Car safety ratings, other than crash tests?
July 14, 2014 11:36 AM   Subscribe

It's time to shop for a replacement car, and I'm interested in sizing up safety. I'm aware of IIHS for crash safety stats. I'd like to evaluate from other angles, especially safety stats from actual accident reports, but possibly other points of view I haven't though of.

This link is an example of the kind of thing I'm interested in, but it's a little niche paper from 2002.

Anything like this that's more recent? Any other safety studies that aren't just crash tests?

(If it helps, we'll be looking at used cars ~10 years old, probably sedans, possibly station wagons.)
posted by mattu to Travel & Transportation (13 answers total)
especially safety stats from actual accident reports

It is unlikely that you will get much information out of them that us usable or that is meaningful. There is a LOT of interpretation and extra factors in understanding accident reports. It is also a rabbit hole in terms of confirmation bias as different factors are not highlighted equally in that kind of thing (depending on what the study is focussing on).

You are far, far better off looking into generic crash safety stats (like IIHS) and spending the rest of the time on this learning how to make whatever car you buy safer than the average one of its kind. This will involve driver training (advanced or defensive driver training, for instance) for you and anyone else driving it as well as reading up on tyre choices for conditions (winter tyres/all seasons etc) and what correct maintenance is like for the car you will buy and how best to afford and plan for it.

You're focussing on the wrong thing. Surviving the accident is a tiny fraction of the problem and also often affected by totally random factors. Spend your time (once establishing a car with good crash structure generally) avoiding the accident. Do this with a car with good handling, effective visibility, brakes and road holding generally and proper training for the people operating the car.
posted by Brockles at 11:43 AM on July 14, 2014 [2 favorites]

Is there a reason you're interested in such old cars? A lot of important safety features have been introduced or become much more popular over the last ten years. And with a ten-year-old car I would think the condition of the individual car (frame, brakes, tires, shocks) is going to be much more relevant to its safety than the overall safety of that model of car.
posted by mskyle at 11:54 AM on July 14, 2014

That's a good point - I just assumed budget was the issue, but it is worth bearing in mind that degradation of the car is a big factor, as well as design life and where it sits in the age range you are looking for - later model styles are designed with much stricter accident and safety standards in mind, because they will need to be compliant through their life.

For instance if you were buying a a 2006 BMW (E46 style, 1998-2006 year model) it is much less efficient (fuel costs) and less compliant with later regulations because it was fundamentally designed somewhere between 1993 and 1998 (with some small updates). The E90 (2005-2011 model) is likely to be compliant with the much stricter crash regulations that have come into force in the last 10 years. It was likely designed between 2000 and 2005. So for the same year of car, you have a decade later design and safety standards available to you.

So please bear that in mind - if safety is a concern, then buy the latest design age car you can afford. Forgoe the bigger engine if it puts you in the newer car. In my recent example of the 3 series, I could have had a 2006 E46 335i BMW - a very fast and very nice car for not much money. But I chose to go for the (smaller engined) 2006 E90 330i because the design is so much more advanced (among other things). A little knowledge like that goes a long way.
posted by Brockles at 12:22 PM on July 14, 2014

Budget is indeed the big reason for buying an older car.

So a 2006 might have been designed in 95 and might have been designed in 2003. What is the easiest way to tell when a car was designed? Is there a threshold year or years I ought to be paying attention to?

I don't mind ambiguous or confusing data. Mostly I'm looking for different points of view.

Safety from things like driver training and tire selection is something I hadn't considered, thanks. I can study this for myself, but any further remarks (or statistics) on this are welcome also.
posted by mattu at 2:17 PM on July 14, 2014 Browse used cars by make and look at the model overview. It will give you a breakdown on what model years are covered by each iteration of that model. Here is the rundown on the BMW 3 Series that Brockles mentions.
posted by VTX at 2:23 PM on July 14, 2014

You can just look up model years and the like on Wikipedia - it's not in depth stuff you're after just phases of model ranges. If you look up BMW 3 series here, you'll see what I mean:

Or Honda Accord:
posted by Brockles at 2:25 PM on July 14, 2014

Generally speaking, the later a model was introduced, the more recent the design is. It generally takes a few years to design a new model. You can find the year that a specific model (or generation) of car was introduced through websites like and even Wikipedia. So for example, if you're looking at a 2006 Honda Accord, Wikipedia shows that that's part of the seventh generation Accord that was introduced in the U.S. in 2002, meaning that it was probably designed starting in 1998 or 1999.

