With regard to your formula, as Paul has indicated it doesn't really apply to R2. R2 has essentially no DC connection to ground at the tube end, so there is essentially no path for signal current to flow to ground. Without a path for current to flow through it, there will be essentially no Voltage drop across it, hence Ohm's law really cannot be applied to it. Which means that the power corollary is also pretty meaningless.
Personally, I have always used carbon comp for this application as it should theoretically have a lesser chance of being inductive. OTOH, I tend to use a higher resistive value on low-level amplifier inputs (line level and lower) on the theory that a higher resistance might have more of a damping effect on potential parasitic oscillations. I have used as high as 10,000 Ohms, in fact, although I can certainly understand the argument for using lower values.
From the preceding discussion, though, you will see that the resistance value cannot have an appreciable effect on the signal level reaching the grid.
It would certainly seem possible to verify this without too much trouble. A constant level of signal, such as a signal generator or a CD with audio tones recorded on it would be necessary. Using a good AC Voltmeter or a scope would be ideal for measuring Voltages in this case, but a DMM could be used on the AC Voltage range, providing you use a test frequency within its capabilities. In a pinch, 60 Hz should suffice.
With regard to your new inquiry, I believe you are speaking of "R1" again, when referring to "grid resistor" or "grid leak". The value of R2, the "grid-stopper" resistor is considerably more arbitrary.
