.
At higher frequencies, the voltage at the speaker terminals does indeed drop by as much as a dB for "normal" wire runs due to the combined effects of resistance and reactance. But at above 15KHz, the minimum detectable change for humans with good hearing is somewhere around a couple of dB. So I don't believe anyone can hear the difference with "normal" wire runs.
I guess another question is just how long is that cable they are testing?
So they could concievably get that graph if the wire is long enough, most likely well betond anything someone would use in a house.