Key factors in a car's safety are rust and frame integrity. Any car you consider should be looked over very carefully by a mechanic to make sure that the frame isn't rusted or otherwise damaged, because a rusty frame isn't going to protect you very well in an accident. Of course, brakes, steering components, tires, and shock absorbers (which are often overlooked) are crucial to safety. If the shocks are old, the car will sort of bounce along the road instead of adhering to it, which can increase stopping distances and make the car harder to control in an emergency.
posted by Leatherstocking at 2:29 PM on July 14, 2014

By far, the single biggest factor affecting survivability in an accident is simply the weight of your vehicle. This is well established by thousands of tests and lots of data. A compact car weighing less than 3000 pounds has a 10-15% higher risk of occupant fatality overall than a heavier car. But age, both in terms of structural degradation and in terms of the evolution of safety features, is also a key factor. If safety is your highest value, you want the youngest possible car (and car design) in any given size class. Side-curtain airbags, for example, added hugely to survivability in small cars involved in the most common kind of fatal accident. Side curtain airbags did not become common equipment until the early 2000s.

Another thing often overlooked on old cars: the seat belts. The belts need to be intact and the tensioners and buckles in very good shape. Worn seat belts failing during a collision is more common than you might think. And without good belts, all the airbags and structural integrity in the world is not going to save you from going into a steering wheel or through a windshield. If you buy used, inspect the belts very carefully, all along their length, for any fraying or stretching. If a car was in a serious front or rear collison, any belts that were in use at the time should be replaced even if they look fine.

If your brakes, steering, shocks/struts, lights, seat belts, and powertrain are all maintained and kept in top condition, then the biggest remaining factor is the quality of your tires and having the right tires for the road conditions you experience.

Modern cars have taken a more technological approach to preventing accidents in the first place, with lane departure notification, blind spot and rear cross traffic monitoring, road-following headlights, and even crash avoidance braking systems that stop you automatically if you are about to rear end someone. Once you start driving newer cars and experiencing all of this stuff, which mostly works pretty well, you do wonder how you lived without it. And some people didn't live, actually, without it.

But bottom line, I think you can basically trust the IIHS data. They test comprehensively for all kinds of different collision scenarios, and they do it in order to estimate risk for their industry's actuaries, so they have a very high stake in the accuracy of their testing, and they don't do it for consumers, so there is no bullshit involved. They have nothing to sell except an interest in you not incurring huge medical bills or causing huge damage to others. Your insurance premium will reflect the IIHS ratings for your car. Their testing is very rigorous.
posted by spitbull at 6:02 PM on July 14, 2014

By the way, NHTSA reports a 26 percent decline in traffic fatalities since just 2005, and last year was the lowest number of deaths on American roads since . . . 1949!

There's a reason for that. Cars are safer now *mostly* thanks to new technologies. If you are serious about safety, it's not a splurge to spend money on a car that is less than 10 years old. It's an investment against paying the highest price a human can pay.
posted by spitbull at 6:06 PM on July 14, 2014

By far, the single biggest factor affecting survivability in an accident is simply the weight of your vehicle.

This is incorrect as presented and is established thinking that has put so many people blindly into hideously unsafe SUV's and trucks in the US in the first place. Unfortunately this thinking is not always true and needs interpreting or further consideration. Given that you are hitting a vehicle with comparable crumple zones and similar construction (ie bumper/chassis height, chassis style etc) this is mostly true for a head on impact, but a 4,500lb car with no good crumple zones and built on a ladder chassis (something like a Ford Bronco, maybe) would not automatically be more survivable than a modern designed monocoque construction 3000lb car (so any proper car, not SUV or truck). In fact you'd be better off in a 2014 spec Ford Fusion than a 2001 Bronco in almost every scenario. It also depends on the type of accident - are you hitting another car? Or a tree? A crash barrier? Or a Big Rig? A straight impact or an oblique one? Was the accident due to wet roads or poor conditions? If you're hitting a big rig or something immobile then it doesn't matter how heavy your car is, as it is proportionally so tiny to what you are hitting it becomes irrelevant. Would the lighter car (with better dynamics by a significant margin) have been less likely to have avoided the accident in the first place?

