Drivers tend to be more aggressive in vehicles equipped with air bags, apparently because they feel more protected. This aggressiveness offsets any safety gains for the driver and passengers. It even poses additional risk for other drivers whose cars don't have air bags.

These contentions were borne out in a study of the 207 fatal crashes in Virginia in 1993 involving post-1989 model year passenger vehicles.

• Only one vehicle was involved in 70 of the fatal wrecks, and in 33 of those, only the driver was killed.

• Forty-four percent of post-1989 vehicles contain driver-side air bags and air bag-equipped cars are 76 percent as dangerous as cars equipped only with seat belts, so 11 of the 33 deaths should have taken place in air bag-equipped vehicles (O.76 x 0.44 x 33 = 11.)

• In reality, 16 of the drivers killed were driving air bag-equipped cars.

• This difference is statistically significant. and the chance that such an occurrence was the result of purely random behavior is less than 1 in 20.

Passengers in cars with air bags also died at a higher rate.

• Thirteen of the single-car accidents killed the passenger(s) but not the driver.

• Nine vehicles (69 percent) were equipped with air bags instead of the five to six (44 percent) that would be expected.

• There is only a 5 percent probability that such an outcome occurred by chance, making these results also statistically significant.

Nor were drivers in non-air bag cars safe from drivers with air bags.

• Of the 137 fatal multiple-car collisions, 30 involved one vehicle with an air bag and another without.

• Random chance would dictate that half, or 15, would be initiated by the driver with the air bag and half by the driver without; instead, 22 were initiated by the driver with an air bag.

• The probability that this outcome occurred by chance is less than 1 percent, showing that the influence of the air bag on driver behavior was significant statistically and otherwise.

These findings do not indicate that more dangerous, accident-prone drivers purchase models equipped with air bags. If that were the case, the personal injury rate should have been highest for the earliest air bag models and should have declined in later years as safer drivers bought air bag-equipped vehicles. In fact, injury claims for air bag-equipped cars rose in model year 1990, fell in 1991, rose in 1992 and fell in 1993. David R. Henderson and Chelsea Haga

Source: Steven Peterson, George Hoffer and Edward Millner, "Are Drivers of Air-Bag-Equipped Cars More Aggressive? A Test of the Offsetting Behavior Hypothesis," Journal of Law and Economics, October 1995.