Hacker News new | past | comments | ask | show | jobs | submit login

N=104 is a fine sample size for some things - especially given how much comfort you have for error.

This isn't a great study in general, because, realistically, there isn't a large range for the work hours. And the majority of them were grouped into one bucket => 4-8 hours.

That's the biggest problem.

The second biggest is that - presumably - not all of the respondents work 5 days per week.

The third is that - without a large sample size - your margin of error is going to cover a pretty big chunk. If your finding is 6 hours +/- 1 hour, 95% CI - that's pretty unsurprising.

You need a large sample size because there are ~115M workers in the US. What are the odds these 104 people are a decent representation of all ~115M voters??

If there's only 10,000 people this applies to that read HackerNews, then - if you fixed everything else - your CI wouldn't be super low.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: