Image: Car Accident © George Doyle, Stockbyte, Getty Images

Sometimes a perfectly good theory can sound nuts. Take this one: Safer cars make the roads less safe.

Sounds as loony as the one about how you're better off not wearing a seat belt, right?

But the first statement is, in fact, a lot more credible -- and it leads to some interesting debate about driving behavior, accident rates and, ultimately, how insurance rates are calculated.

It's called the Peltzman Effect, a generally accepted principle that suggests that as people feel safer they take on greater risk. People have a set tolerance level for risk and adjust their behavior accordingly. Build a high-impact football helmet, and players will ram their heads harder. Improve car safety, and people will drive faster.

The danger comes when the added risk cancels out the benefits of the safety measures or, worse, adds new and greater risks. In football, for example, evidence is mounting that the fierce head butts of the modern game are causing long-term brain damage.

And it's increasingly clear that we feel safe enough in our cars to consume a meal, phone a friend or argue with our passengers at 75 mph.

So what if I crash?

University of Chicago economist Sam Peltzman conceived the theory in 1975 after finding that accident rates had remained stagnant in the 1960s following the introduction of federal auto-safety regulations. When drivers felt protected, they drove more recklessly, Peltzman concluded, putting pedestrians, cyclists and others at greater risk.

In the years since, vehicular fatality rates have dropped dramatically even as safety laws have mounted, refuting the notion that the Peltzman Effect offsets the benefits as a whole.

"There is some evidence that safety features that people get direct feedback from do cause them to drive less carefully, and the best example is studded snow tires," says Russ Rader, a spokesman for the Insurance Institute for Highway Safety. "But there's no evidence that this is widespread with things that don't give drivers direct feedback, like air bags" and electronic stability control.

In 1980, three years before the first state mandatory seat belt law went into effect, 51,091 people were killed in auto crashes in the U.S., a rate of 3.345 per 100 million vehicle miles traveled. In 2010, an estimated 32,788 were killed, a rate of 1.09 deaths per 100 million vehicle miles traveled. (For a graphic illustration of how times have changed, check out this test crash of a 1959 Chevy with a 2009 Chevy. Hint: The driver of one would have been killed instantly; the driver of the other would have sustained a knee injury.)

"If it were affecting drivers significantly, then we would not have seen the dramatic reductions in crashes and fatalities that have occurred over the last 50 years, since auto safety regulations were first implemented," Rader says.

More from CarInsurance.com: