Understand that I am all for procedures that improve the accuracy but this one, in the case of the Prius c, is going the wrong way:
There are a statistically significant number of user reported mileage to show the 2012-16 was reasonably accurate. Even using a larger pool from
Fuelly - Track and Compare your MPG
The earlier EPA estimates are pretty darn close. But the 2017 version is ~8% too low.
Speculation on my part but at one time the EPA applied a 'correction factor' to derate vehicle mileage. I'm wondering if this correction factor was applied to the more accurate results and thus driving it to error?
Bob Wilson