I don't understand for sure whether your answer means your observations have been coming from prints or from monitor images. If the former fine, if the latter, you may find it revealing to print the various tests you've made and re-assess your observations.
Yep, you have the critical element here. As far as my observations go, well I hope I've provided enough data for you to judge for yourself. What really matters is what you see on the print, as you quite rightly pointed out.
If you keep within decent dpi bounds, then there's little to nothing to choose between the different methods of sharpening.
I do believe, though, that ACR 4.1 offers an advantage if you need to stray beyond those bounds. I'm not, however, so foolish as to believe that my belief would be shared by all.
I just want to round out this thread by stressing that ACR 4.1 is
tha BOMB. This is
the reason to fork out for a CS3 upgrade if you're a photographer, it's that good (and if Adobe paid me to promote their products I'd have to spend a lot of time working out how to be more subtle without any 'look dude, decide for yourself' examples ).