[Piotrsko]

Im gonna say that the available test measurement total accuracy is the only thing that affects the test. If it's a nonstandard test parameter, then you need to sample more and different to get the boundaries of that sample, so going up and down, left or right gives you meaningful data, but only if you measure all the data available. Besides, nobody takes only one data point for a test, you use the mean. If I use only one data point, my F250 gets 99mpg. (Down a steep hill towing a 7,000lb trailer with a tailwind, instantaneous)