The following are my observations only, featuring my opinion, and I don't claim to be offering any expert advice or information.
Without getting into a lot of details, the way that the different manufacturers set their compression levels is a closely guarded trade secret. They will not tell you exactly when and where things like oversampling or subsampling kicks in on compression. It is not important that you understand how this works, it is only important that you have a set of eyes. You can zoom in to see what (if any) compression artifacts are visible, and choose the level you feel most comfortable with.
Compression generally is different even between the image sizes of your own camera... for example, the software MAY compress Large-Basic and Medium-Basic differently.
I generally shoot Large-Basic on my D80's and Large-Fine on my D40, because the D80's Large-Basic is virtually impossible to tell from Large-Fine on even a 24x20 print, because the Large-Basic really isn't very basic. On the D40, there is a HUGE difference... obviously, the subsampling settings are quite different between the two image processor.
I don't have a clue on Canons, and I don't use photoshop for compressing JPEGs, so I don't know where the quality really starts to degrade... but there is definitely a point that it does.