If the AA policy causes the company to lose business or competent staff, then it won't be the proponents of AA who suffer. As the proponents of AA are not accountable as a company's management (and shareholders) for the consequences of their policies, it is unwise to substitute their preferences for those of the hiring staff.
One my favorite misconceptions about Affirmative Action. You do have to have to be comptent to benefit from AA. Any other system is a quota.
Candidates for positions in a company will always be a mixed bag, and different HR departments will have a different idea what qualifications are most important for the role that the candidates are competing for.
The HR departments will have a better idea what qualities are desired and what are not, and in what proportions, than someone who is pressuring the company to pass an AA policy.
It may very well be that the culture of a company is such that hiring any woman would disrupt the group dynamic. One example of this would be hiring for Chippendale dancers, although I'm sure there may be others. It's better to let the company's management, rather than an unaccountable 3rd party, decide which candidates could best contribute to their bottom line.
Economics isn't the same thing as engineering.
And the fact that single female economics professors w/ PhD's earn slightly more than their male counterparts does support my argument that if you hold various other factors constant, the pay differential between sexes vanishes. This evidence doesn't conclusively prove my argument, but it certainly supports it by giving one instance where the observable consequences of the argument occur.
I didn't attempt to try and deductively prove my argument with my example, but merely gave an example that supports it.
My bad on the economics/egineering. Why would you give data to prove a point that can't be deductively proven?
Even though my assertion can't be deductively proven, most assertions cannot be deductively proven, so that point doesn't weaken my assertion in particular. Most assertions are given as knowledgeable due to the preponderance of evidence that supports the assertion, not because they are deduced from relationships to other assertions.
For example, if I came across a man holding a bloody knife and standing next to a dead woman lying on the ground with a bloody knife wound in her throat, that observation doesn't deductively prove that the man killed the woman, but it would strongly support an assertion that he did.
This is clearly not the same type of reasoning that characterizes the argument "All dogs are mammals. Since Spot is a dog, Spot is a mammal."