Please take a look at hp application note 77-1. Transistor Parameter Measurements. Info in the note still applicable today as well to passive devices. Yes, you need to take care that the device input power CH0 is reduced from the -9 to -12 dBm power level. Say, -20 to -30 dBm. As well, the output power set by the DUT gain is not excessive as delivered to CH1.
In AN 77-1, the grand daddy of the vna, the vector voltmeter is discussed, however, test fixtures and the application of the bias tee network is presented. Note 2- directional couplers are featured on the cover. The vna you use has in effect only one of those couplers (a resistor bridge). Hence you need to turn the device around to obtain the remaining s data. Or, see another post where another user has built a switch function for doing this task.
Also, see LT Spice simulator which is free. LT handles active devices with models and provides S parameter output data. So you can run a simple 2N3904 at 30 MHz or so, where package parasitic of the transistor is not so critical and see the expected results.
There is quite a bit of information to digest before you can expect to get measured results to agree with calculation. However, the eureka moment is sure worth the effort!
FUN.
