The world would be a better place if more people understood statistics.

This is not a dig at the OP, I agree with him.

I don't think all of the reviews are complete in part because they haven't accounted for the mounting variations mentioned.

What is needed is at least 50 different tests each with the same blocks and different mounts. So the same 10 blocks with different reviews, for example, but each block is remounted five different times.

The aggregate data can then be analyzed and the spread can be computed.

Here is an old but fair example of testing methodology.