A SAS macro for testing differences among three or more independent groups using Kruskal-Wallis and Nemenyi tests

Abstract

As a nonparametric method, the Kruskal-Wallis test is widely used to compare three or more independent groups when an ordinal or interval level of data is available, especially when the assumptions of analysis of variance (ANOVA) are not met. If the Kruskal-Wallis statistic is statistically significant, Nemenyi test is an alternative method for further pairwise multiple comparisons to locate the source of significance. Unfortunately, most popular statistical packages do not integrate the Nemenyi test, which is not easy to be calculated by hand. We described the theory and applications of the Kruskal-Wallis and Nemenyi tests, and presented a flexible SAS macro to implement the two tests. The SAS macro was demonstrated by two examples from our cohort study in occupational epidemiology. It provides a useful tool for SAS users to test the differences among three or more independent groups using a nonparametric method.

Keywords

Kruskal-Wallis / Nemenyi / multiple comparisons tests / SAS macro

References

[1] KruskalW.H., WallisW.A.. Use of ranks in one-criterion variance analysis. J Am Stat Assoc, 1952, 47: 583-621

[2] RuxtonG.D., BeauchampG.. Some suggestions about appropriate use of the Kruskal-Wallis test. Anim Behav, 2008, 76: 1083-1087

[3] Theodorsson-NorheimE.. Kruskal-Wallis test: BASIC computer program to perform nonparametric one-way analysis of variance and multiple comparisons on ranks of several independent samples. Comput Methods Programs Biomed, 1986, 23(1): 57-62

[4] AndrewsF.C.. Asymptotic behaviour of some rank tests for analysis of variance. Ann Math Statist, 1954, 25: 724-736

[5] ChanY., WalmsleyR.P.. Learning and understanding the Kruskal-Wallis one-way analysis-of-variance-by-ranks test for differences among three or more independent groups. Phys Ther, 1997, 77(12): 1755-1762

[6] HettmanspergerT.P.. . Statistical inference based on ranks, 1984, New York, John Wiley & Sons Inc.

[7] Nemenyi P. Distribution-free multiple comparisons. PhD Thesis, Princeton University, 1963

[8] DemšarJ.. Statistical comparisons of classifiers over multiple data sets. J Mach Learn Res, 2006, 7: 1-30

[9] GibbonsJ.D.. . Nonparametric Statistical Inference, 1971, New York, McGraw-Hill Inc.

[10] SAS Institute, Cary, NC, 2007

[11] SAS OnlineDoc 9.1.3. SAS Institute, Cary, NC. Available at: http://support.sas.com/onlinedoc/913/docMainpage.jsp

[12] ChenW., HnizdoE., ChenJ.Q., et al.. Risk of silicosis in cohorts of Chinese tin and tungsten miners, and pottery workers (I): an epidemiological study. Am J Ind Med, 2005, 48(1): 1-9

[13] SantosC., NorteA., FradinhoF., et al.. Silicosis — brief review and experience of a pulmonology ward. Rev Port Pneumol, 2010, 16(1): 99-115