gavinbirss, that's not a test of the MP3 encoding at all. That's a test of the mp3 encoding AND the decoding to .wav AND the frequency analysis of Cooledit. Too many variables to prove anything, even if you trust all the components. Have someone help you perform a double blind test, playing back both CDs and MP3s from the digital output of a sound card into an external DAC like the ART D/IO or the USB stereolink - but even then you haven't eliminated the variable of two different players being used.
Assuming Cooledit is showing you correct results, you will probably find if you read the documentation for your MP3 encoder that the rolloff above 16kHz is intentional. It is discarded by default because most people can't hear those frequencies at all. It can be turned off and you will then see much the same performance as Ogg, if not better - but will you hear the difference? Have you used a signal generator recently to see if you can still hear 16kHz?
LAME is also free, and can be played back with any number of portable and non-portable devices, which Ogg can not. There is no portable Ogg player that can hold the 3000 tracks I have loaded over the past two days onto my PJB, for example. Unless of course you mean free in a religious and ideological sense, in which case you will be perfectly willing to use any format as long as it is "free" and disregard the performance entirely - but please save the political advocacy for Slashdot.
I don't think that anyone disagrees that MP3 is not as good as the original - that's what "lossy compression" means. Ogg and wma and mp3 are alike in that regard.