http://eetd.lbl.gov/ea/teepa/pdf/TRB_Safety_1-03.pdf
Are SUVs Safer than Cars? An Analysis of Risk by
Vehicle Type and Model
Transportation Research Board 82nd Annual Meeting
Washington DC
January 15, 2003
Summer of Findings
Average midsize and large cars have same risk to drivers as average SUV
Safest subcompact and compact cars have same risk to driver as average SUV
Pickups and SUVs (and minivans) impose high risks on other drivers
because of their incompatibility with cars
Average subcompact and compact cars have similar combined risk
as average SUV
Driver behavior influences what we call risk
low risk to drivers of minivans and high risk to drivers of sports
cars
Driver sex and age do not appear to influence our main findings by
vehicle model
However, other driver characteristics or environmental conditions
(rather than vehicle design) may explain some of our findings
Quality of vehicle design appears to be a better predictor of risk than
vehicle weight
http://www.cars.com/carsapp/national/?&srv=parser&act=display&tf=/news/archive/suvs_analysis.tmpl
Study Concludes SUVs Not Safer Than Cars
Posted 09/06/02 10:17 a.m. CDT
By Joe Wiesenfelder
cars.com
While sport utility vehicles do represent a greater risk to drivers of smaller vehicles, they are not safer to drive, according to a study by the University of Michigan and the Lawrence Berkeley National Laboratory. In comparing specific models of all vehicle body styles sold from 1995 to 1999, the study supports one of the accepted societal shortcomings of SUVs and pickup trucks that they endanger others but refutes the perceived upside of safety for their drivers.
Funded by the Energy Foundation, a non-profit partnership of sustainable-energy concerns, the study also refutes conclusions drawn by the Insurance Institute for Highway Safety (IIHS), in which vehicle weight overshadows other factors. We thought it was wrong to take the weight of a car as the only safety characteristic, said Marc Ross, a University of Michigan professor of physics, who co-authored the new study with Tom Wenzel of Lawrence Berkeley. The report states: The argument that the low weight of cars with high fuel economy has resulted in many excess deaths is unfounded; by paying careful attention to safety in vehicle design, smaller cars can be, and indeed have been, made as safe as larger ones.
The subcompact Honda Civic was as safe as the average SUV during the study period of 1995 - 1999.
The study is based on driver death rates from National Highway Traffic Safety Administration (NHTSA) data, and includes only the most-popular models sold for the entire five-year period, for statistical validity. Unfortunately, this excludes all SUVs with unibody (car-based) construction, which were introduced in the middle of the period analyzed. The best snapshot of the study is its Figure 2, in which the risk to the driver is plotted against the risk to drivers of other vehicles (the latter only in two-vehicle collisions), for all vehicle body styles.
SUVs vs. Cars
In terms of risk to their drivers, the average SUV, large car and midsize car were statistically equal. The risk to drivers of other vehicles was significantly higher for SUVs than for cars of all sizes (excluding sports cars), which supports earlier studies that tallied 1.8 times as many deaths in SUV-to-car crashes as in car-to-car crashes. Compact and subcompact cars posed a risk to their own drivers that was higher, on average, than that of the larger cars and SUVs. Note, however, that lines on the Fig. 2 graph depict ranges to reflect the risk ratings for all models studied (the circle is the average for that class). The subcompact cars varied profoundly in their driver risk factor, and overlapped most of the safer vehicle types.
The study notes that the safest subcompact (Honda Civic and Volkswagen Jetta) and compact (Mazda 626 and Nissan Altima) car models were as safe for their drivers as the average SUV. The Chevrolet Cavalier (and Pontiac Sunfire), Ford Escort and Dodge and Plymouth Neon models carried twice the risk of the Civic and Jetta, which raised the average risk for the whole class.
Even the safest full-size SUV of the period, the Chevrolet Suburban, was no safer than the best-rated midsize and large cars, the Honda Accord and Toyota Camry and Avalon. Likewise, the average vehicle from each of these classes was roughly equal in driver risk.
Other Light Trucks
So where do other light trucks fall? In opposite directions from the SUV, both in terms of risk to driver and to other vehicles. The average pickup truck posed a risk to its own driver comparable to that of the average compact and subcompact cars, but its threat to drivers in other vehicles was almost double that of the average SUV and triple that of cars. Minivans represented a lower risk to their drivers than all trucks and cars except import luxury cars, which are a separate category. The vans posed a higher risk to other drivers than cars, but not as much of a threat as SUVs.
Import Luxury and Sports Cars
The extremes in this study are represented by import luxury cars, which have the lowest risk to their drivers and other drivers, and sports cars, whose risk factor for their own drivers is nearly double that of the next worst class. The risk sports cars pose to other vehicles is roughly the same as the average SUVs, making it the only type of car that comes close to the SUVs threat.
The sports car rating is perhaps the most glaring example of the effects of behavior on driver fatality, and thus the studys results. The study, like many before, explains that the high risk factor comes not from the inherent design of sports cars but from who drives them and how. Sports cars tend to be driven aggressively, and are owned overwhelmingly by young males, who are considered the most dangerous drivers. Behavior plays a part in all of the classes rated.
