He did not say they were lowering the bar, but that by rejecting (proportionally) a greater number of qualified male candidates than qualified female candidates, the bar is effectively lowered.

If what he says is true, that there is a higher false negative rate for men, it's hard to imagine a system where the bar isn't effectively lowered.

The one possibility I saw argued elsewhere is that you could take all qualified men, and randomly reject some of them. At that point, you would expect the bar to be level.

If however you rejected qualified men in a non-random way, which is more plausible, the effect would be to change the bar.

I hadn't ever really thought about this kind of selection effect on the statistics of populations, so would love to hear if this sounds wrong or what the real expected outcome should be.