Tony, I believe Dave is saying that he needs to calculate a correction factor on both sides of stoich. I believe the implication is that the sensors are not off equally on each side and that the data scatter is such that a regression program is needed to come up with the correction factor.
Now, being the devils advocate that I am I have to ask "so what?" What is really important to the performance minded is a cal that produces a safe, best power condition. That means only the rich side of stoich is important and significantly richer than stoich. That is where the accuracy is most important. That is really the opposite of a calibrator, wher the proximity to stoich is very important. And as you lessen the f/a ratio, the resultant variation is much more important. The difference between .1000 and .0950 may be a little black smoke. But the difference between .0700 and .0650 may be nicely holed pistons.
So, for most of us the most important range is .0750 to .1000 (10 - 13.3 a/f for the guys that look at it the other way). Dave how accurate did you find the wide band sensors in that range?