Formula Predicts Rollover Risk


Advocates want roll-ratings; automakers say process is faulty

By Gary Stoller, USA TODAY

July 17, 2000

Rollover crashes kill one of every four people who die in auto accidents, the federal government says. Yet while consumers can find out how vehicles perform in government crash tests, they cannot tell which vehicles are most prone to rolling over in an accident.

Now they may be able to. An exclusive analysis performed for USA TODAY provides rollover predictions for 189 model-year 2000 vehicles.

The analysis was done by Joe Kimmel, an industrial economist. It is not based on road tests, but rather a mathematical formula that takes into account factors such as a vehicle's height, weight and width. Accidents are defined as those serious enough to be reported to the police and cause property damage, personal injury or a fatality.

Most rollovers occur when a vehicle runs off a road and turns over at least on its side.

Under Kimmel's analysis, sport-utility vehicles — a category that includes some of the USA's best-selling vehicles — account for 17 of the 20 vehicles with the highest likelihood of rolling over. Twenty-two cars, nearly all high-priced, such as Cadillacs, Mercedes-Benzes and Lincolns, scored lowest for rollover probability. The formula predicts:

Most sport-utility vehicles (SUVs) are likely to roll over in one of every four accidents. Some may turn over in nearly one of every two accidents.

For most cars, the chance of rollover in an accident is less than 10%. Some cars, such as the Chevrolet Metro and the Suzuki Swift, can roll over about 20% of the time — as often as some SUVs.

Minivans are more likely to roll over than most cars. They're less likely, however, to turn over than SUVs and pickups.

The analysis was done by Joe Kimmel, a Villanova, Pa., management consultant who has independently developed a mathematical formula for predicting the likelihood of a car, minivan, sport-utility vehicle or light truck rolling over in an accident.

At least three sources who reviewed Kimmel's K Index — the Center for Auto Safety, a consumer-advocacy group; an independent consultant who has done research work for the National Highway Traffic Safety Administration; and NHTSA's former chief mathematician — say it is as good as or better than a formula NHTSA plans to use.

Automakers criticize the work, generally saying that rollover rates for particular models can't be predicted mathematically.

Kimmel's analysis comes as the auto industry fights NHTSA's plan to use a mathematical formula to rate the rollover potential of new models this fall. Last month, the Senate Appropriations Committee voted not to fund the rating system and said the National Academy of Sciences should study rollover propensity. The issue awaits further action by Congress.

NHTSA already rates vehicles in crash tests and has the results on its Web site (www.nhtsa.gov).

Kimmel's calculations depend largely on a vehicle's dimensions and estimated center of gravity. Those factors, insurance industry and safety advocates say, play a major role in determining rollover probability.

Others, including auto manufacturers, dispute the validity of Kimmel's index and all mathematical formulas used to predict rollover. "The Kimmel Index seems to be an interesting academic exercise, which does little to provide meaningful information to the public," Toyota says. "Using a single calculation as the sole determinant of the potential rollover of a vehicle is not only misleading, it yields inconclusive results," says Mitsubishi Motor Sales of America spokesman Kyle Bazemore.

NHTSA statistics show that 9,771 people were killed in rollovers in 1998, the latest numbers available. That's nearly one-quarter of the 41,471 killed in all types of accidents. About 60% of fatalities in SUVs, and 40% in pickups, were in rollovers, NHTSA says. About 22% of fatalities in cars, and 30% in minivans, were rollover-related.

NHTSA has long warned that SUVs are more prone to rollovers than other vehicles. However, some consumers buy them because their height provides more visibility and they are regarded as good vehicles in snow. Experts, though, say their four-wheel-drive capability may not protect against icy roads.

Three SUVs, the Chevrolet Tracker, the Suzuki Vitara and the Toyota RAV4, according to Kimmel's K Index prediction, are likely to roll over in more than four of every 10 accidents — the highest rollover propensity of any type of vehicle.

The least likely SUV to roll over is the Ford Excursion, the K Index predicts. The vehicle will probably roll over in less than 13% of its accidents, Kimmel says.

General Motors, which sells the Tracker, says formulas predicting rollover may be consistent with the physics of rollover resistance, but they omit real-world factors that can affect the probability of a rollover. Vehicle design, GM officials say, plays a small role in avoiding rollovers. The best way to prevent rollovers and injuries, they say, is to increase safety belt use and curtail aggressive and impaired driving.

Brian O'Neill, president of the Insurance Institute for Highway Safety, a non-profit research group funded by auto insurers, disagrees. "'Let's blame all rollovers on the driver' has been the standard line out of auto manufacturers forever, and it's just nonsense," he says. "Each car type is more or less likely to roll over than another type. Consumer information provided by the federal government would be a first step, and there's a need for standards to get the most unstable vehicles redesigned."

George Ball, managing counsel for American Suzuki Motor, calls Kimmel's results unreliable. Contrary to what Kimmel predicts, he says, Suzuki's Vitara and Grand Vitara SUVs are more stable than other SUVs. Ball says Suzuki bases its conclusion on a formula called the static stability factor (SSF), which automotive engineers have used for decades and which NHTSA plans to use. That's the ratio of half the track width (the distance between right- and left-side tires) to the height of the vehicle's center of gravity. Subaru officials also point to the SSF, and spokesman Richard Marshall says Kimmel's calculation for its Forester SUV is "higher than expected." The Forester, says Subaru's Don Bearden , has a low-lying engine and a lower center of gravity than other SUVs.

