No, Tesla, Subaru, and Ram drivers don't have the most accidents

Recent claims that never should have been spread, now fully debunked

Four days ago, LendingTree published an article on their website with the headline, “Ram, Tesla and Subaru Have the Worst Drivers, While BMW Drivers Have the Highest DUI Rates”. The article is published with the following disclaimer, but also highlights that it was edited by two LendingTree staffers, including their “managing editor for studies and surveys”.

The article states as “key findings” claims such as:

• “Ram has the worst drivers”

• Mercury, a brand which hasn’t existed since 2010, “has the best drivers”

• “Tesla drivers have the highest accident rate”

• “BMW drivers have the highest DUI rate by a wide margin”

They also included this ranking, labeled “10 car brands with the most driving incidents per 1,000 drivers”.

In addition to accidents, the article also includes rankings for “incidents” (as in the image above), which include a variety of citations such as speeding tickets and DUIs. They claim that BMW drivers have the highest rate of DUIs.

Seems like quite the scoop!

These results got picked up by wide variety of news outlets, most of which focused solely on Tesla. The first I saw was Forbes, but since then it’s been all over social media, EV blogs like Electrek, mainstream news sites like CNBC, local Fox affiliates, and even apparently was mentioned on TV programs including Jim Cramer’s show.

If the social media chatter is any indication, people are very interested in these results. Some seemed surprised by the findings - why is Subaru so high on this list? For others, it confirms their preconceptions - of course all the Tesla claims about their low accident rates must have been false all along. And nobody seems surprised by the idea that BMW drivers, who seem to be universally despised, are always driving drunk. It explains so much!

The problem? None of it is true, as a LendingTree spokesperson has now confirmed.

As of this writing, the LendingTree article still remains published and not retracted or corrected, and the same is true for all of the others as far as I can tell. One would hope this will change soon, though the holiday break may be delaying things. Or maybe it’s already old news and nobody cares if it was true? I guess we’ll see.

So, what exactly happened? Read on to find out.

Reasons for initial skepticism

When I first saw these claims, many things surprised me. First of all, as someone who’s written detailed analyses of NHTSA’s accident data, the results seem to be not only a stark contrast to all the data I’ve seen, but in many ways the exact opposite. This alone doesn’t mean their data is wrong, of course, but it certainly made me want to look closer.

I also thought it odd that the “safest” drivers were supposedly those of three long- discontinued car brands (Saturn, Pontiac, and Mercury). The data isn’t presented as absolute numbers, but rather as a rate, so even if these vehicles are rarely still in use, a properly conducted study should account for this. This told me that either something was fishy with this study, or some very interesting pattern had emerged here that warranted further investigation. Since you’re reading this, you can probably guess which turned out to be the case.

The “methodology”

Something else that immediately caught my eye was that the article says the data comes from insurance quotes. Some of the articles referencing this have misstated the source as being insurance claims, which would make a lot of sense as a way of tracking accident data. Though it wouldn’t make sense as a way of tracking traffic citations.

The article has a section at the bottom about its methodology, which says:

Researchers analyzed tens of millions of QuoteWizard by LendingTree insurance quotes from Nov. 14, 2022, through Nov. 14, 2023.

To determine the best and worst drivers by car brand, researchers calculated the number of driving incidents per 1,000 drivers by brand in every state. This main category included accidents, DUIs, speeding and citations. We examined the 30 brands with the highest number of quotes.

We looked at the four categories combined and individually. Our individual analyses don’t add to the driving incident total because of drivers with multiple incidents.

There’s also a list of included traffic citation types, but that’s it.

Once again, they say this is based on insurance quotes, and specifically those requested through LendingTree’s “QuoteWizard” online insurance shopping tool. But they still say that they “calculated the number of driving incidents per 1,000 drivers by brand”, which everyone seems to interpret as meaning they knew what brand of car you were driving when the incident occurred. I mean, they’d have to have some way of doing that, right?

Immediately I came up with a possible theory explaining what had happened here, and I felt pretty confident that I was right.

