I'm not sure if I'm reading this correctly, but I think a 40 mile sample is too low to accurately figure out your mileage. I mean you should probably do it over a whole tank or so, and see how much it differs from the OBC then. I find that the OBC is very slow to update its values, and usually as once I start riding on the highway the economy only goes up after about 4-5 min. With a bigger sample you can see more precisely what kind of correction you need and correct the OBC's values so that it gives true mileage.