Kruskal Wallis H test assumptions

Rank based non-parametric test
Called one-way ANOVA on ranks
Test determines statistically significant differences between ≥ 3 groups of independent variable and continuous or ordinal dependent variable
Can be used to compare groups in questionnaire (5 or 7-point Likert scale)
Before performing the test, ensure that the 4 assumptions are not violated.
Assumptions
1) One Dependent variable is
Ordinal variable(5 or 7-point Likert scale) or
Continuous variable (not normal distribution or outliers)
2) Another independent variable (nominal, 3 or more categories)
Ethnicity
Profession
3) Independence of observations
No paired samples, each participant is part of only one group
4) Same distribution shape for all groups
If distribution of data in each groups have
Similar shape (same variability)
Use medians to compare differences between groups
Different shape
Use mean ranks compare differences between groups
Examine data shape using
Histograms
Boxplots
See also
Non parametric Kruskal Wallis test + POST HOC TEST in SPSS TUTORIAL
• Non parametric Kruskal...
ON-PARAMETRIC KRUSKAL WALLIS TEST FOR MORE THAN 3 GROUPS IN SPSS TUTORIAL
• NON-PARAMETRIC KRUSKAL...

[@mathematicsanddataanalysis5791]

Thanks a lot