One sample will sound a little 'thin', the other one 'punchier', 'more powerful', or 'with slightly more detail'.
In reality one sample has just been attenuated by 1.5 db (nothing else). Without that knowledge and time synched AB switching capability (like in Foobar's ABX) it is almost impossible to tell, wether there is really 'thinner' sound or just a difference of absolute volume.
To exclude possible PCM truncation issues, the same shaped dither (UV22) has been applied to both samples.