I find it telling that the GoogleMemo author cites sources, while the initial rebuttal from Google's VP Diversity simply states somethign to the effect "The statements so incorrect its not even worth addressing".
That was exactly the kind of shouting down that the author was talking about.
So, here are some of the researchers that were referenced in the memo.
There's rather a lot of kneejerk reaction on both sides here.
conclusions are questionable
I agree he takes many of his statements way farther than the evidence (that I'm aware of) would support.
he seems sincere ... pitchforks appear to be on sale
... and that might just be my personal take away. If you are going to make bold statements, you better cite specific sources.
There are two reasons for that: when the blow-back comes, your case is less dismiss-able; also, the act of cross-referencing your statements to citations acts as a self-editing mechanism (regarding the accuracy vs boldness of your statements).
He takes things very far. It's clear that despite his apparent biology background he actually has little experience handling statistical evidence (for what it's worth, I've seen similar errors in engineers for decades - I think it's a category error that goes with the territory earned by training). But his critics are making similar and sometimes even worse errors. It's like watching the blind fight the blind.
You're right: his biggest mistake was in taking the obligation to cite too casually. Not only does it weaken his argument and open him to criticism but it also lures him into thinking he understands things better than he does.
But Google better hope he can't persuade a court that he thought he had legitimate labour grievances shared by colleagues. At will dismissal is one thing, but union busting is quite another.
As for statistics - frankly I would trust an engineer more than a social scientist. Sorry Kevyn - maybe you're an exception. I am far from an expert in statistics yet I have seen so many obvious errors in statistics (errors in basic math, really) by well-regarded researchers in the social sciences that it is hard for me to take the social sciences seriously. I take those who study things like the physical structure of the brain more seriously because, as near I can tell, they are far more rigorous - and typically can do basic arithmetic.
In a room, there is a naked woman, a mathematician, and an engineer. The mathemetician says he can never reach the woman because he would have to cross 1/2 the room, then 1/2, etc. The engineer says something like, "I can get close enough for my purposes".
After reading +Kevyn Winkless's comment, it occurred to me that part of "The Art" of engineering is to make assessments and decisions on imperfect information. It can lead to imperfect, though practical, interpretations.
That is the gift of engineering, but it is also a sword that can cut both ways.
When it comes to actually analysing raw data or assessing somebody else's analysis? You're right: engineers' training os more likely to apply, and actually most modern social scientists have very limited stats training if any. (Yes, I may be an exception - I certainly was in grad school. I blame my microbiology and biochemistry training.) But social sciences (I mean actual social science - not literary criticism in mufti), biology, and medicine deal regularly with stochaistic logic and situations where A therefore B is not the pattern, where "scatterplot" phenomena interact in messy ways. I wouldn't say they're better equipped to get right answers, exactly - but less prone to making some basic category errors and misleading assumptions simply because they're more apt to be on the lookout. Some of the errors I see in the parts of the manifesto I read are exactly this kind of thing, mostly mistakes in understanding the relationship between population trends, trait prevalence, and non-random sample.