But the tests are still a useful tool and allow a basic
mileage comparison to be made.

Of course. What I said was that the *fleet averages* are nonsense, because
they don't even include the cars that people want to drive, by calling them
SUVs instead of passenger cars and leaving them out of the counting. Like
most technical analysis provided by the government, the information is not
the plain truth, but what promotes political and commercial expedience.