Ok, scientifically this is the correct procedure.
(Is normalisation always necessary and if so how do I do that?)
But from a practical view, if the 24/96 file sounds better than the 16/44.1, I do not need to care to much why this is the case (because of the digital format or different handling of the file by my DAC), right?
And if I hear a difference (better?) at 24/96 and both formats are sonically equal, it would make sense to up convert all my 16/44.1 files to 24/96 to take advantage of the different (better?) handling of these files, right?