seeing the compression ratios/file size after unarchiving the samples is a bit too much of a giveaway, and could lead to biases, but otherwise the test was pretty good, imo. You can differentiate between cd quality and anything encoded in a lossy format (even at 320 kbps) by inspecting the waveforms with basic audio editing software, but I don't think people would actually go to the trouble of doing that. I hope, that is.
If you were to run the test again, I'd recommend using something that's densely layered with many distinct sounding instruments that stand apart in the mix, and with little or no guitar distortion - basically, something with many fine details .Katatonia - tgcd has great production, but due to the heavy distortion and relative simplicity, the audio quality suffers much less from compression than something with more detail, which requires a greater amount of space/information to be expressed accurately.