As an example, consider two pairs of vehicles. The first pair is two vans, one rated at 11 MPG and the other at 13 MPG, and the second pair is two cars rated at 29 and 49 MPG. Many people intuitively believe that conditional on gas price and miles driven, the difference in fuel costs between the second pair is much larger, because the difference in MPG is much larger. In fact, the fuel cost differences are almost exactly the same: the difference between each pair of vehicles in gallons of gasoline consumed per mile driven is 0.014. Interestingly, if this misperception has any magnitude outside of the laboratory, it would make us more likely to buy a Hummer and more likely to buy a Prius, while making us less likely to buy a medium-MPG car. This is because it causes us to underestimate the relative costs of the lowest-MPG vehicles and overestimate the relative savings of the highest-MPG vehicles.
In short, when it comes to making cars more fuel-efficient, what might look like modest MPG gains for the lowest-MPG vehicles actually matter quite a bit more than whether someone who is already in a high-MPG vehicle makes additional gains.
And maybe some smart marketing person or heads-up government regulator should be figuring out a way for car buyers to learn about how many gallons it takes to travel a mile–or 100 miles?–not how many miles to a gallon.