Go ahead, and measure the levels. They're 1:1 within 0.2db for the first example (unless the Youtube codec messed that up). I get why you would THINK that the levels MUST be different as the loudness change is CRAZY...they're not though.
Here's a screengrab from the footage we used for the first (full mix) example (I disabled the crop and transform in FCPX, this is the actual video that makes that part of the trailer):

Note the 0.2db difference visible at the level meters on the lower left. Kinda not so much for the amount of change audible in my book...
