You never seem to consider that the EPA estimates are the most accurate way to measure differences in fuel economy between engines (or vehicles, for that matter). Of course your mileage or my mileage may vary. That's why the EPA continues to publish a range in addition to an actual number for both highway and city mileage. Look on any window sticker and you'll see an actual number in addition to a range that is goes above and below that figure.
If Joe Blow says he got XX mpg and you're getting XX mpg, that really means nothing because there's no basis for comparison. There are simply too many variables to even make an accurate comparison. But under a controlled study, two or more vehicles are put under the SAME EXACT conditions, which makes the results the most accurate possible. So if the EPA publishes a 2.5 mpg difference (combined city/highway) then that's about as real world as you're gonna get. Remember, it's not about the actual mpg, it's about the difference between the two vehicles.