Yet we don't know what jobs women and men do at Google. There can be bias elsewhere in the system. For example women might be disproportionately hired as administrative assistants and men might be disproportionately hired as senior engineers. Both groups could then be compensated fairly and it would result in men receiving a disproportionately higher percentage of the money.
You've missed the point entirely. The comparisons are among peers in the same jobs. But it sounds like Google can't get its own data together in a sensible way, so you can't really trust any numbers in this article.
I think we're moving the goalposts here: are we talking an abstract man, on average, in general earning more than an abstract, average women or we're talking about a different pay for the same work?
My understanding is that we're talking about the latter.
Does "same work" also include the hours worked? Usually not. If someone completes x projects in y hours, and someone else completes the same x projects in z hours, are they the same? Both did "the same work," but one was more efficient and did it faster (presuming the quality of the work by both parties was equal).
This is a good hint, but it's not enough information. Specifically, the jobs where the pay imbalance is likely greatest also would also likely have a more skewed gender ratio. But we can't know, because the article doesn't say, and just trusts Google's word on everything.
