I agree that it is an oversimplification but that's the basics.

As far as the gear multiplication, that's why you use the gear that is closest to 1:1 on a chassis dyno. The dyno can monitor plenty of outside variable such as ambient air temp, humidity, etc as well as car related variables like exhaust gases, engine RPM, wheel speed, etc but with all this the only true measurable factors it has to give power readings is the twisting force on the rollers and the engine RPM.