What I thought had happened

My theory was pretty straightforward:

• LendingTree lets drivers request insurance quotes for a vehicle

• When requesting a quote, drivers specify their accident and citation history, because this affects the rate they’ll be quoted

• LendingTree tracks this data

• Someone at LendingTree decided to perform an analysis of the data, looking at the accident and citation histories of people requesting quotes for different vehicle brands

• Someone else misinterpreted the analysis and decided to write a blog post about it making claims not supported by that data

• Lots of people, including many journalists, uncritically accepted the claims and parroted them, and in some cases even further embellished them

If I was right, then at no point along the way did anyone consult a data analyst, traffic safety researcher, or anyone with any ability at all to vet the claims. Surely a trustworthy news source wouldn’t publish such potentially damaging claims about vehicle manufacturers without vetting the data, right?

In case it’s still not clear to you why making claims like this would be problematic, let me give an example:

• Let’s say you’ve been driving a Ford Fusion for 5 years

• 4 years ago while driving your car, you got rear-ended

• Sometime between November 2022 and November 2023, you decided to look at getting an electric car

• Before buying, you want to know what it will cost to insure, so you search for “car insurance quotes” and end up at LendingTree’s site

• You fill out their form and request a quote for a Tesla Model Y

You now have contributed to this data as a driver with an accident associated with Tesla. Or as the article falsely claimed, you are a “Tesla driver with an accident”. Even though you were driving a Ford when you had the accident.

After all, one of the main reasons people shop for car insurance is that they’re considering, buying, or recently bought a new car. In which case, their accident history has nothing to do with that car. Even if they were searching for insurance quotes for their current vehicle, that doesn’t mean any accidents or citations they’ve had in the past occurred while driving it. Indeed, for a new and rapidly growing brand like Tesla, it’s even more likely that a driver’s accident and citation records reflect driving in one or more prior cars they’d owned (or leased, or rented!).

I hope you can see the problem.

I won’t even get into other issues such as the non-representative sample, the fact that anyone can request a quote and put whatever information they want into the form (I personally did this a few weeks ago using this very tool while discussing the topic of high insurance rates, based on another LendingTree article about EVs being expensive to insure!), or the clearly poor data practices leading to long-dead car brands coming up as having the “safest drivers”.

Two days ago, when I published my latest review of Tesla safety data, which shows among other things that Tesla vehicles have among the lowest accident rates using high quality data sources (like NHTSA’s FARS database), I discussed this claim and my theory. At that time, though, I did not have confirmation that my explanation was correct.

Now I do.

The confirmation

Yesterday, I decided to e-mail Lora Kolodny at NBC, who ran one of the articles spreading these bogus claims. In her piece (which still remains unchanged!), she had a quote from a LendingTree spokesperson who helpfully answered a question about rental cars that I don’t think anyone asked (and for me, raised even more questions):

The researchers analyzed quotes from people looking to insure their own vehicles, and did not include accident or incident data involving drivers of rental cars, a spokesperson for LendingTree told CNBC by email on Tuesday.

I told Lora about my concerns and asked if she could help me get in touch with someone at LendingTree to clarify what they did and address my concerns. To my delight, she did exactly that.

Yesterday afternoon, I received the following e-mail from a Public Relations Manager at LendingTree:

Hi Brandon, 

Lora Kolodny from NBC forwarded me your email regarding our recent study on driving behaviors by car brands. I spoke with my editorial team about your questions on our methodology. 

You're correct that we can't know if the vehicle they're looking to get insurance for is the vehicle in which they had the driving incident. Ultimately, the people looking for auto insurance for a Tesla using QuoteWizard from Nov. 14, 2022, to Nov. 14, 2023, had the highest percentage of accident incidents. We made no claims about the safety of Teslas and/or the safety features of any other brand mentioned.  Our study is solely about QuoteWizard users during the previously mentioned time frame, the brand of car they're looking to insure, and the driving incidents of those users. 

I hope that helps clear up any questions you may have. Please let me know if there is anything else we can help answer. 

Happy holidays,
<name> | LendingTree, LLC
Public Relations, Manager

Emphasis added

So, there we have it. Every claim made in their article and those echoing it is totally bogus. Somehow, nobody along the way decided to make even the most cursory attempt to vet these claims. I suspect if Tesla had not been at the top of the list, someone would have looked closer, or it wouldn’t have gotten much traction in the first place.

I’m not a lawyer, so I don’t know if LendingTree or anyone else is exposed to legal risk here. But I wouldn’t be surprised if LendingTree’s lawyers are having their holidays interrupted by some unpleasant phone calls this weekend.

Last night, I replied and asked the LendingTree PR person if they planned to post a retraction. I’ve not yet received a response.