Lee,
Yes, sampling is dependant upon two coordinates - the amplitude (y) and the positioning (x). Amplitude error clearly manifests itself as amplitude error (duh). Timing error, however, is more complicated.
If the samples are taken at varying intervals of time - say there is error in the sampling mechanism - then the resultant samples will be interpreted as having happened at the correct times, and the data will travel through the rest of the system like this. The result is that bad amplitude data will be captured, and this data will be like any other amplitude data error. In other words, bad timing data during the capture process manifests itself as bad amplitude data. This is because bad timing data causes bad amplitude to be captured.
OK, now on the output side - this doesn't cause bad amplitude data, per se, but it causes the same results. Jitter on the output means that the samples are turned into analog voltages at the wrong times, thus recreating an erroneous version of the waveform. This erroneous version, however, is of the same variety that happens when bad timing is used on the input - the waveform that is represented by the samples is erroneous in the same way. The timing error on the input causes the waveform to be misshapen - stretched or compressed - representing bogus frequencies (that's the noise part) and the more error there is the more amplitude those other frequencies have. The same occurrs on the output - the waveform - by means of being converted at the wrong times - ends up compressed or stretched in the exact same way. In other words, the manifestation of jitter on the input or the output both produces a result that is akin to simple amplitude error of a certain variety.
Am I making sense? Ask if you have more specific questions.
Nika