OK. Let's say you have two options:

(a) encode using standard MP3 with, say, 128Kbit/s
(b) encode using MP3pro with, say, 224Kbit/s (or whatever produces the same file size as standard MP3), at the expense of losing compatbility with just about anything in this world.

(a) and (b) would take the same size, but according to the study you mention nobody would tell the difference.

If you choose a lower bitrate on (b), the size would be smaller, the quality (according to your study) would still play no role, but the compatibility problem would still be there. Easy solution: just lower the bitrate on (a).

Obviously, if bitrate actually makes a difference (and I suppose it does at some point, to *my* ears mp3 with 4kbps is sufficient , then you want to maximize the information a -- standard -- player can use, and that's something you can only achieve with a standard format, i.e. not mp3pro.