[simias]

If you keep picking the 128kbps track it does mean that you can hear a difference, and maybe that you subjectively prefer the more compressed audio.

That's why usually good audio tests use ABX where it's not about seeing if you can identify what is best, but rather if you can identify a difference at all. See for instance: http://abx.digitalfeed.net/

EDIT: I actually just tried the test and... ended up doing the same thing you did. On the Neil Young track I could hear a definite difference in the high pitched violin-like background, but I ended up selecting the 128kbps track as the "highest quality", maybe because the compression made it sound smoother.