All of it at once.
Same. I don't do any dominance training though, but I do have expectations, sorta NILIF.
I start with Positive Reinforcement for teaching
all new behaviors. I used to use a clicker, but I am bad at fumbling treats and leashes and clickers so my dog's 'click' is actually me using "Yessss!" as a marker. When I shape something new, "yessss" + treat is how we communicate.
After a behavior is already known I use a balance of corrections and rewards to proof and really solidify exactly what I want. "yess" when correct + reward, "No" + correction when wrong. I don't need to correct often anymore, much like "yes" has been conditioned as a reward, she understands "no" means she's not doing something right.
I didn't introduce corrections until my puppy was older. As my trainer explained it to me, if you only reward when they do something right, you only teach half the command. what happens if they choose to ignore you? they do not know what that result is - in some cases "oh I will miss out on a reward" is not enough incentive for my dog to do the command. By introducing a correction it's extremely black and white: they know exactly what outcome is for each thing they decide. This worked for my particular dog, probably isn't necessary for other dogs, just what works best here
She likes to train a lot and has gained a lot of confidence with the balance of knowing exactly what I"m asking.