Mann-Whitney and Kruskal-Wallis tests
Description: These tests can be considered non-parametric alternatives to the standard parametric independent samples t-test and one-way ANOVA. Rather than testing for differences in group means on a continuous dependent variable, these tests compare the mean RANKS on the dependent variable across groups. If the shapes of the underlying continuous distributions (on the DV) for your groups are equivalent, then you can consider significant results from the Mann-Whitney or Kruskal-Wallis test as reflecting differences in group medians. If the shapes are different, you can still use these tests. However, the test becomes one of stochastic dominance of at least one group over the others on the dependent variable. Stochastic dominance refers to 'the probability that a randomly drawn observation from one group will be greater than a randomly drawn observation from another' ((Dinno, 2015, p. 292).
Newest presentations:
How to perform a Mann-Whitney test using SPSS (February, 2021): video , SPSS data
How to perform a Kruskal-Wallis test with Dunn's post hoc tests using SPSS (February 2021): video , SPSS data , Powerpoint
---------------------------------------------------------------------------------------------------------------
Older video presentations
Video presentation of Mann-Whitney test using SPSS, SPSS dataset 1, SPSS dataset 2
Video presentation of Kruskal-Wallace test using SPSS, SPSS data
*Citation
Dinno, A. (2015). Nonparametric pairwise multiple comparisons in independent groups using Dunn's test. The Stata Journal, 15, 292-300.