands simply by driving to the store. Neither assumption would be correct.

  7. One can bet the statisticians and actuaries with insurance companies breath this data and adjust your rates accordingly. So how bad or well your neighbors drive affects your bottom-line.

  8. Douglas Wiken

    “Crash” is preferred to “accident”. “Accident” has a random no-fault aura to it or implications not carried by “crash”.

  9. Oh ho! An eager reader points me toward state and federal DOT Study SD2005-14, “Improving Motor Vehicle Crash Reporting on Nine South Dakota Indian Reservations.” Issued May 2007, this study said that “Crash report rates from tribal lands in South Dakota are generally low relative to the number of crashes estimated for those areas. This study sought to quantify the number of actual crashes on tribal lands in South Dakota for 2005, describe the barriers to better crash reporting from tribal lands, and suggest remedies. The research team visited all nine reservations and worked with law enforcement agencies to retrieve crash data for 2005 as available. A total of 737 crashes were documented in some fashion by tribal and BIA law enforcement agencies, though only 52 were reported with enough detail to be included in the South Dakota Accident Reporting System.”

    The study recommended more training for tribal police, software solutions, and somehow overcoming political friction between tribal and state officials to improve reporting. I will need to do more reading to determine whether those solutions were implemented; the data from the crash report suggest they have not.

  10. To further emphasize the problem, that 2007 study found that from 2001 to 2005, American Indians constituted 26.4% of South Dakota’s highway fatalities. I’m looking for comparable data on recent crashes.

  11. …nertz! The 2013 report breaks down crash drivers by age (simple solution: no one drives without Mom until age 30) but not by ethnicity.

  12. …but that report also notes that in 2008, “SD Highway Patrol begins submission of all reportable crashes using TraCS (Traffic and Criminal Software) system. The Office of Accident Records will expand TraCS to add municipalities & counties for more efficient reporting during 2008.”

    The 2012 Tribal Transportation Safety Summit reported that the Oglala Sioux Tribe had adopted TRaCS and “several Summit participants expressed an interest in hearing more about OSTs experiences with TraCS, and would like to discuss in the future how they may use the system.”

  13. Interesting note on insurance, John. I take it folks in Spink and Faulk counties could face higher insurance rates based on the higher ratios of crashes to population? Care to speculate how many dollars those differences might raise an annual auto premium?