Kruskal-Wallis test is a popular nonparametric ANOVA method ,
and its corresponding post hoc analysis or multiple comparison methods are rarely known by part of scientific researchers,
now we recommend you try the Steel-Dwass test.
The pre-requisite for statistical analysis is very important and easy to be ignored, such as whether the data is normal distribution, whether the variance is homogeneous in multiple sets of data, etc., if your data meet the above conditions, parameter test will be OK, otherwise, use nonparametric test.
For the parameter test case, Tukey’s HSD test is common for post hoc test after 1-way ANOVA ; correspondingly, nonparametric test method of Steel Dwass test is a good choice for post hoc test after Kruskal wallis test.
BTW, many high-IF SCIs’ editors seem to recommend Steel Dwass test for non-parametric Kruskal wallis post hoc test, rather than others (e.g., Tamhane’s T2 test).
Steel-Dwass test, a multiple comparison method after Kruskal-Wallis test.
R codes are as follows:
# Steel-Dwass test
| t | p
1:2 | 2.680234 | 0.036960431
1:3 | 2.539997 | 0.053980573
1:4 | 1.282642 | 0.574011771
2:3 | 3.746076 | 0.001031145
2:4 | 2.046776 | 0.170965537
3:4 | 3.384456 | 0.003976894
① This article represents the author’s personal point of view, does not represent any other position!
② Feel free to cite or share this article but please specify：The original URL of this article is http://meiweiping.cn