Quote Originally Posted by carraway View Post
If you're performing controlled tests, you shouldn't purposely introduce variance.

When I tested STR vs. ATK at 2:3 for LNC, I targeted one mob at one level (L50 Amalj'aa Drubbers) and performed ~200 attacks (along with ~75 Moonrise IIs) to establish confidence to my satisfaction, with no buffs used that would affect the results. Parsers don't calculate for buff and level variance because we can't calculate for buff and level variance.

The tests performed in this thread are sufficient to demonstrate that DEX affects ranged attack in a similar fashion to how STR affects melee attack. Since STR:ATK is roughly equivalent at a 2:3 ratio, I also think it's safe to say that DEX:ATK at 2:3.

Where it gets a little fuzzier for me personally is how dLVL affects the ratio, but looking at the boosts from stats against L1 targets in this thread and comparing them to analogous tests against L50 targets, my theory is that the STR/DEX:ATK ratio is preserved, and dLVL operates separately.
Agreed, I never stated specifically that you would deliberately add variance. However, mobs aren't guaranteed to respawn at the level you killed them, nor can you control whether or not they hit you with a debuff that may affect your damage. My point regarding MoE was that there are always factors in testing that you may have to account for.