More confidence intervals

[simonster]

At the moment we only have confidence intervals for the t-test and binomial test. We should also have confidence intervals for:

• [x] Sign test
• [x] Fisher's exact test
• [x] Signed rank test
• [ ] Mann-Whitney U test

[BigCrunsh]

I am thinking about a confidence interval for the sign test. Currently, there are actually two sign tests, one for testing the median of a distribution and a paired one that tests the hypothesis that the difference median is zero. In the first case, I would expect something like

x = [-4.7,3.7,22.4,13.6,8.7,9.1,-7.8,10.8,15.6,23.5,14.4,20.2,6.5,10.1,-6.9]
ci(SignTest(x, 0.0), 0.05)
>> (3.7, 15.6) # confidence of 96.5%

But what output do you expect for the paired test?

[simonster]

For the paired test, I think you'd give a confidence interval for the median of the difference.

[bdeonovic]

Just want to point out we also have confidence intervals for multinomial proportions (PowerDivergenceTest)

[rawls238]

happy to take this on unless it's already been completed

[rawls238]

Signed rank test is done now, Mann Whitney U test is still uncompleted though