[crossposted from my blog]
Background
The Insurance Institute for Highway Safety just released its 2011 status report regarding driver death rates. (I blogged about the prior report in January.) With this new report has come a rash of news articles, such as this one in Autoweek, with the rather bold claims that SUVs have turned the tide, have stopped rolling over, and are now safer than cars.
Well, is this true? The short answer is that "it depends on the vehicle" and as such I encourage you to look at the report's vehicle-by-vehicle driver death statistics itself in order to find your vehicle and see where it may stand. For a more general discussion read on.
So what are the safest vehicles?
The real burning question in your mind is which vehicles, as driven by real life humans, are the safest. (More on that later.) Well, here's that list:
True to the headlines, there are indeed a lot of SUVs and large vehicles in general amongst the safest. This isn't to say that SUVs are necessarily safer as a rule:
While overweight and obesity continue to drive our health care costs into the stratosphere, excess weight seems to be good when it's in the form of a vehicle. Heavier vehicles do seem to lead to lower rates of driver deaths, and at any given vehicle weight SUVs as a segment have lower driver deaths than cars. Furthermore, these data appear to show that SUVs are no more likely than cars to cause death by rolling over, independent of vehicle weight. This reduction in rollovers is likely due to better, more car-like unibody designs and widespread adoption of electronic stability control.
While this seems to be common sense, it actually isn't. As I blogged about before, this wasn't always true: while typically safer in multi-vehicle collisions due to their mass, SUVs used to roll over and kill their drivers at rates far higher than now seen, with resultant high single-vehicle accident death rates. This is something new in the past 6 years of vehicle design and construction, in other words.
Not everything makes sense to me regarding the "bigger is better" argument. If height and weight are the only factors at play, why do pickup trucks have higher driver death rates at any given weight than SUVs, let alone higher than cars? Could it be that bro-tatos in pickup trucks drive worse than latte-sipping soccer moms in SUVs? Also, does the subset of cars that only includes luxury cars, midsize cars, and large cars do better than the aggregate group of "cars" in general (including tiny cars and sports cars)? I don't know the answer to these questions.
Limitations/Glitches in the Matrix
The data are adjusted for "driver age and gender, calendar year, vehicle age, and vehicle density at the garaging location," but even so are no doubt very skewed. Speeding, substance abuse, and careless driving habits vary significantly across the population, and specific vehicles will be driven and crashed by skewed samples of the population. Even after attempting to correct for demographic factors as the researchers have done, the driver death data are an imperfect proxy for vehicle safety in isolation, which is what we really care about.
Another limitation is that there is great variation in driver death rates within classes of vehicles (e.g., small 4-door cars). This variation within a vehicle class is often greater than the global variation between cars, SUVs, and trucks. For instance, a Honda Accord and a Chevrolet Malibu are both midsize 4-door cars, yet Accord drivers die at a rate of 19 [confidence interval: 13-25] while Malibu drivers die at a rate of 99 [80-118]. Therefore one should not make across the board claims about the safety of "cars" or "SUVs" without a specific example in mind.
There also must be driver effects that are not being controlled for adequately. The best evidence for this is the varying death rates between badge-engineered platform-mates that only differ superficially. Examples: the Ford Crown Victoria's driver death rate of 33 [27-38] compared to the Mercury Grand Marquis's 57 [50-65]; the Subaru Legacy sedan's 83 [66-100] compared to the Subaru Outback's 40 [34-47]; and the Chrysler Town & Country's 28 [14-42] compared to the Dodge Grand Caravan's 63 [42-84]. Also odd is the sometimes very large variance between 2WD and 4WD models that are otherwise identical: the Nissan XTerra 4WD's 26 [15-38] vs. the Nissan XTerra 2WD's 72 [55-89] is an example in point, with the Nissan Pathfinder demonstrating a similar but less extreme trend.
From the above, we can see that the data aren't perfect. So be it.
Conclusions, or What cars should I avoid?
After staring at these driver death data for several hours I've come to some conclusions:
I was wrong.
It turns out that, all other things being equal, physics trumps all... and now, thanks to lower rates of SUV rollovers, "all other things" are indeed equal.
Background
The Insurance Institute for Highway Safety just released its 2011 status report regarding driver death rates. (I blogged about the prior report in January.) With this new report has come a rash of news articles, such as this one in Autoweek, with the rather bold claims that SUVs have turned the tide, have stopped rolling over, and are now safer than cars.
Well, is this true? The short answer is that "it depends on the vehicle" and as such I encourage you to look at the report's vehicle-by-vehicle driver death statistics itself in order to find your vehicle and see where it may stand. For a more general discussion read on.
So what are the safest vehicles?
The real burning question in your mind is which vehicles, as driven by real life humans, are the safest. (More on that later.) Well, here's that list:
True to the headlines, there are indeed a lot of SUVs and large vehicles in general amongst the safest. This isn't to say that SUVs are necessarily safer as a rule:
All that said, it is indeed striking that there no small cars at all amongst the safest vehicles, only "small" trucks that aren't really small at all... This begs the question: Is bigger better?1. Statistically all of the vehicles with driver death rates this low are basically in a dead heat: "0" here probably just reflects sampling error. (As a concrete example, the confidence interval for the Mercedes E-Class AWD death rate is [0-43].)
2. Also note that in this list of 26 vehicles there are 7 cars, 3 minivans, and even 1 pickup truck. It's not all SUVs.
3. The cutoff point of 22 is arbitrary, and, as I mention above, not really relevant statistically. Consider vehicles with death rates under 40 instead, still a very good figure, and you get a different mix of vehicles, with 31 SUVs and 30 cars or minivans among the 72 vehicles in total. You can check my work here if you're interested in the raw data.
