Featured Prius, Tesla and other, 23 most dangerous cars

There's already a lot of speculation that improper use of driver assistance systems is the only reason Teslas are topping that list.

The data does not distinguish between cars that were being completely controlled by human drivers vs. those using any kind of ADAS. We will probably have to wait a few more years before that kind of data is available and statistically useful.

 

It is quite possible that the click bait used the wrong number of vmt. Tesla said that tesla's traveled twice the vehicle miles as was used in this article, and the worst traveled over 3x more. Same may be said for Prius, for when bad data is used bad results follow.

 

The source: The 23 Most Dangerous Cars On The Road - iSeeCars.com

...
iSeeCars.com analyzed fatality data from the National Highway Traffic Safety Administration's Fatality Analysis Reporting System (FARS) of model year 2018-2022 cars with car crashes that resulted in at least one occupant fatality to identify the most dangerous vehicles on U.S. roads today.
...
“Most of these vehicles received excellent safety ratings, performing well in crash tests at the IIHS and NHTSA, so it’s not a vehicle design issue,” said Brauer. “The models on this list likely reflect a combination of driver behavior and driving conditions, leading to increased crashes and fatalities.”
...
Automakers With the Most Frequent Occupant Fatalities - iSeeCars Study
Rank Make Fatal Accident Rate (Cars per Billion Vehicle Miles)
1 Tesla 5.6
2 Kia 5.5
3 Buick 4.8
4 Dodge 4.4
5 Hyundai 3.9
Overall Average 2.8
...
iSeeCars analyzed fatality data from the U.S. Fatality Analysis Reporting System (FARS). Only cars from model years 2018-2022 in crashes that resulted in occupant fatalities between 2017 and 2022 (the latest year data was available) were included in the analysis. 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. Heavy-duty trucks and vans, models not in production as of the 2024 model year, and low-volume models were removed from further analysis.​

Both the Model Y and S were listed in the worst cars list of models suggesting they have a higher, occupant, fatality rate than the Model 3 and X. Curious, the Model Y production began in 2020 but the study covers 2017-2022. This suggests potentially "infantile" or new model attributes. But we know Tesla changes cars "on the line" instead of waiting for model year upgrades. Regardless, I am not in the market for a Model Y.

Bob Wilson

​

 

All though are some of the safest due to safety improvements, compared to the old heaps out there, of which I own one. Without any data to back it up. But I think they are safer.

 

Key takeaway;

"estimated mileage"

 

The real challenge is the sample set size. If one fully loaded Model Y accident occurred with less than 2 years FARS data, that one accident over so few miles would give an inaccurate statistical rate.

Their results beg independent replication … which any of can do.

Bob Wilson

 

"i see cars is a data driven and research company that helps find shoppers find the best car deals."

enough said

 

both Tesla and iseecars are using US government data for US operating vehicles. We do not know what time period isee cars used for vmt but Tesla included data to match the US government data for 2017 through 2022. Model Y was released in 2020. No one has been able to repeat the bad numbers from iseecars. I assume it is bad data or they would respond with their data to get the much worse rates.

with all these systems drivers need to be aware that the systems have flaws.

 

It is always best to go to the original source as sometimes, journalists lack engineering skills. So I downloaded the Fatality Analysis Reporting System (FARS) data for the year 2021: Fatality Analysis Reporting System (FARS) | NHTSA

• 16 Model S accidents with fatalities
• 5 Model X accidents with fatalities
• 34 Model 3 accidents with fatalities
• 14 Model Y accidents with fatalities

So how could they get such wrong numbers for fatalities per miles? ... Division!

As the denominator gets smaller and smaller, the subsequent results get larger and larger. Let me give two practical examples:

• 14 Model Y / 1000 miles = 1.4% accidents per mile
• 14 Model Y / 10000 miles = .14% accidents per mile
• 14 Model Y / 100000 miles = .014% accidents per mile

Any time someone reports a relative rate WITHOUT the raw numbers, be wary ... they may have "garbage" data which you can't see.

In this case, the FARS data is readily downloadable and easily loaded into a spreadsheet of your choosing. Given I have deep reservations about the "Bell the Hybrid" Act, I may go back and look at 24 years of data instead of what we had in 2009 when Congress had a "panic attack." But then, I kinda like the synthetic warning sounds.

