I realized that I was unclear with how I presented the plot I posted above--not just with AA readers but with myself. I made this measurement a while ago and had not recently thought much about exactly what I'd done. I decided to revisit the issue. With a bit more confidence--not to say understanding--I compared Tidal files: the MQA file (decoded) and the CD-res file, with the MQA decoder turned on--but this of course is not an MQA file, so that shouldn't matter.
I repeated this several times. I'm confident of the result. There are two spectra in the chart below: MQA is red, CD-res is yellow. These are Tidal files delivered bit-perfect to the converter--the Mytek Brooklyn--via Roon.
The two spectra are identical, to within small fractions of a dB.
