yes, and that is consistent with what @Rolf said.
It initially tries to centre both rotors (ie. both pointing straight up). To do that, it needs to work out what PWM it needs to put in to achieve the centering. If you look at the nice diagram from Rolf, you'll see that as Q_TILT_YAW_ANGLE increases that the PWM needed to make the servo vertical moves more towards SERNOn_MIN.
The way the maths works is this:
* for an example, lets assume Q_TILT_YAW_ANGLE is 15
* it knows that at SERVOn_MAX that the servo is at -15 degrees (15 degrees angled back).
* it knows the total angle the servo goes through it 90+15 which is 105 degrees
* so to get it straight up, it needs to move the servo to SERVOn_MAX-(15/105)*(SERVOn_MAX-SERVOn_MIN)
As you can see, if you increase Q_TILT_YAW_ANGLE without actually changing the servo movement then it will cause the rotors to tilt fwd more.
In other words, Q_TILT_YAW_ANGLE is something you measure about the frame, not a desired angle. If you want to change the desired angle you need to move the servo endpoints.