Kruskal–Wallis Test

Introduction: Kruskal–Wallis Test

The Kruskal–Wallis test is a non-parametric statistical method used to determine whether there are significant differences between three or more independent groups. It serves as the non-parametric alternative to one-way ANOVA when the data does not follow a normal distribution or when sample sizes are small or unequal.

Background

Developed by William Kruskal and W. Allen Wallis in 1952, this test extends the Mann–Whitney U test to multiple groups. Instead of comparing means, it compares the median ranks of each group to assess whether they come from the same population. Because it does not assume normality, the Kruskal–Wallis test is especially useful in social sciences, biology, and medical research where data may be ordinal or skewed.

Key Elements/Features

• Type of Test: Non-parametric; based on ranks rather than raw data.
• Assumptions: Independent samples, ordinal or continuous data, and similar distribution shapes across groups.
• Test Statistic (H): Calculated using the ranks of the combined data set. A higher H-value indicates greater differences between groups.
• Post-hoc Analysis: If the test is significant, pairwise comparisons (e.g., Dunn’s test) are used to identify which specific groups differ.

Applications/Examples

The Kruskal–Wallis test is used in situations where ANOVA assumptions cannot be met. Examples include comparing customer satisfaction levels across several stores, examining differences in plant growth under various soil types, or analysing patient recovery scores across treatment groups. Researchers often visualise the results with boxplots to show median and variability.

Relevance/Impact

This test is an essential tool for robust data analysis when dealing with non-normal or ordinal data. It enhances reliability by avoiding misleading conclusions that may arise from parametric tests under invalid assumptions. Its simplicity and versatility make it a staple in non-parametric statistical analysis.

See also

• 1-Sample Sign Test
• Wilcoxon Signed-Rank Test
• Mann–Whitney U Test
• Non-Parametric Methods