John Farley
It is unarguable (on boeing types) that in the situation that you described, rudder would be the control input that is required. After all, any roll observed during an engine failure is only a product of the yaw generated by the assymetric thrust condition. That is the theory.
The practice however is far different. We are not, or at least should not, be training guys to recognise and deal with sim failures. We should be giving them the tools they need to deal with a failure in the real case.
There are two major issues with recognition of engine failures and application of the correct technique. Firstly, as has been previously pointed out, if the incorrect rudder is applied, the aircraft will become unrecoverable in a matter of seconds. Secondly when operating on the line (i.e not in the state of heightened alertness for Eng failures as you are in the sim), the first indication of an engine failure airborne is usually the roll associated with the yaw. The problem being that in the real world, many other things cause roll as well, like gusty wind conditions, wake turbulence etc. As the AA crew discovered in New York, rudder is not the best course of action to deal with every wing drop. This, I believe, is why Boeing advocate initially aileron, followed closely by rudder. The natural reaction by any transport category pilot when noticing a roll, should be to correct initially aileron. When it becomes apparent (as it will almost immediately) that the roll has been caused by an engine failure, then apply the appropriate rudder to stop the yaw. Stopping the yaw is the primary objective. Not necessarily to level the yoke, it just so happens that a level yoke is a big, easy to use, and obvious indicator of when you have got approximately the correct amount of rudder.
I hope i have got this almost right as i am giving a brief on it soon!