This is strange, I can only think non center mount or no mag mount is a kludge to cover impedance problems. Sounds like poor antenna design.
For best omnidirectional pattern, the center of the vehicle profile is the best, proven over decades and modeling.
And again, the magnet mount issue. I once thought, Hm, poor/no ground, that cant work well.
That is true talking about DC current flow, but this is RF, high frequency AC. The magnet base is grounded to the antenna, so you have like a capacitor, coupling to ground. You get above 20 MHz, it doesn't take but a number of picofarads to be like a short circuit.
Tuned circuit base loaded antennas will see a near ground at the magnet. Some well designed antennas evevn incorporate a capacitor to ground side of the coil.
A ground plane less than reasonant 1/4 wave makes the impedance inductive, the fix for this is a slightly longer driven elemet (antenna).
A good example is a CHP 40 MHZ antenna on a motocycle, practically no ground plane. The whip is longer than say a cruiser with the same setup.
Either mounting with mag mount, there would be so little difference, you need special equipment to see the difference.
I tested a Wilson 1000M on a mag mount, centered on the trunk lid of my old Cadillac, tuned with near no detectable reflected power 27.200 MHz. somewhat direcetional to the front of the car.
Moved the antenna to center of the roof, (This was vinyl covered BTW) and the difference was best SWR a 27.000 MHz. Just about omnidirectional. So.....
