Isn't the other way around... the faster the roll-off, the greater the group delay?[\quote]
Oops...
Yes Dave, you are right, I am wrong!!!( again!!!)

I must have had a temporary mindstorm.
To restate the definition, for myself and any other people confused out there:
Group Delay is proportional to the phase response of the box, driver and crossover combination.
Or more simply, as Q gets higher, (sharper roll off), phase change between frequencies increases, leading to a greater Group Delay.
Sorry folks