Bob Wilson

 

Data: it inspired the likes of Art Spinella.
.

 

I didn't see the odometer reading in FARS from the crash vehicles. The article claims to get the miles from another source " the total number of vehicle miles driven, which was estimated from iSeeCars’ data." I suspect this "estimated" divisor it the problem. However, there is a workaround.

Basis of mileage estimate:

According to the Federal Highway Administration (FHWA), the average number of miles driven in the United States in 2022 was 14,489 miles per year. This is an increase from the previous year's average of 13,500 miles and 14% more than the 2020 average of 12,724 miles. ​

One could assign a model year vehicle in an accident as 1/2 the FHWA miles per year. Then any previous years would get the average number of miles plus 1/2 of the current accident model years. For example:

• 2022 model vehicle in 2022 accident - 14,489 / 2
• 2021 model vehicle in 2022 accident - 13,500 + (14,489 / 2)
• 2020 model vehicle in 2022 accident - 12,724 + 13,500 + (14,489 / 2)

Imprecise, at least it has a open basis of estimate until better odometer data is available from FARS.

We are coming up on a holiday. It might be interesting to map pedestrian accidents in 2022 as a function of vehicle model. Then divide by the above estimated miles, we can get an idea of who is really most lethal to pedestrians. Unlike the usual suspects, I would publish the raw data as well as rates. Did I mention I like the low speed tunes?

My hypothesis is the size of the vehicle "A" pillar is a primary determinant in pedestrian deaths. But if the data shows that to be a valid hypothesis, what to do with it? How to reduce it? Anyone here care?

Bob Wilson

​

 

Interesting. I can see the Teslas being driven fast and hyped as the safest car contributing to their dismal safety statistics.

 

Interesting. On the DIY EV forums, the most common donor vehicle is a Tesla. Are there that many more Tesla's on the road compared to any other EV in the US? Or are they more commonly crashed and parts out at a dismantling yard?

T1 Terry

 

Hard to reconcile the Ops dredged up article in light of Tesla's data, ostensibly accurate as their Mothership knows where each vehicle is and how far it's been driven:


https://www.tesla.com/VehicleSafetyReport

Well as an example - California is the largest market for vehicles of any kind & Tesla is the largest volume seller there ....
https://insideevs.com/news/727245/california-ev-car-sales-2024q2/#:~:text=In%20Q2%2C%20Tesla%20noted%20a,51%25%20share%20in%20Q2%202024.
So that may very well have something to do with it
.

 

It's 2018 to 2022, IIRC. Iseecars uses the miles and age from used car listings that pass through their site. This study is a step up in that it didn't just use a single year.

I'd say mosy drivers won't read the manual.

About a magnitude more Teslas; 100s vs. 10s of thaousands. Yearly Model 3 sales are around that of the Camry in the US. The Model Y outsold the Corolla(car) globally recently.

 

The claimed "dismal safety statistics" is not backed up by facts and data from the Fatality Analysis Reporting System (FARS). The FARS database of fatal accidents is open to anyone who wants to test their hypothesis.

We are coming up on a long, holiday weekend. I may analyze the 2019, 2020, 2021, and 2022, fatal accidents. I own a 2019 Model 3 and the Model Y started sales in 2020. The latest data is 2022 when I was still driving my Model 3. So what I'll do, is analyze all makes and models involved in fatal accidents. For mileage, I'll use the average, annual miles, probably a constant for full years since the model year, with 1/2 of that given year for the miles driven. These miles will be applied for all vehicles.

Once the miles per fatal accident are calculated, I list them by most fatal accidents per mile to least in hopefully a CSV file (along with the spreadsheet.) Then I'll post the worst 10, the Prius, and Tesla metrics. But I'll probably start a new thread with the results followed by a second post detailing how it was calculated going back to the FARS databases for each year. Since it will include the worst, Prius, and Teslas, I'll probably post the thread in "News" and tag our moderator on where he thinks it should go.

Sound like a plan?

Bob Wilson

 

I found it very interesting that the Y was so high on the list, but the 3 was nowhere to be found.
I suspect bad data as posited above.

I could see those cars fairly high on the list simply because of the power and performance of those cars.
I look forward to seeing Bob's calculations