The 2024 update in this post (which claims the worst brands are Tesla, Kia, Buick, Dodge, and then Hyundai, in that order) uses data from an iSeeCars blog post[1], which claims to have gotten their data from the U.S. Fatality Analysis Reporting System[2]. There's one important detail they mention in the blog post:

> To adjust for exposure, the number of cars involved in a fatal crash were normalized by the total number of vehicle miles driven, which was estimated from iSeeCars’ data of over 8 million vehicles on the road in 2022 from model years 2018-2022.

Note that the mileage data for each model comes from iSeeCars, not FARS. We don't know how accurate it is. There could be selection effects because iSeeCars only gets odometer data for cars that are for sale or have been sold, not vehicles that people keep. If for example, someone reserved a Model Y and sold it above the original price as soon as they got it (a common practice for new Tesla models), and the second owner kept the car indefinitely, iSeeCars would only see the odometer reading at the early sale. Tesla mileage would also skew low in sales data for anyone who bought an electric car, then regretted it for one reason or another, and exchanged it for a gas car. And since the rate they are measuring is fatalities per vehicle mile traveled, not per passenger mile traveled, any model that tends to contain more people will skew higher in the stats.

If you query FARS data to get all Teslas involved in occupant fatalities from 2018-2022, there are 141.[3] If you split out the data by model, there were a total of 22 Model Ys involved in occupant fatalities from 2018-2022 (actually 2020-2022, as the Model Y was released in 2020). If there were 10.6 Model Ys involved per billion miles, then (according to iSeeCars) the total number of miles traveled by Model Ys was 2.07 billion. By the end of 2022, the US had approximately 484,100 Model Ys on the road.[4] If you divide the total vehicle miles traveled by the number of vehicles, then the average Model Y had less than 4,300 miles on its odometer. That seems suspiciously low, as Tesla's 2021 report put the average Model Y at 13,000-14,000 miles per year.[5]

This artifact in mis-measuring mileage would also explain why other recently released models seem to have high fatality rates. The Honda CRV Hybrid was released in 2020. The Buick Encore GX was released in 2020. US production of the Hyundai Venue didn't ramp up until 2020. The Toyota Corolla Hybrid was released in 2019. Most people who buy a new car either sell it quickly (because they dislike it and want something else), or they keep it until it gets old. So anyone collecting milage data from car sales will have data that is skewed low for new models because the higher mileage sales haven't happened yet.

It may be the case that I'm using different reports than iSeeCars did. It's hard to know whether they're looking at vehicles involved in fatal crashes, drivers involved in fatal crashes, drivers killed in fatal crashes, occupants involved in fatal crashes, or occupants killed in fatal crashes. FARS lets you query all of these, and the iSeeCars post doesn't make it clear which data they used.

Another odd thing about the iSeeCars data is that they claim an average of 2.8 vehicles involved in a fatal accident for every billion miles traveled. But in 2022, the US had 13.3 deaths per billion miles traveled by car. Unless the average accident involves 5 fatalities, it seems to me that the iSeeCars data has some significant issues.

1. https://www.iseecars.com/most-dangerous-cars-study#v=2024

2. https://www.nhtsa.gov/research-data/fatality-analysis-report...

3. https://cdan.dot.gov/query I don't know how long they retain generated reports for, but mine was available at https://cdan.dot.gov/files/files/b14dcfb4-1f45-4eeb-abda-744... when I made this comment.

4. https://en.wikipedia.org/wiki/Tesla_Model_Y#Sales states 172,700 sold in 2021 and 231,400 in 2022. Other estimates put around 80,000 sold in 2020, for a total of 484,100.

5. See page 79 of https://www.tesla.com/ns_videos/2021-tesla-impact-report.pdf

How do cars do in out-of-sample crash testing? (2020)