Behavior vs. Design
Ross said in an interview that behavior accounts for a variety of factors, including: vehicle modifications; vehicle maintenance; seat belt use and where, when, how and how much a vehicle is driven. Factors inherent to the vehicle include: its performance (how well it stops, handles), rollover propensity, weight, safety features and how well it manages crash energy and protects the occupant in a collision. The real-world fatality data account for all these things, and reflect single- and multiple-vehicle collisions of all vehicle sizes and shapes from a wide range of speeds and angles of impact something crash test ratings dont do. That it accounts for all this behavior and usage as well as vehicle design is the datas strength. Its weakness is that variables are difficult if not impossible to separate.
The study acknowledges that pickup truck data are highly influenced by the how and particularly the where of their use. Pickups are most popular in rural areas, where speeds may be higher and road conditions poorer. Two-lane rural roads host a disproportionate percentage of fatal collisions because no space or barrier separates opposing lanes, and vehicles pass each other and intersections at high speeds. On interstates, the speed differential is less among nearby vehicles. Still, the study cites earlier findings that a substantial part of the risks light trucks impose on other drivers is associated with the design of trucks. The report says the same is true for SUVs: Some of the higher risk in SUVs relative to cars is due to the tendency of SUVs to rollover and the danger of these types of crashes to unbelted drivers.
The other end of the scale has behavioral components as well. The study notes that minivans are driven with care to protect children, and seldom are piloted by young males. As for import luxury cars, The joke is that if you buy a Mercedes or a BMW, you want to avoid getting the car scratched. Therefore you dont drive it in a way that will get you killed as often, Ross said. That�s part of it. On the other hand, they are [inherently] very good vehicles.
Risky Drivers Influence
The study includes an examination of risky age/sex groups: male drivers under age 26 and drivers over age 65. We dont see any evidence that elderly drivers are dangerous drivers, but we see plenty of evidence that they�re frail, Ross said. Because this group made up more than 50 percent of the drivers in the fatality data for four of the large cars, the researchers suspect this car class would appear safer if the results could be adjusted for driver frailty. The results arent always as expected.
Some of the subcompacts like Civic and Jetta that have very low risk factors actually tend to have a lot of young male drivers, Ross said. But that doesnt make them dangerous cars. Those relationships are not simple.
Most important, the study concluded that buyers of SUVs and midsize cars are similar enough in age and gender that the margin of error for these two groups is small. Some cars have a lot of drivers in high risk groups. SUVs dont, and midsize cars dont, Ross said.
What the study doesnt, and arguably cant, do is separate the influence of any behavioral difference between midsize car and SUV drivers in part because the same driver may behave differently in either vehicle. But the researchers emphasize that behavioral considerations dont overwhelm the studys conclusions, most notably the importance of vehicle design over the weight criterion. It states: Careful vehicle design plays a more dominant role in vehicle safety than size or weight. For instance, the foreign subcompact and compact cars have almost the same risk as the domestic large and luxury models, while the foreign midsize cars have a substantially lower risk than the larger domestic models. While at first glance the figure may suggest that, at least within each manufacturer group, vehicle risk decreases as vehicle size increases, the design factors overwhelm the influence of size.
That Was Then . . .
This assertion may help explain the minivan categorys supremacy over other light trucks, because most of the vans in the data were unibody designs. (Only the Chevrolet Astro Van and Ford Windstar were body-on-frame [truck-based] vehicles, and their sales figures were dwarfed by the others.) So is it possible that the minivan�s low-risk rating comes in part from its having the weight of an SUV but not its detrimental high center of gravity or construction? I agree with that, Ross said. We see hints of that everywhere. The Aerostar and Astro van are somewhat riskier [than the unibody minivans]. He said the Jeep Grand Cherokee, which combines attributes of unibody and body-on-frame construction, is less risky than the Jeep Cherokee, which is purely body-on-frame.
It bears noting that no SUVs in the study were unibody designs. Roughly 20 of the 2003 model SUVs are car-based, and have more in common with minivans than conventional SUVs. Ross noted that changes in truck-based SUV design are now hitting the market as well, in the form of the Ford Explorer and other models. With reinforcement in the roof, and a wider track and a lower center of gravity, I think its probably true that SUVs will in the future become quite a lot safer, he allowed. Its going to take some years, and the vehicle then has to be around for several years in order to collect data on it. It will be some time before some effect is evident. Sooner than that, well be able to see if Honda and Toyota unibodies have lower risk.
Ross agreed that todays perception of SUV safety may be on its way to becoming reality, but added, Im outraged that this problem that was well known 10 years ago is only being dealt with now.
Copyright 2002 by cars.com;
Image courtesy of the manufacturer
Most of the vehicles I see at my younger children's school are SUVs, vans are in the minority. A lot of people use SUVs to haul boats though. Also, you'll find a lot of super-shiny (without a spot of dirt on them) Hummers. 
I just checked fueleconomy.gov and there's only one minivan with MPG 19/27. Most are in the 18-19/25-26 range. My SUV is 17/22 -- yes, lower than minivans, but not by much. And there's a bunch of SUVs, including most of the smaller ones, hitting 23-24/28-29.
