AB testing has terrible potential for errors. See my sig line for summer and winter average mileage in my truck, and this thread for where those numbers came from:
http://ecomodder.com/forum/showthrea...yon-17070.html
I regularly make a 60 mile trip. My MPG for that particular trip ranges from 30 to over 42, depending on wind and temperature. I can get even lower mileage in a good snowstorm.
Look at my fuel log. You can easily see the seasonal variation and the tank to tank variation. For example, yesterday's fill was lower than normal for the range of temperatures because most of the miles (almost 90%) were against a headwind, while the previous fill had slightly warmer temperatures and no wind.
Yesterday's fill: Average temperature about 40 degrees and 31.6 MPG.
Previous fill: Average temperature about 65 degrees and 35.2 MPG.
A Scangauge or Ultragauge is useful for short tests on a straight stretch of road where you can test a change ABABAB several times before the temperature and wind change. Tank averages tell what happens in the real world, but it is difficult to control the variables well enough to truly measure the results of a change. I discuss this somewhat in my mod thread linked above.
Don't give up and please keep us updated.