While overweight and obesity continue to drive our health care costs into the stratosphere, excess weight seems to be good when it's in the form of a vehicle. Heavier vehicles do seem to lead to lower rates of driver deaths, and at any given vehicle weight SUVs as a segment have lower driver deaths than cars. Furthermore, these data appear to show that SUVs are no more likely than cars to cause death by rolling over, independent of vehicle weight. This reduction in rollovers is likely due to better, more car-like unibody designs and widespread adoption of electronic stability control.
While this seems to be common sense, it actually isn't. As I blogged about before, this wasn't always true: while typically safer in multi-vehicle collisions due to their mass, SUVs used to roll over and kill their drivers at rates far higher than now seen, with resultant high single-vehicle accident death rates. This is something new in the past 6 years of vehicle design and construction, in other words.
Not everything makes sense to me regarding the "bigger is better" argument. If height and weight are the only factors at play, why do pickup trucks have higher driver death rates at any given weight than SUVs, let alone higher than cars? Could it be that bro-tatos in pickup trucks drive worse than latte-sipping soccer moms in SUVs? Also, does the subset of cars that only includes luxury cars, midsize cars, and large cars do better than the aggregate group of "cars" in general (including tiny cars and sports cars)? I don't know the answer to these questions.
Limitations/Glitches in the Matrix
The data are adjusted for "driver age and gender, calendar year, vehicle age, and vehicle density at the garaging location," but even so are no doubt very skewed. Speeding, substance abuse, and careless driving habits vary significantly across the population, and specific vehicles will be driven and crashed by skewed samples of the population. Even after attempting to correct for demographic factors as the researchers have done, the driver death data are an imperfect proxy for vehicle safety in isolation, which is what we really care about.
Another limitation is that there is great variation in driver death rates within classes of vehicles (e.g., small 4-door cars). This variation within a vehicle class is often greater than the global variation between cars, SUVs, and trucks. For instance, a Honda Accord and a Chevrolet Malibu are both midsize 4-door cars, yet Accord drivers die at a rate of 19 [confidence interval: 13-25] while Malibu drivers die at a rate of 99 [80-118]. Therefore one should not make across the board claims about the safety of "cars" or "SUVs" without a specific example in mind.
There also must be driver effects that are not being controlled for adequately. The best evidence for this is the varying death rates between badge-engineered platform-mates that only differ superficially. Examples: the Ford Crown Victoria's driver death rate of 33 [27-38] compared to the Mercury Grand Marquis's 57 [50-65]; the Subaru Legacy sedan's 83 [66-100] compared to the Subaru Outback's 40 [34-47]; and the Chrysler Town & Country's 28 [14-42] compared to the Dodge Grand Caravan's 63 [42-84]. Also odd is the sometimes very large variance between 2WD and 4WD models that are otherwise identical: the Nissan XTerra 4WD's 26 [15-38] vs. the Nissan XTerra 2WD's 72 [55-89] is an example in point, with the Nissan Pathfinder demonstrating a similar but less extreme trend.
From the above, we can see that the data aren't perfect. So be it.
Conclusions, or What cars should I avoid?
After staring at these driver death data for several hours I've come to some conclusions:
I used to resent the large vehicles that soccer moms throughout the US tend to favor. I thought that they only lent a false sense of security, what with their bulk, high seating position, and, often these days, high "cocooning" door sills. I thought that superior active safety, the increased nimbleness and maneuverability that a smaller, lighter car offers, would trump passive safety when the time of reckoning arrived.1. Size does matter, and, as such, a SUV, luxury car, or a large car might not be a bad choice. As a rule, lightweight vehicles do not do well when crashed. Even the best models in these minicar, small car, and sports car segments are only middling, and the argument that active safety (maneuverability) trumps passive safety is simply not borne out in the numbers. At the other end of the spectrum, luxury cars as well as midsize and large SUVs generally are very safe in general, as demonstrated by the list of overall safest vehicles atop this post, with only a few poorly performing models in those categories that buck the trend.
2. Design and country of origin matters. Small Korean cars are particularly unsafe although their larger models are acceptable. Based on this and the prior IIHS study's data I would not recommend that anyone ever buy a Kia Rio, for instance. The Chevrolet Colorado, Ford Ranger, Nissan Titan, and Nissan 350Z are also notable for their tendency to roll over and kill their drivers in inordinate numbers. Although these are the most egregious examples (and likely very influenced by driver demographics in the case of the Z car), the moral of the story is that it still pays to look at an individual vehicle's performance rather than just by going by its segment.
3. Most SUVs indeed have been cured of their rollover habit, but some pickups still turn turtle often. The vast majority of SUVs now roll over at comparable if not better rates than do cars. Keep in mind that there exist a few cars and minivans that tend to roll over, with some of the worst offenders-- worse even than any SUV!--the Nissan 350Z, Mitsubishi Eclipse, Mazda 5, Chevrolet Corvette, Pontiac Solstice, and Chevrolet Aveo. Pickup trucks still represent a sore thumb, as some models such as the Nissan Titan, Dodge Ram Super Duty, Ford Ranger, and Chevrolet Colorado roll over with alarming frequency and resultant high mortality.
4. "Going green" will neither kill you nor save your skin. The Toyota Prius is actually the nominally safest small car. It's in a statistical dead heat with the non-hybrid Camry but is edged out slightly by the Camry Hybrid. Similarly, the Civic Hybrid is essentially exactly as safe as the standard Civic. None of these vehicles is particularly safe in the grand scheme, however, even though the hybrid part, per se, isn't doing any harm.
I was wrong.
It turns out that, all other things being equal, physics trumps all... and now, thanks to lower rates of SUV rollovers, "all other things" are indeed equal.
Last edited: