Hacker News Reader

Comment by vacri | Hacker News Reader

It's very hard to make conclusions for the whole company from such a small sample.

Statistically speaking, an n of 100 is a reasonably decent sample of 5000 data points. It seems to me that you're looking for excuses, as if this concept of 'dry-run' meetings only happens to be within the author's "hundred-person bubble".

View on HN

My experiences at Apple are vastly different than the author's; As a developer I've never had such a low-stress position. My hours will sometimes drift over the 40-43 mark but not often.

as noted elsewhere, apple actually employs over 90,000 people. it's less significant in that scope.

Also, while I have no doubt his experience sucked, he was also in customer service as opposed to being a developer/engineer. I suspect those areas differ somewhat. (Not suggesting it's ok at all, just that it may not be completely representative of the entire company)

Actually, the size of the population as a whole is nearly irrelevant when considering the significance of a particular sample size. 100 people has just about the same significance for a population of 1000 as it does for a population of 1 billion.

That said, since the people he met certainly weren't randomly selected, it'll be a heavily biased sample.

My point is that the parent considered a sample of 100 to be too small for a population of 5000. Statistically speaking, it's a decent sample size.

To mikeash and vacri,

A sample size of 100 is equally meaningless for 500 or 5 billion if the sample is skewed--which it definitely is in this case.

Reply on news.ycombinator.com