I'd say take one image and try it both ways and print each.
If you can find a difference for the good, that's the tract to take. Otherwise, it may be simply a toss-up as to which method works for enlargement.
However, if you are simply going small, always work on the native res of the file as your largest image. Work it once and be done with it. You can always shrink the image down to a single pixel quite well, but enlarging an image you put sweat into at 4x6 and then realize you need much larger, you need to start at square one.
When it comes to shrinking a master image down, the image looks just fine using bicubic sharper as the method to throw away your pixels.
For large images, in the past few years, I've tried opening an image at the max in ACR and later taking my image and enlarging it in PS. Sometimes I've had to use both methods to get the needed size.
Between enlarging using Bicubic in PS and going for the max in ACR, it's hard to tell which one is better, so I don't worry about it too much. Both seem to look just fine when the ink hits the paper.
In any case, the last step of any of these processes is to sharpen your output print file as the last step and save the file with a new name. Your master files should remain unsharpened and with all your layers intact so that you don't skew the pixels any more than you need to.