I guess there is a big difference between processed samples from a given source and a recorded sample. You get better results when convert 16 to 8 bit or direct 8 bit calculation. There are some variables that can make 8 bit records sound bad (mostly noise added). But it depends on source quality. So I don`t see much sense in measure comparing because at the end you hear what you hear and that differs from person to person.
If you record the analog Amiga output that plays a 16 bit sample via 14 bit calibration. What do you think you can measure or what result will expected?