Diego Javier Zea

Other thing about `OneSampleADTest`: The _p value_ is calculated according to _"D'Agostino, Goodness-of-Fit-Techniques, 1986 p.127"_, which are _p values_ for testing normality. That book has other tables, with the other...

I ran all the methods in this pull request, and I measured the differences against R's P values with the following function: ``` julia function P_test(N) data = DataFrame( ties...

I guest I'm getting Int overflow here... How can I solve it? The problem is in this line: https://github.com/diegozea/HypothesisTests.jl/blob/master/src/spearman.jl#L121 ``` julia julia> x Spearman's rank correlation test -------------------------------- Population details:...

This gives a better result, but maybe is not the best solution: ``` julia (((x.n+1)^2)/36) * (x.n^2) * (x.n-1) ```

Clever ideas @andreasnoack and @fiatflux ! Since `N > 0`, `sqrt((N-1)*N^2*(N+1)^2)` is just `sqrt(N-1)*N*(N+1)`. I will also use `float`, just in case it could avoid the integer overflow.

Thanks! I updated the pull to avoid Int overflow ;) ``` julia julia> x = rand(10153); julia> y = cos(π*x+0.5π); julia> corr = SpearmanCorrelationTest(x,y) Spearman's rank correlation test -------------------------------- Population...

When there is no ties, _S_ values are the same between R's `cor.test` and this `SpearmanCorrelationTest` and they are calculated in R using the equation (14.6.4) in Numerical Recipes. Here...

Travis shows an error with `FisherExactTest` unrelated to this pull request: Maybe tests for `t.ω` should use `@test_approx_eq_eps`... ``` ERROR: LoadError: LoadError: assertion failed: |t.ω - 7.3653226779913386| 8.881784197001252e-12 ... while...

@fiatflux I updated the code, deleting paddings and extra spaces. I also setted a seed for each random test ;)

@fiatflux I added more tests to improve test coverage and I fixed bugs that I found in the process.