Watching Jeff and Michael in "from Camera to Print", and reading "Real World Camera Raw CS3" (nice job that, Jeff), I think they've made a good case not to resample when sizing an image for printing. Let it stay at "native" resolution while resizing and let the final output resolution fall where it may; choose the nearest resolution and intended surface gloss when selecting PhotoKit output sharpening; and print.
Much of my printing is done via ImagePrint. My understanding (maybe this is a re-RTFM moment) is that ImagePrint up- or down-rezzes to 360ppi for output.
So if resampling is going to take place, would it be better to do so in Photoshop using whichever version of Bicubic is appropriate to the direction one is re-rezzing; or let IP do it on its own?
Forgive me if I've missed the obvious somewhere...