- From lambda 1.000 to 0.800 it is within 1%
The closer you get to 0.700 its out by over 7% or so on average
For lean running ~1.200 its about 4%
If you are using a dyno tail pipe probe to calibrate your sensor you need to think about pressure effects in the exhaust (I have a video on this that shows in detail more), is the exhaust system perfectly sealed *no air leaks*? if you are in a Nazi type gas chamber set up the lack of pure air around the sensor which the reference cell uses is something allot of people neglect to factor in. When was the last time your dyno person calibrated/refreshed their sensor? how old is it?, is the reference meter able or the user/operator able to program in the daily difference in %O2 in the air (it changes by the day and has large effects). There is a stack of things to consider before blindly following what the 'dyno operator' says as gospel. Again these differences are only of greater importance the further you deviate away from stoich, anything richer than 0.800L and you really want to be knowing what you are actually measuring.
Hope this helps a bit.