What is the only consensus on samples? Samples lie. Even if you go through all the steps you have in the finished piece, samples will still lie. Either because the width and length of the cloth is different, so the fabric behaves differently or the loom is different, the washing takes 5 minutes longer, or the sun didn't shine enough, one never knows exactly why, but samples are devious beasts.
I've been thinking about comparing apples with pears and such. Would you get more accurate data if you could mark a 10 cm piece of yarn with two very tight choke ties to be able to do most of your measurings on the same centimeters of yarn all the time? Do this in a few spots with differently marked or colored ties.
Just trying to add to the crazy. ;^D