So with a head on accident between Fusion and Bronco? The Bronco is likely to be least damaged. However this crucially doesn't necessarily mean that the occupants experience a smaller accident pulse (ie sudden g loading during the impact). THAT is what kills people and crumple zones are essential in progressively reducing the accident pulse before the humans inside have to absorb it. Larger or older vehicles with poor crumple zones can be less survivable than smaller ones with more efficient crumple zones as a result. Also consider an oblique impact with a Fusion and a Bronco, where both have a secondary impact or perhaps roll, then all bets are off in terms of weight. It's all about construction of the vehicle and the safety features. My money would be on the Fusion being more survivable.

This doesn't change the fact that due to the hideous primary safety of the Bronco (primary safety= vehicle handling, so the ability to avoid the accident in the first place) would mean that the Fusion is significantly (like A LOT) more likely to be controllable in a near accident to a greater degree or even avoid a given accident all together. You are more likely to have an accident in a large, heavy and dynamically poor car than a smaller, lighter and more nimble one. It's all about parameters - do you want a vehicle that is more likely to survive a head on impact with a big rig? Or one that is less likely to hit a tree on a wet mountain road and get you home without even scratching the paint? Which scenario is more likely to be a factor in the life of the car in your use case?

It is not at all as simple as it is presented there. It is not just a matter of mass unless there are comparable configurations and a straight head on accident. Everything else blurs the line or makes that established 'wisdom' incorrect.
posted by Brockles at 6:58 PM on July 14, 2014 [4 favorites]

Thanks, everyone. We'll lean toward newer cars as we shop.
posted by mattu at 5:53 AM on July 15, 2014

Sorry but the laws of physics remain true. Absolutely there are many other interlocking factors, but all else being equal, weight matters most. Anti-SUV ideology notwithstanding (I share it).

Other than the driver's skill of course. That is far and away the most important factor you can control. The meat in the seat as they say.

The other major statistical variable is how much and where you drive. The linked article thoroughly reviews the subject from a statistical point of view.
posted by spitbull at 8:00 AM on July 15, 2014

That article neatly demonstrates the bias on which it is based in the second paragraph: "Large vehicles have longer hoods and bigger crush zones, which gives them an advantage in frontal crashes.

So, assuming similar crush zone technology and also in frontal crashes only. The article repeatedly states that similar cars crashed together show the larger one winning, not the heaviest car wins regardless, which is the clarification I was making. Also, in the US larger vehicles do not necessarily have larger crush zones (ie trucks built to commercial regulations rather than passenger car's much tighter crash requirements).

Given the following- equal construction style, equal safety features, frontal, simple (ie single) impact (as I said above) - yes the larger car wins, but that is a very small proportion of collisions and the result of an even smaller proportion of 'accidents that could potentially result in collisions' (where primary safety is a massive factor). Also the video they link to showed different sized cars from the same manufacturer (so of the same construction style and comparable safety features) being shown and (as I said above) in that context the larger car wins.

The survey is also not at all factored for percentage of cars of each style on the road, nor number of occupants/likelihood of front passengers being killed weighted accordingly. Nor what the vehicles hit - did the smaller cars that showed fatalities hit trees? Bigger cars? Or other small cars? What were the speeds of the accident? It's the perfect example of how a conclusion can be plucked out of the air without any real consideration for the actual factors that determine survivability. It also doesn't give any detail on how injured the occupant may be in a NON-fatal accident. It isn't all about being killed - it doesn't factor how lower speed survivable accidents leave the occupants - is it progressively more damaging to them based on speed? Or is it a sharp change between 'Knocks and scrapes' and 'dead'.

It is just not possible to accurately make conclusions based on these sorts of surveys as they make assumptions that are not clear, and use statistics that are either not useful or are not factored/variables considered correctly to give a usable answer. Aspects are missing or ignored and it is hard to know *how* the statistics were gained. This is why I made the first response I did on the subject.

The statistics and facts represented as blanket require a series of large assumptions to be true. That is skimmed over in the article, which is why a lot of people (yourself included) erroneously conclude that the largest car ALWAYS wins. It is just not as simple as that. The article even demonstrates the bias/assumptions it has made within it, if you read between the lines.
posted by Brockles at 9:25 AM on July 15, 2014

« Older I'm sorry, I can't help you.   |   I need clean crisp percussive songs. Newer »
This thread is closed to new comments.