I've also noticed that they are not consistent with WHERE they measure jitter. For instance, in their review of the Sonos ZP80, the TOSlink output was used, rather than the coaxial SPDIF output. This seems strange, as TOSlink outputs are normally far higher in jitter than coax.

Hard to know the truth.

Main system: Cable, Apple TV, HD621, G68D, AC11, 218, DSP7200SE (L+R), DSP3200 (L+R Rear)
2nd system: Cable, Apple TV, MC200, HD621, 568.2, M3100 (L+R)