“Miles per gallon” (mpg) is the most common measure of a car’s fuel efficiency. The typical U.S. consumer, in shopping for a car, uses mpg as a way of calculating gas consumption and carbon emissions.

Because the concept is in such wide use, mpg has become as familiar to the American ear as FBI, CIA, or ABC.

But miles per gallon is not the best way to measure how fuel-efficient one car is compared with another.

That’s according to Richard P. Larrick and Jack B. Soll, management professors at Duke University whose math-intensive argument, “The MPG Illusion,” appeared in the magazine Science last summer.

Larrick was at Harvard on April 9 to make the case for an alternative metric of automotive fuel efficiency — one that favors volume over distance: gallons per mile (gpm). He and Soll, in fact, like using this measure per 10,000 miles. (That’s the average number of miles Americans drive in a year.)

Car buyers wrongly assume that gas consumption measured in miles per gallon goes down in a straight line: The higher the mpg, the lower the gallons of gasoline burned.

But in reality, fuel efficiencies are curvilinear, said Larrick. The higher mpg ratings go up, after about 20 mpg, the more efficiencies flatten out.

Because of this misperception — Larrick called it “the mpg illusion” — people underestimate the value of improving a gas guzzler’s fuel efficiency. Even improvements of a few miles per gallon help, said Larrick.

He offered an example: If you trade in a 34 mpg car for one rated at 50 mpg, you reduce gas consumption by about 94 gallons over 10,000 miles.

But if you trade in a car rated at 16 mpg for a model rated at 20 mpg, you reduce gas consumption by 125 gallons over the same distance.

You get “big gains with small changes in big vehicles,” said Larrick.

But the mpg illusion means that consumers scoff at improving mpg ratings at the low end of the efficiency scale — and too readily praise improvements at the higher end of the scale.

Larrick and Soll (who are carpooling friends) conducted studies of hypothetical car purchases based on perceptions of fuel efficiency.

They found that buyers are willing to pay a high cash premium for fuel-efficient cars based on misperceptions of how much fuel is actually saved.

“People are willing to pay too much for a very efficient car,” said Larrick. “But they should also see the value of moving (mpg) out of the teens and into the 20s.

He made his case for gpm to an audience of 25 at the Harvard University Center for the Environment (HUCE). Joining him in a dialogue was behavioral economist Max Bazerman, Jesse Isidor Straus Professor of Business Administration at Harvard Business School.

“There are certain deceptive qualities in mpg,” agreed Bazerman. “Those low miles per gallon numbers all look the same.”

HUCE periodically sponsors such “green conversations” as part of a series of events on the personal and public dimensions of energy usage. OK, admitted Larrick: In a perfect world, everyone would buy a 40 mpg Honda Civic, or a similar high-mileage vehicle.

But the reality is that a lot of American cars already on the road are not as fuel-efficient. So part of the policy over fuel consumption should be making even small improvements in fuel efficiency for less-than-efficient vehicles.

“Small improvements in big cars are good,” said Larrick.

Beyond these small improvements, he advised using buy-back programs and market incentives to phase out the worst gas gulpers — those with mpg ratings in the teens. The idea, cash for clunkers, has already been widely adopted overseas.

European countries, Larrick pointed out, already use a gpm measure of fuel efficiency. (In Great Britain, for instance, a car’s fuel efficiency is expressed as liters per 100 kilometers.)

Adopting a gpm measure underscores the beauty of a cash-for-clunkers plan, he said. Replacing a car that gets 14 mpg with one that gets 25 mpg, for instance, saves 300 gallons of fuel over 10,000 miles — the equivalent of avoiding 3 tons of carbon dioxide going into the atmosphere.

As of January, three cash-for-clunkers bills are on the table in Congress. One would give new car buyers a credit of up to $5,000 for buying a U.S.-made car that gets at least 27 mpg. The car traded in must get worse mileage, must be at least eight years old, and must be junked after the new-car sale.

As for miles per gallon: Keep it, said Larrick. It’s good at least for calculating the number of miles you’ll get out of a car’s gas tank.

“Miles per gallon is a powerful number in the consumer’s imagination,” he said. “It’s simple, meaningful, salient, and ‘fixed’ for a given car.”

There are ways to calculate gpm that are similarly “sticky” for the consumer imagination, said Larrick.

In the meantime, he advised simply adding the gpm figure to standard car listings, like Kelley’s or Consumer Reports magazine — and equipping dealerships with ways to calculate it for the benefit of new buyers.

Adding the unfamiliar gpm mile number is a tough sell, admitted Larrick — analogous to getting Americans to adopt the metric standard. “No one wants to explain that much math to make the switch,” he said.

“We’re not miscalculating” by using just mpg, said Bazerman, who likes the gpm concept. “We’re failing to calculate.”

For more on the gallons-per-mile concept, go to www.mpgillusion.com. And to do a little figuring for your own car, go to www.gpmcalculator.com.