The most dangerous car in America is... | Hacker News Reader

The most dangerous car in America is...

blog.leaky.com

30 pointsjasontraff29 comments

Is it the car that's deadly, or does the car attract deadly drivers?

For example, are there other, stronger predictors of deadly accidents such as age, geography, or occupation? And are those people more likely to buy certain cars? I notice Buick has fewer deaths per million, but could that be because Buick attracts more cautious or less frequent drivers?

I assume the IIHS has competent statisticians. But I'd still be interested to see how strongly the differences between drivers factor into this.

I wondered about this too. The other part is -- how much is a car driven. Even if you look at behavior, is a 350z that much safer than a Porsche 911? Or is it just that 350z are driven [X] number of times more often, in the same manner, and thus result in a much higher death rate.

The porsche 911 line definitely handles better than the 350Z. Even in bad weather, which is where I would expect most of the accidents would have occurred.

Hmm, guess what I just bought.....

By itself this data isn't much use though, I suspect. For years Concorde was statistically incredibly safe, then one crash made it statistically incredibly dangerous, simply because there were only 14 flying in the first place - small datasets are rather prone to wild swings. Without knowing at least incidents per car sold, let alone incidents per passenger mile, it's not very useful.

There's then the cause of incidents. Without wishing to cast undue aspersions on my fellow owners, I suspect it may be the sort of car more likely to be driven in a manner likely to put it at risk of serious accidents. It is, after all, about the fastest car for the money and almost certainly the fastest 2WD car for the money. I hope I drive relatively sensibly in it, I certainly try not to be an a##e just because I've got a fast car and am seriously considering advanced driver training. Do all owners though? Probably not, and I'm probably not being as good as I think I am either. Point is, none of that directly implies the car itself is dangerous, merely that it tends to be driven in a way that increases the risk of incidents. When did you last complain about some tearaway hooligan in a Toyota Yaris or Kia Rio?

>By itself this data isn't much use though

Yes! Data means nothing until it is explained. It cannot speak for itself; its only use lies in helping us select from competing explanations.

> By itself this data isn't much use though, I suspect. For years Concorde was statistically incredibly safe, then one crash made it statistically incredibly dangerous, simply because there were only 14 flying in the first place - small datasets are rather prone to wild swings.

They don't have the metrics you mentioned, but all of those cars have at least 100k registered vehicles, as is stated in the third paragraph: "The study looked at vehicle models from 2005 to 2008 that had more than 100,000 registered vehicles and examined the driver death rate between 2006 and 2009."

Hmm, yes, I'm a twit - they did specifically say this is deaths per vehicle. Still not against per passenger mile or as deviance from the demographic norm for the driver which would be what we'd really need to find how dangerous the car inherently was, but still, this is better than I said.

I drove a 350z for a couple of years and never felt more invisible on the road than when I was driving it.

It was a weekly occurrence to have a big truck merge right into my lane without seeing me, causing me to have to swerve to avoid getting creamed. I'm not sure if it was because it was so low to the ground or what, but I always had to pay extra attention every time I got behind the wheel.

I've had mine for close to 3 years now and I can't recall any instance where I felt like this. I'm curious where you live, or at least where you did most of your driving where you felt invisible in it? I had it for the first year in a small town in Ohio (thought it was garaged for the winter as it's impossible to keep the tires planted in the snow) and I've been down in Florida's Space Coast for the last 2 years with no real issues revolving around not being seen.

I'm in Phoenix, AZ. We have terrible drivers in general, and I spend a lot of time on I-10.

The problem was usually someone with a cellphone planted on their left ear during rush-hour, occluding the view of their rear-view mirror.

Couldn't agree more, and mine was "Le Mans Sunset".

How could people consistently miss a bright orange car in their mirrors?

I think the data is misleading. It implies there's something safer about the physical car, its structure etc while the trends may be due to the age ranges, emotional maturity, sense of responsibility etc of the average driver for the particular model. Just look at the "safest" car (Chrysler Sebring). IMO that's a car for relatively conservative drivers. While the most dangerous car (Nissan 350z) is IMO driven more often by younger, less responsible people.

These are some surprising results.

1) A lot of Nissans. A real lot.

2) A lot of trucks.

3) A lot of the same makes (Nissan, Chevy), while some notable absences (only one Toyota, no Hondas)

4) Kia Spectra is listed twice, with no differentiation. Looks like a mistake, as it is the only one like that. (I'm assuming different years) Edit: It's actually station wagon vs. sedan, with the wagon higher.

They are smaller trucks, though.

The key difference between the 'safe' vs 'unsafe' cars was electronic stability control. We talk a little about it in another post. The other difference was size (namely weight and height of a car). Those were the key themes that emerged.

It would have been nice to see a plot of all data, not just the selected region of the 26 most dangerous.

That's a great idea, and I'm curious as well. I may have to get back to you on that. You can find the report at: http://www.iihs.org/externaldata/srdata/docs/sr4605.pdf

It would be interesting to see what were the most common vehicles to be involved in crashes with these 'dangerous' cars.

Scanning the raw data, I see the Toyota Sienna has zero reported deaths in any category (Minivans seem to do well in all categories).

Like others mention, this likely has to do with some degree of selection bias (i.e., Drivers who are safer choose certain vehicle types like minivans) as well as (in the case of the Sienna) a very small sample size, likely due to the model redesign that fell within the period, limiting the data - survey only chose most recent model data during the period).

What's also frustrating is they don't mention all deaths/injuries. For some insight into that, it's probably best to look at the IIHS other reports on insurance losses per model: http://www.iihs.org/research/hldi/composite_intro.html

The 350z is bought by the racer types. Another thing you'll notice is the really high occurance of salvage title sports cars on craigslist. I want a WRX wagon for my next car, and a solid 50% of the used WRX sedans I see on craigslist are salvage titled. I bet the numbers are the same or higher for the 350z!

I'd be curious to see whether it takes crash speeds into account. Isn't it wrong to name a car dangerous purely based on the number of people died in it? I'd argue the driving habits of the particular demographic that drives a particular car should be a huge factor.

Perhaps like looking at the number of driver deaths per million registered vehicles? ... oh. Wait. That's on the graph. And is the axis on which the "most dangerous car" is furthest out on.

You're letting your smartassitude get in the way of things. By demographics, gp is talking about things like age, gender, region, etc. If Nissan 350Z owners skew heavily towards irresponsible, twentysomething males pulling 140 on the nearest freeway, then it's not really fair to equate crash statistics (even if normalized) with a characteristic inherent to the vehicle itself. I could build a car of of paper mache and duct tape, but if I only sell it to grandmothers who drive to the store once a month it would barely even show up on this graph. All of which is to stay: the sole purpose of an infographic is to mask nuances in complex data. Draw conclusions from them at your own peril.

It's pretty hard to measure the speed of a crash after the fact. By the time any third party notices the scene, the speed is generally 0.

Skid marks and debris can actually provide a pretty accurate picture of the speed of the collision.

Remember, these are only the most dangerous cars that passed all the DOT safety requirements and (presumably) were maintained and passed inspection. It's easy enough to make a more dangerous car, but it's hard to be legal about it.

I am so glad I can't afford an expensive sportscar.

Get a Honda Accord V6. Basically the same weight. About 1 second slower in its 0-60 acceleration. Cheaper, better mileage and safer.

Just to add, the Honda Accord scored really well in the test with 19 driver deaths, which is considerably lower than the average of 48 and the 350z high of 143.

Reply on news.ycombinator.com