Since I did not make the measurements, and do not know the actual procedure, the most likely situation is there is a DAC, which has its jitter measured. If not, I do not see any way a single TX component can generate that much.
While I do understand why these sort of measurements are made at the audio output, as that is the only metric that matters, the way they are done is prone to lots of confusion. Most involve reading the sidebands, of some signal.
First, you have to know not only the numbers (like x pSec, or -y dBc), but also the spectral content. Then, to add confusion, the frequency of the signal is part of the equation. A fixed amount of sidebands will have lower jitter numbers, as the signal frequency increases. If one guy uses 11 kHz, and the next 16 kHz, trying to make a direct comparison is not possible, without knowing the fudge factor, to put both on equal footing.
So, even if you know all of this, does this still mean it makes any sense? The folks who design this stuff don't agree on what is more important than other matters. We have our believe that the sideband level has to be "so low", and that sidebands at "this frequency" are more harmful than those at "that frequency". But, that is just an opinion, and we all know how useful that is. There is not an absolute metric, cast in stone, that must be obeyed, and all else is crap. Nope, doesn't work that way.
Designers I know argue this stuff, back and forth, all day long. If we can not agree, how can the layman/customer? It is simple: you can't. Just take those nice plots, in the magazines, and take them for entertainment value. Trying to make sense of them is not going to happen. As a designer, we find value in them, but not for the obvious reason(s).
Steve has the right idea: buy something, and try it. Most folk have some sort of trial period. If it sounds good, buy it. If not, try the next guy. Leave the worrying to the professionals. We like have something to argue about.
We don't even agree on how to measure it, and what kind of equipment it takes.