It's definitely not representative. For reasons that would be obvious if you knew the history of the document, but that i can't share, it consists almost entirely of a sample of people from certain mailing lists.

It's definitely not valid to draw any conclusions about google as a whole from it.

> from certain mailing lists

eng-misc?

I believe you. I also believe the 2nd chart would basically look the same if you polled 10,000 representative people from inside Google. The 1st one, not so much.

Maybe?

Remember though, that groups that are underrepresented may also be disproportionately affected.

IE women may feel the document is much more harmful than men, but by numbers, ...

You'd have to break it down by a lot of variables, and once that happens, in practice, everyone will just draw whatever conclusion they want anyway :)

Anyway, not gonna speculate on what it would look like, just gonna point out it's not valid data.

a) I'm not sure what you base your believe in. My gut feelings tells me the opposite. If you polled a representative sample of employees you'd get very different results in the 2nd chart

but that's not important because:

b) The idea that average people are good at understanding the consequences of such document is inherently flawed. It's my problem with a lot of aspects of American culture (I'm an European living in SV) - regular people are not in position to evaluate complex systems.

Every time I see an article about, say, a shooting, that ends with "Joe Black, who lives nearby, believes that police should do X and Y to stop the increase in crime", my blood boils. Joe Black is in a terrible position to understand the crime dynamic in his city, to understand how the police operates and what does and doesn't work in the pursuit of lowering crime rate.

Now, coming back to the topic, I'd argue that 95% of Googlers are completely not qualified to respond to the second question. At all.

There's a bunch of people in the HR department, with degree in sociology, group psychology, organization behavior and design, maybe people who studied micro-discrimination, gender imbalances etc. that are qualified to answer it. But the idea that you can poll employees to learn about whether this document is harmful is completely nuts.

Sorry for venting :)

p.s. But yeah... while writing this comment I saw an article about the poll asking Americans if the US should attack North Korea [0]. I honestly believe that the culture of doing that, while pretends to stay close to the roots of the country of freedom, where citizens decide about their country, is in fact terribly harmful because it builds conviction in people that they're in the right position to make such judgements on topics they have insanely little knowledge about.

[0] http://www.cnn.com/2017/08/08/politics/cnn-poll-north-korea-...

Voluntary response bias is definitely possible. People who feel the most wronged are more likely to respond to the survey.