A huge bulk of analysis relies on system linearity for validity. Without paying attention to the non-linear nature of your system, you will end of up comparing apples to oranges, so to speak. 

The best way to "linear-ize" your system is by normalizing your measurement method. Introducing a scaling factor after data has been collected is wrong. The scaling factor must be introduced at the time of measurement to measure the correct system behavior.

The most common example of this is the use of constant current drive conditions to calculate the mechanical Q-factor using power bandwidth calculations. And you guessed it...the Q factor calculation using bandwidth requires system linearity. 

More on this type of measurement in the next post. Stay tuned!