11 Nov
2014
11 Nov
'14
8:13 a.m.
It should be mentioned that p-levels are not really probabilities. --Dan On Nov 11, 2014, at 6:48 AM, Seb Perez-D <sbprzd+mathfun@gmail.com> wrote:
As Brent said, statistical tests are usually not independent, and properly calculating the p-level of a combined set of tests is complicated.
That being said, if test i has a probability p_i of failing, and the tests are independent, isn't the probability that they all fail equal to the product of the p_i ? and the probability that at least one of them fails is 1-\prod (1-p_i) ?
And somewhat on-topic, an amusing demonstration of non-independent tests: https://macartan.shinyapps.io/fish/