I'm by no means a scientist, but wouldn't the density of the gas change, like everything else, from hot to cold? Everything I can think of expands and contracts as the temperature changes. That alone will give you different results based on season, unless not.
Then, I believe in the summer they throw in ethanol which has almost 30% less energy than gasoline, which translates into a few % less fuel economy, which I wonder if that negates the temperature fluctuations to some degree
Anyway, the point is there a lots of ever changing variables, which is why you need long term data, that is fairly consistent to be accurate.
Hell, I always wonder how much remote starting effects your numbers, 10 minutes of gas being burned going zero miles has to screw it all up.