I've talked to one expert about a similar result in a company where I worked, and the flip side to results like this is sometimes promotion velocity - women looked "overpaid" for their level relative to the men, and that was often because they had been at that level much longer than most men at the same level.
If you set initial salaries fairly with respect to gender but promote men more quickly, then you end up with a company where it looks like women are paid more when you control for job title.
So would you say the accurate way to study this is to examine "lifetime pay"--look at the pay of demographics over a time period at the company? That would show if men are making more in 5 years of accumulated employment vs women, right, and potentially unearth some of this "promotion velocity" numerically?
That is one part of the picture. This situation is so complex that I don't believe everybody has actually figured out all the important parts to study in the first place.
Step 1: Let's agree that this is a complicated problem, and that quick feel good knee jerk reactions may end up having negative impacts.
Step 2: Work on solutions to complicated problems.
Easier said than done, but few things worth doing are simple, and there's already a lot of people doing great work on sustainable, informed solutions (but too many people just going "lets pay everyone the same, its the only way to be fair!!!" hurting everyone else)
> women looked "overpaid" for their level relative to the men, and that was often because they had been at that level much longer than most men at the same level.
Usually it's the opposite but for the same reason: women seem underpaid but when you correct for experience on the job, they make something like 98 cents on the dollar compared to men.
A little late on this thread, but did want to add - I'm not assuming that's what's happening here, but I know this has been overlooked in similar analyses. I think Google's hr team is probably smart enough to account for things like this, but I would really like for the report to say what other variables they controlled for in order to conclude that women are overpaid.
And I think someplace like Hacker news is exactly the right place for the community to discuss the many ways a company can get to a result like this - via Simpson's paradox, via differences in promotion velocity, or via simple economics, based on the fact diverse teams are desirable and women are more scarce.
Attacking someone for suggesting one not-yet-discussed possibility is not helpful to the discussion, imo.
Why do you feel like you were attacked? I think that's an ongoing issue today where people do not like to be challenged.
My observation is that news about men being disadvantaged comes out and people don't want to believe it.
I've seen people insinuate that the study was done by men and therefore the outcome would naturally happen. Another mocking that is was "scientific" and "numbers-based" Another claiming the women deserve more money. I bet if I challenged any of them they would claim to be attacked too.
If you set initial salaries fairly with respect to gender but promote men more quickly, then you end up with a company where it looks like women are paid more when you control for job title.