70 people polled out of a population of 4700 would have a 12% margin of error at a 95% confidence level, if it was a random sample. 500 people polled lowers that to about 4%.

If Blizzard were a lot bigger--say 100k people, a random poll of 70 would still be 12% and a random poll of 500 would still be 4%. You can play around with the numbers here [1].

The margin of error on a poll, if the population is large compared to the sample size, depends mostly only on the absolute sample size, not on how big a fraction of the population is sampled.

The flaws in this poll are how the sample was selected, not the size of the sample. It's not a random sample.

[1] https://www.surveymonkey.com/mp/margin-of-error-calculator/

> if it was a random sample

And here lies the problem. How do we know that those people were not ahem, "selected"?

Well they almost certainly weren't selected. But there might be latent correlational factors that cause HK views to correlate with blind-membership.

They would still need to be willing to answer the survey. I have ignored plenty of Blind polls/surveys in my time there.

1% is not that low, a typical poll has much less than that. E.g. a Gallup poll generally interviews about 1000 individuals to represent the whole of the US.

Selection bias is a much larger problem here.

Because it wouldn't be economical to hire hundreds of thousands of call takers to reach everyone in the US in a single week.

There's no reason a company could not send out a one-question survey link to all of it's employees with one click if they wanted to get reliable data.

An important point - at least Gallup publishes some details on its methodologies.

How do they know these weren't 499 shills claiming to be employees?

Blind requires verification with a work email.

Which then introduces some doubt that this is truly anonymous. In other words, if the respondents feel that there's a chance that their answers could be linked back to themselves as employees, they're probably more likely to toe the company line.

> Blind requires verification with a work email.

Thankfully corporations are unable to control their own email addresses, right?

Huh? Is your suggestion that these are not employees, but "people paid by a company to perform work for them"?

I suppose the FTE/contractor distinction is relevant here, but otherwise I don't get your point

I think the suggestion is that they may not be individual employees stating their personal opinion, but accounts created by even just one employee (who can create multiple accounts) under company direction.