[MetroMPG]

As much as I blather about A-B-A tests, I wouldn't rule out long distance (and also long term) testing as a way to measure meaningful trends in fuel economy... provided it's a big sample.

There are a few members here who are meticulous record-keepers (recording the results of every single commute, both ways, with environmental and other info as well) who can spot a deviation from the norm just because the sample is so good.