I think the EPA should show several different estimates for different driving styles. Say something like conservative, average, and spirited. For example a bigger engine say 3 liters is not exactly less efficient than a small engine say 2 liters in every application. Going 50mph the small engine might get 50mpg and the bigger engine might get 40mpg so the smaller engine is 20% better in this example, but what about a more typical highway speed of say 80mph. Now you might see them both getting 25mpg. The fact is bigger engines tend to be more efficient at producing power, that is why big trucks have a big engine and not 20 small engines.

An extremely simplified way of looking at this would be the lines 2x + 50 and 3x + 20 where the line is fuel usage or cost and x is the horsepower you need. So if you want 10 horsepower you are going to use 70 units of fuel with the big engine and 50 units of fuel with the small engine, but if you want 100 horsepower you are going to use 250 units of fuel with the big engine and 320 units of fuel with the small engine.

 

That is not a bad idea.

The current EPA testing is about "average". You can guess that the mileage numbers for "spirited" would be lower and that the numbers for "conservative" would be higher.

 

I am sure what is preventing this is partly the time it would take to conduct multiple tests. The other problem is that most would be too stupid to deal with that many numbers.

There is probably some sort of estimating overall economy based on driving style and the type of driving, but that would be too confusing for most too. Unfortunately there are several variables involved in actual average fuel economy, you would be lucky to be accurate within 5-10% for 95% of drivers. Although you could probably use a gps to log speed and elevation for every second the engine is on. Then you would know the type of driving, whether highway cruising, city, uphill, downhill, idle time, etc. You could then simulate how much gas each car would use based on last years driving habits.