Hi everyone, I’m using an Anderson-Darling test to check for normality and am unsure about the results. I am expecting to reject at the .01 level with a p-value around .008. The pvalue() call on the OneSampleADTest returns .323. When looking at the test summary the test statistic is 1.06, expected.

I’m new to Julia and am thinking I am causing the problem. Any reason the returned pvalue is different than what Agostino and Stephen’s would give?

I did try this. I also double checked off excel, python and JMP. The test statistic is the same across all bu the returned p-value is different, in this case not significant, in Julia vs the others.

Pardon my formatting, still learning:
using CSV
using DataFrames
using Statistics
using StatsPlots
using Distributions
using HypothesisTests
using Pingouin

data = [6.67, -0.69, -2.03, -2.6, 2.19, -1.17, 9.93, -2.68, 10.41, -1.34, -0.16, -0.26, 1.67, -4.04, 6.46, 2.94, 7.32, -2.49, -1.26, -0.17, 7.06] mean_samp = mean(data) std_samp = std(data) d = Normal(mean_samp, std_samp) tested = OneSampleADTest(data, d) tested pvalue(tested)
Returns the right AD value but the pvalue returned is .34. Should be .008.