I think you're right about there being selection bias... people who are happy with their package don't spend their time filling out these spreadsheets (I didn't!)
Also, I note that if I had taken just my initial offer, my TC would have been much closer to the listed. But I went through the long and arduous process of sending one email to my recruiter asking for more money and here I am.
I guess I'm technically a "person of color" (Hispanic) but I have a ton of useful, valuable experience and so far as code monkeys go I'm pretty good at negotiation. I thought Microsoft was a horrible pit of vipers and a generally terrible company, but I definitely did NOT experience any racism.
Maybe the people who responded to the survey are generally just really bad at negotiation, and the HR people should've gone easier on them.. I don't think it's feasible to get a 50% pay bump without changing jobs, so they're all in a really awful position
I got unlucky and ended up on a particularly bad team, so I left before I could get any firsthand experience with it. It's weird if it works that way, though, since it's so easy to jump ship to another company and get way more money.. Microsoft's interview process is as hard as pretty much any other high end company (except maybe Google, which seems to expect you to grind on leetcode a lot more..)
Seeing that in 2020 reminded me of this https://www.penny-arcade.com/comic/2002/07/22
> A majority — 194 out of 310 — total respondents said they were people of color. Also, 69 respondents reported feeling marginalized or at risk of being marginalized due to their gender or gender identity. In total, 22% of respondents said that their gender or gender identity had or might hinder them at work.
This indicates that it is not a representative sample of M$. Though which conclusions to make from it are much harder to say.