Rees:
Certainly a lot of times the decision is driven by the scorecard and by the other holes you've already found. If you've already got a couple of good short par-4's in that nine holes, and you don't have two par-3's, the decision is pretty obvious. Other times it might be driven more by the green site: if it's big and open, you'd tend to go with the long 3, if it's a tight spot, maybe the short 4.
Sometimes the client will have an opinion, too. Pat Mucci loves to cite the 18th at Sebonack, where Mr. Pascucci preferred a par-5 to a par-4; that sort of discussion happens on many projects, probably as often as not. If you're in Asia, whatever makes the course a par-72 is what the client will ask for.
One I wrestled with recently was the routing for our new project in New Zealand, Tara-Iti. The 7th hole there comes back underneath the clubhouse, and in the original layout it was a 200-yard par-3, though there was some room to push the green site back to 250-275 yards. [The tee site was fixed by some dunes.]
After a couple of versions of the back nine routing, we settled on a plan with three short holes on the back [10th, 15th, 17th], so the 7th was going to make five par-3 holes, unless we went for the very short par-4. The green site further back connected quite well to the 8th tee, so we decided to give it a go.
The hole will be only 260-270 yards from the normal tee, though we did find a small tee hanging out off the dunes so the big hitters can play it at around 300 yards. The green is quite small with a bit of apron at the right front, but it will take a great shot to be putting for a two. The hole was just seeded last week and I will be excited to get back and play it next winter.