I would say that the position of the shock doesn't really matter.
What matter is :
- The path of the wheel axis
- The shock dynamic (How the shock is compressed regarding the position of wheel)
1) The path of the wheel axis
If the path is circular then the braking force will bend the suspension.
If the center of this circle is lower than the front of chain (where the force comes from), then the force of pedaling will also bend the suspension (added to the weight of rider). This is called "pumping".
To circumvent these problems the engineers have :
- rised pivot point above chain, so the force of chain counter the weight of rider. This was the first solution. It is found today on cheaper bikes as it just require to move pivot upper, this does nothing against brake force.
- virtual pivot point and linear path. This is more expensive as it need more pivots working together. My favorite design (from engineer point of view) is parallelogram design that you could find on Giant Trance for example. The rear is a full triangle connected to front part with two links. (look on close ups from left side, just above pedal axis, you'll see that second link).
2) The dynamic of shock
A metallic spring has a linear strength each time you apply one unit of mass, you get one unit of movement (approx). This is good for the suspension. But metallic spring weights a lot.
Air springs are lighter but they are not linear, the more you compress them, the harder they get. To circumvent this engineers use levers and play with angles (virtual pivot point).
Lot of designs are mechanically equivalent. With levers you can place shock where you want, but some places makes more sense as they apply forces where it is easier for the frame to bear them. The goal of the game is to get a good behaviour, fewer pieces, less weight, better reliability and strength, and not infringe a competitor patent. In the end, the actual position of the shock is as much a question of patent and look than real mechanic.
Please feel free to rewrite this answer as my english is far from perfect.
It would be also better to explain forces with schemas.