Two things that I have seen lead me to believe that bias is a big problem here. First very few users get significant experience with one high end package let alone two or more packages of a similar vintage. Second the priorety level for ground plane,trace length equalization component position criticality and other rules is extreamly design dependant. I have concluded that the in house requirements will determine the package choice.

Then the operators experience level and understanding of the design engineering teams specific requirements will control the time required to achieve a sucessfull board.

As with most evolving fields the demands placed on this area have caused the software to evolve into a separate discipline.
Finaly after a selection is made your company develops a huge investment in the chosen product(einished designs, custom librarys ect.all of which must be supported or utilized to justify the investment). That investment prevents casual migration to a diferent package. Terry