You miss my point which has nothing to do with warm fuzzies or political correctness. I'm not accepting as fact that there are innate intelligence differences,

You miss my point which has nothing to do with warm fuzzies or political correctness. I'm not accepting as fact that there are innate intelligence differences, but for the sake of argument let's postulate small differences between race/ethnic/gender populations. When you screen that way you throw away almost half your candidate pool for no good reason. You achieve worse results than if you use some other, not quite as lousy, method of culling. You'd probably have to work at it to find as weak a correlation. Sexism and racism aren't only evil, they're stupid too - even if the tiny intelligence differences that disputed authors claim to have found were real. So in what way are stereotypes "pretty accurate" in the context of hiring? You aren't hiring a population so small population differences aren't super useful - except perhaps at the tails of the distribution. If you attract just the cream of the crop, you might have an argument. That population is likely to be very skewed.