Mathematics desk
< October 17	<< Sep \| Oct \| Nov >>	October 19 >

Welcome to the Wikipedia Mathematics Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

October 18

statistical significance with small n

If you want calculate the statistical significance of a distribution against the number of samples, say the confidence that a coin toss is biased after only ten tosses, what would the formula be? — kwami (talk) 23:14, 18 October 2012 (UTC)[reply]

See [1]. Bo Jacoby (talk) 00:09, 19 October 2012 (UTC).[reply]

Thanks! That answers my question. — kwami (talk) 00:24, 19 October 2012 (UTC)[reply]

But don't rely on Binomial distribution#Confidence intervals for small n. Bo Jacoby (talk) 17:18, 21 October 2012 (UTC).[reply]

Right. The main article it links to, Binomial proportion confidence interval, gives various ways of getting a confidence interval; in the section Normal approximation interval, for one class of confidence interval calculations, it says "A frequently cited rule of thumb is that the normal approximation works well as long as np > 5 and n(1 − p) > 5".

StuRat, I'm not sure what the point is in putting on resolved tags. How do you know whether someone else will have something relevant to say?Duoduoduo (talk) 19:57, 21 October 2012 (UTC)[reply]

Well, this does answer my question. If someone can improve the answer, great, but this is sufficient. We had a page move based on a GBook search with only 40, where according to the first answer above, the diff was only 1σ. I think we should make moves based on RSs, but oh well. Next time I'll be better prepared. — kwami (talk) 20:52, 21 October 2012 (UTC)[reply]

Resolved