I read an interesting analysis decades ago on the thousandth digits and the brain's bias. With enough shots the thousandths should appear with equal frequency, but they don't. We all have our own bias, so .001 and .009 may appear twice as often as .002 and .008, etc.
I agree about the unconscious bias. I saw it in the 80's when we were running a lot of three wire levels for photo control for a 100 mile proposed highway. At the time we were not using GPS for elevations, so we ran 100's of miles of levels. I could see that some numbers occurred more often than would be expected if there was no bias.
I had one level party that would only check the digits in the thousandths place, and not check the hundredths or tenths between the three wires. Once I figured out what they were doing, they had to rerun a lot of lines, because the i-man misread these other digits occasionally and it was not caught. That was not using a micrometer.
I believe that the use of a micrometer mostly removes the bias, since you are no longer estimating as much. You are only estimating between two graduations (i.e. one or the other), not interpolating to a tenth of a graduation.
That is interesting. I hadn't considered that before. I always tended to think that my estimation of millimetres, on a metric E rod (in non-precision applications), was unbiased and therefore and random errors would cancel each out over a run.
