IIRC, most cam companies suggest making lash changes for testing in .004" steps (assuming the lash ramp allows for that much range of adjustment); .002" makes a difference, but not much of one. That's part of why I posted the link above, in order to quantify some actual at-the-valve duration changes resulting from the different tests referenced.

Tell four different people to set the lash to .018" on the same engine and you could end up w/ actual lash ranging from .016" to .020" simply because of the variations in their technique.

For engine dyno testing, I believe that repeatability within 1% is considered very good. But that's still saying that if pull #1 made 600 HP and pull #2 made 606 HP w/ absolutely no changes between them, they're still considered to have "repeated". I doubt a lash change of .002" can be validated w/in a 1% variance, even if there is some other characteristics that change as a result (e.g., if you can A-B-A test an idle speed increase).