But in all those threads I didn't see anyone who said: I tried RNG/* at lvl ## - it worked out like this ~~~~, so I suggest using [this/another] job combination. It just seemed that all the posts were backed with theory or 2nd hand experience rather than first hand (even though I am sure at least some of the responders had first hand knowledge of some of the job combinations). I was guilty of this myself in a couple of posts.
You will see this a lot. The reason why is because there is a good deal of time that needs to be invested in order to say, "Yes, I've tested it, and RNG/SAM is, in fact, worse than RNG/NIN at LVxx."
Essentially, you don't need to equip the subjob in order to see the difference. The differences in stats between the jobs are relatively trivial when you take into account all the bonuses you should be getting from equipment. The big differences are in job traits and abilities.
Berserk will give you +25% to attack.
Dual Wield effectively gives you +10.5 RACC.
Utsusemi gives you ~3 attack evades.
Pre-60, /SAM gives you one passive TP up (+1% per hit when SAM is subbed) at 20/10, and Third Eye at 30/15. Third Eye is nice to use before Barrage, but it's on a 60 sec timer, and only dodges one hit. Utsusemi is clearly superior (and rightfully so, it's a spell); 3 hits, 30 sec timer. When it comes to TP, /NIN will still outgain you with DW daggers. DW daggers attack at 90% of original delay, which is a bigger percentage increase than the 1% TP gain from the first Store TP.
Pretty much all of the above is simple mathematical number-crunching. Now, if two people go out and actually use RNG/SAM and RNG/NIN pre-60, how can the comparison be fair? They will most likely be using different equipment, fighting different mobs, with different party compositions. You can have two people with identical
jobs go out and report back with different results, so it is to be expected that two different jobs will have different results, as well.
Anecdotes are not the best way to make a comparison (and in order to make a controlled comparison, we would have to have a lot of information about the game engine that Squenix obviously would not allow).
The most unbiased way to compare the jobs is to take the (limited) hard statistics that we do have access to and crunch those.