Kimmel maintains that its light weight contributes to a 32% rollover probability. Kimmel says his K Index is better than the SSF because it considers the weight of a vehicle and its cargo, and past rollovers for different sizes of vehicles. K Index results for many older model cars, he says, are in line with state rollover statistics that NHTSA collected and SSF-based calculations by NHTSA.

Kimmel is aware, however, that his K Index and the SSF have their limitations. SSF doesn't include a vehicle's weight in its formula. And neither SSF or the K Index takes into account factors such as driver performance, weather, terrain, tire size, suspension and all-wheel-drive systems. It's possible, Kimmel says, that some of those factors may average out among model types.

The index "isn't the be-all and end-all rollover formula," says Kimmel, an auto buff and a member of the Society of Automotive Engineers. "It's a simplified, relatively reliable way to predict rollover propensity based on a vehicle's dimensions and center of gravity."

Kimmel, who became interested in rollover research in 1991, has been using statistical techniques to analyze problems in various industries since 1966. His clients have included seven governors, the U.S. Postal Service and the American Bus Association. He testified once for the plaintiff in a rollover-related lawsuit involving the Ford Bronco II.

Clarence Ditlow, executive director of the Center for Auto Safety, a consumer advocate, says, "The K Index is better than the SSF because it adds another factor (weight) into the equation."

Donald Friedman, a California consultant and rollover expert, agrees. Friedman formerly worked for General Motors Research Laboratories and says his own company received $30 million in NHTSA research and development contracts from 1968-86. Friedman says Kimmel's formula is consistent with other formulas for analyzing rollover propensity and praised him for taking rollover accidents into account.

NHTSA's official position is it doesn't know whether Kimmel's rollover index is valid. But Terry Klein, who evaluated K Index results for 30 vehicles while he was NHTSA's top mathematician, says the index "performed about as well" as the SSF. The K Index "appears as good as any other means for predicting rollover propensity," says Klein, who left the agency several months ago.

The K Index's conclusion that SUVs are the most likely, and cars the least likely, to roll over is in line with NHTSA data. A 1999 study , co-authored by two NHTSA officials, states that SUVs have "the lowest SSF values" and that "cars clearly have the highest SSF."

The Alliance of Automobile Manufacturers, whose 13 members claim 90% of U.S. vehicle sales, says it, too, believes SUVs are more likely to turn over. "You accept the fact that SUVs and pickup trucks are more prone to roll over because they are higher off the ground and have a higher center of gravity," says spokeswoman Gloria Bergquist.

Honda contests the K Index's prediction that its model year 2000 CR-V will likely roll over in one of every three accidents. Company officials point to a 1999 NHTSA study that shows the 1998 CR-V with an SSF better than many SUVs. Honda spokesman Art Garner says SSF and the K Index do not accurately predict real-world rollover propensity.

Kimmel says that other types of vehicles (from the same and other model years) have had relatively high SSF scores that were similar to the model year 1998 CR-V. A sample of state accident statistics showed they rolled over in 24% to 50% of accidents.

Top cars

The vehicles least likely to turn over, according to Kimmel's index, are Bentley and Rolls-Royce. They will probably roll over in 1% of their accidents, he says. "These are very interesting findings, however, not entirely unexpected," says John Crawford, a Rolls-Royce and Bentley spokesman. A vehicle's design, low center of gravity and weight play a major role in determining rollover likelihood, he says.

The cars most likely to turn over, Kimmel says, are three subcompacts: Chevrolet Metro, Suzuki Swift and Toyota Echo. They will probably roll over in at least 20% of their accidents and are about six times more likely to turn over than cars with the lowest likelihood of rollover, he says.

Ball, Suzuki's lawyer, says the auto manufacturer has spoken to "outside technical experts" who "have been unable to decipher Mr. Kimmel's equations because they are confusing, nonsensical and not presented in any kind of standard mathematical form."

"I respect his opinion and that he is doing his best to represent his client," Kimmel says. "We have people like Terry Klein, Clarence Ditlow and other people who think it's pretty darn good. "

Subaru takes issue with Kimmel's calculations predicting that the Impreza is more likely to roll over than the Legacy. It says that its calculations, based on the static stability factor, give the Impreza has a lower rollover propensity than the Legacy. Kimmel predicts that the heavier Legacy will likely roll over in 7% of accidents and Impreza, 11%.

Minivans

Most minivans, according to Kimmel's K Index, can be expected to roll over 10% to 17.9% of the time in an accident. The Honda Odyssey will likely be the most stable minivan, probably rolling over in 8% of its accidents, Kimmel says. The Ford Windstar follows closely behind, only a single percentage point higher.

Many auto manufacturers oppose a system based on a mathematical formula. Bergquist, the spokeswoman representing 13 manufacturers, says manufacturers "support giving consumers more rollover information, but it's important that consumers realize the merits and limitations of the information."

Since the 1970s, safety advocates have criticized NHTSA for lacking rollover standards for new vehicles and not providing consumer guidance. "The number of deaths keeps climbing, and with the great increase in the number of SUVs on the road, it's only going to keep increasing," Ditlow says .

NHTSA last year changed warning labels it requires inside SUVs. Text-only labels must now be in bright colors and graphics, and must read: "WARNING: Higher Rollover Risk," and "Avoid Abrupt Maneuvers and Excessive Speed."

NHTSA spokesman Rae Tyson says there's a reason it's taken long to provide rollover information by model to consumers. "We had to make sure we had an accurate, repeatable method of predicting rollover propensity," he says. "These things take time."