Crash tests don't tell full story

Starting with the 2010 model year, the government will use dummies akin to 108-pound women in crash tests. (Peter DeMarco/Globe Correspondent)

For the past five years I've driven a big, bulky Chevy Tahoe. It's been a great vehicle, but for my next car I'm determined to buy something far more energy efficient - and probably far smaller. The question is, will I be trading safety for better mileage?

Looking at a Prius and my Tahoe side-by-side, I think I know which car would fare better in an accident. But do crash test scores back my assumptions? This week, we find out.

Car vs. SUV

We covered crash test essentials in our last column, explaining how the National Highway Traffic Safety Administration rates vehicles on a scale of one to five stars in three separate crash tests: frontal (both driver and passenger sides are rated), side impact (both sides), and rollover (details at www.safercar.gov).

So, if a Prius and a Tahoe both get five-star frontal crash ratings, they're equally safe, right? Not exactly.

The government crashes each vehicle at 35 miles per hour into a solid barrier. The resulting score shows how that vehicle's occupants would fare in a head-on collision with a vehicle just like it. It does not show how that vehicle's occupants would fare in a head-on accident with a larger or smaller vehicle.

To figure that out, the government would need to crash each vehicle against a litany of other vehicles of different sizes and weights, which it just doesn't have the resources to do.

Side-impact and rollover tests are different. In the government's side-impact test, a barrier moving at 38.5 miles per hour is slammed into a vehicle's doors. Since the barrier is the same for all vehicles, the results can be compared across vehicle classes. So you're just as safe in a Prius with a five-star rating as you are in a Tahoe with a five-start rating if you get T-boned. Same goes for rollover tests.

Almost everyone agrees that a larger vehicle will protect you better in a head-on accident. But SUVs typically roll over more often than cars with lower centers of gravity, so it's a crapshoot as to which type of vehicle is safer overall.

"You can get killed in a rollover as easily as you can get killed in a side-impact crash. You have to look at the big picture," said Rae Tyson, spokesman for NHTSA.

The safest bet - but certainly not the most fuel efficient - might be a large sedan with five-star ratings across the board, Tyson said.

Offset ratings

The Insurance Institute for Highway Safety, a nonprofit organization funded by the insurance industry, is the nation's only other major car crash tester (www.iihs.org). The government recommends consumers look at their test scores as well, so you know the institute's methods are sound. But they conduct tests a bit differently.

Instead of crashing cars directly into a barrier, the institute angles them so that less than half of the front grille area hits the target. The idea behind such "offset" frontal tests is that head-on collisions aren't as common as, say, striking a utility pole. And vehicles that score fine on a head-on test don't necessarily do as well if the crash is concentrated on one side of car, the institute has shown.

Their frontal tests are conducted at 40 miles per hour, 5 miles per hour faster than NHTSA's tests.

The institute's side-impact test is also "more sophisticated and more severe" than the government's, Consumer Reports magazine says. The institute uses a high barrier, shaped like the front end of a pickup or SUV, as well as more advanced dummies, including "small females," equipped with sensors that measure head and neck injuries. The government's barrier, by comparison, resembles the front end of an old car, so it can't replicate an SUV hit. The government's crash ratings also don't factor in potential head and neck injuries, a significant failing, the institute says.

Scoring update

NHTSA officials, aware of the shortcomings of some of their tests, announced big changes to its program this summer.

Starting with the 2010 model year, the government will use dummies akin to 108-pound women and child dummies in some tests. Dummies used in side-impact tests will also be equipped with sensors to measure neck and head injuries, as the institute does. A pole-striking test will also be added to the side-impact test, though the lower, car-style barrier will remain.

Vehicles that are equipped with certain anticrash technologies, such as electronic stability control and forward collision warning systems, will get special mentions.

Perhaps most important to consumers, NHTSA is also going to award each vehicle with an overall safety rating, a first for the government. The overall score will factor in the results from the three crash tests, with scores weighted according to the likelihood of each particular accident in the real world.

"With the system we have now . . . you might have a vehicle that got three or four stars in both the frontal positions, and three stars in the side positions, but only got two stars for rollover," Tyson said. "If you're in a crash with another vehicle, your protection is going to be high. But the bad news is, this vehicle is more prone to roll over. So we're really forcing consumers to make choices about what is most important to them.

"Our new rating system is going to help people because we will do that for them and give them an overall score."

The Insurance Institute for Highway Safety, it should be noted, says the government's changes don't go far enough. It has issued a warning not to base buying decisions strictly on the government's new, overall vehicle ratings.

"Poor performance in one test could be masked by good performance in others," institute president Adrian Lund said. The institute will continue its policy of awarding "top safety picks" only to vehicles that score highly in every crash test.

Tyson said NHTSA will make further changes to the testing program if needed over time.

"You're creating artificial conditions to measure safety. Any time you do that, there will be shortcomings," he said. "But we've developed what we believe to be potentially, beginning with the 2010 model year, the most meaningful consumer safety ratings you're going to be able to find."