Perhaps this discussion of acceleration versus top speed should be split off...
Regarding top speed versus acceleration, I think it's important to remember what (I believe it was BJC) said in a whitepaper: the goal of a drive train is to get you where you need to be, when you need to get there. To that end, I'd focus on time to go a given distance: a very nice calculator for this can be found
here, which was written up by Andrew Schreiber (based on the drive train acceleration model, which Ether provides a link to in the next post). It's a little bit out of date but it's good for giving reasonably accurate numbers.
Other than that, it's all about team preference. Obviously there's some math behind it (if it's going to take you longer than 25 feet or so to reach your top speed, it probably isn't worth it), but nothing that hasn't already been mentioned elsewhere.