One thing almost everyone always forgets is the fatigue multiplier. Bethesda was pretty lazy with most of the formulae based on chance (skill + stat*0.2 + luck*0.1 vs the same for NPC or minus the cost), so it's pretty easy to figure out if you determine the GMSTs' effect correctly. GMSTs regulate most of the game's formulae, so it is logical to start with them, not with testing. I'm thinking you guys started from the wrong direction.