To GGAPBE96,
1. When Network Analyzer measures the S-parameters it actually measures the voltages. Because imaginary part parameters depend on trigonometry there may be very steep slopes when small changes in argument provide big changes in function result. As close you measure to component’s SRF area as steeper will be the slope of reactance and bigger measuring inaccuracy is possible. Now, suppose you are measuring S-parameters of Murata 0402 inductor on some frequency F. Or you may obtain corresponding S-parameters file from Murata web site, which is preferable because measuring and then de-embed S-parameters is not a trivial task. Now, when you need to simulate resonant circuit with your inductor of interest you need to find suitable capacitor desirable (but not absolutely mandatory from the same manufacturer). In this case both S-parameter files were measured in the same conditions and by the same technology, therefore you may expect roughly equal inaccuracy in measuring process. When you combine these two S-parameters files for L and C in one simulation of resonant (it is very important) circuit, reactance slopes will be equal, but of opposit sign and mutually compensates. Accordingly you may expect a good compensation for measuring inaccuracy of L and C components S-parameters. Finding high Q capacitor usually is not a problem. If you have any difficulties to find a cap, you may chose Q-factor in the simulation program and usually 1000 is a good number for small caps below 10 pF. Practically this method allows determining Q-factors at high frequencies with good accuracy, but if you need to do it for low frequencies, it is probably easier to use X over r ratio.
2. Capacitor as a circuit element do not absorb inaccuracy, the latter just compensated without any physical participation of capacitor. There is much more trigonometry than physics.
Best regards,
RF-OM