Honest question: in what way? As in it might make sense to have an omni-directional antenna for the transmitter but a directional antenna for the receiver rather than omni-directional for both?
Easy answer is, for a fixed load (I.e. a resistor) and frequency, one varies amplitude with power. (perhaps absolute value of area under the curve might be a better description)
For more information see here: http://hyperphysics.phy-astr.gsu.edu/hbase/Waves/powstr.html This is for a physical string, which I think is easier to wrap your head around. The math is very similar for electrical waves.