Just a couple thoughts... And great topic btw. The longer tumbling tends to be faster and, as you've noticed better for bounding passes. The other benefit it that a short landing is less punishment on the body because the rotation is faster and the body will continue to turnover instead of jamming into the floor.
I like front passes to be low and long until the last skill in the pass. So for a full, lay, double full... I'd try to be long, long, high. Same for back tumbling in that sense. Back 3/2, lay, 2/1 would be low, low, high. The last skill should go higher in my opinion to make the stick easier. So if they're just doing a back 2/1, take it up and stick.
Now for the reality of it... If the kid is purely JO and that's always the goal (especially if the won't be looking for a scholarship later) I usually train everything to be nice and high and floaty. They won't need to bound and they won't do incredibly high numbers and it's a hard concept. I teach those kids in a pretty traditional way. Future Hopes, elites, scholarship 10s... I teach passes to be about where you said, shoulder height until the final skill.
My thoughts pretty much mirror
@Hollowarchkick's. I feel as though single saltos, and the final salto in a series should always be as high as possible. This is to avoid amplitude deductions, to enable the athlete to "drop into" their landing, thus making it easier to stick, and because it's just plain awesome and looks impressive. I also think that most techniques should be taught with an emphasis on height initially during the learning stages, because at first it is a tricky concept learning how to set, and to create all the required forces necessary upon take-off in order to coast through the rest of the skill. Once the gymnast has become really proficient with the skill and the block and set work, they can start to play around with the technique and learn how to modify it depending on the objective of the skill. I would not advocate teaching distance on saltos initially. What I always do is I explain briefly that it can and will change in the future, but that's about all I say about it before nagging them to death for the next year about height.
As far as physics and biomechanics are concerned, leaning into the skill upon take off will cause travel, which actually assists rotation. This is why it's so important that connecting skills be long and low, because it maintains as much linear and angular momentum as possible, and it's just way more efficient. This is especially important when it's to set up a big, high, floaty skill at the end of a pass. Furthermore, height is not desirable for connecting skills because it adds more impact force to the rebound phase into the next skill, which will lessen plyometric ability. This will make it very difficult to connect skills unless the athlete's legs are ridiculously strong and quick twitch, but even so, it's still bad for the joints because they get more of a pounding from the added forces.
One more thing: when getting into double saltos, they should actually travel more and be a bit lower than their single salto counterparts. This is because you have to trade a little height for distance since it aids rotation. That's why double back layouts also travel noticeably more than double back tucks, because so much more angular momentum is needed for the larger shape used in a layout. Double lays are also lower as a result of this. Next time you see an elite do a double layout and it looks really high in the air, chances are the gymnast landed a little short, and probably had to take a step forward on the landing. Landing short hurts, too!
@Geoffrey Taucer, I love these types of discussions! You're usually the one to bring up challenging and thought provoking technique questions. Love it.