1. Have two or more controllers.
2. Bind the controllers to the same axis actions. I used helicopter cyclic for my testing.
3. Get in a helicopter and test the axes. Notice that the sensitivity is lower than with one joystick, showing the axes are being averaged, as intended.
4. Disable all but one joystick and repeat the test. No change to sensitivity.
5. Unbind all but one joystick and repeat the test. Full sensitivity is restored.
Repeat the whole process, but instead of disabling the other joysticks, disconnect them without unbinding their axes. Notice the low sensitivity until the non-present controllers are totally unbound from the axes.