Anna Varvak

58 reputation

	58 reputation	bio	website		visits	member for	1 year
4 badges		location			seen	May 16 '12 at 20:42

10 Actions

May 16	asked	Need someone's endorsement to publish a pre-print on arXiv
May 9	awarded	Scholar
May 9	accepted	Recommendation for peer-reviewed open-source journal?
May 9	comment	Recommendation for peer-reviewed open-source journal? Yes, it's not just inverting a bootstrap confidence interval, which actually has terrible validity for small samples. Null hypothesis: mu=mu_0. I take the original sample (size n) and reflect it around mu_0, getting a symmetric distribution with 2n measurements; then I take bootstrap samples of size n from that symmetric distribution.
May 9	awarded	Nice Question
May 9	comment	Recommendation for peer-reviewed open-source journal? Thanks, this is really helpful!
May 9	comment	Recommendation for peer-reviewed open-source journal? Typically, permutation methods for testing hypotheses rely on two simultaneous assumptions for testing the statistical significance of the sample evidence: (a) the null hypothesis is true, and (b) the sample is representative of the population. This obviously doesn't work for testing hypotheses about one mean. I introduce Mirror Bootstrap method which assumes that the samples individuals are as representative as they could be with the null hypothesis being true, without assuming more extreme individuals than observed. I've tested validity and power, and it works surprisingly well.
May 9	awarded	Supporter
May 9	awarded	Student
May 9	asked	Recommendation for peer-reviewed open-source journal?