What is the Kruskal-Wallis Test?

The Kruskal-Wallis Test (also known as the Kruskal-Wallis H test) is a rank-based non-parametric test that can be used to determine if there are statistically significant differences between two or more groups of an independent variable on a continuous or ordinal dependent variable. It is essentially an extension of the Mann-Whitney U test to cover more than two groups. The test statistic approximates a Chi-square distribution.

When to use it?

Use the Kruskal-Wallis test when:

• Multiple Groups: You are comparing 3 or more independent groups.
• Failed Normality: Your residuals do not follow a normal distribution, making One-Way ANOVA invalid.
• Measurement Scale: Your dependent variable is ordinal or continuous but skewed.
• Homogeneity: While it doesn't assume normality, it assumes reasonably similar distribution shapes across groups.

How our tool works (Accuracy & Stack)

This tool leverages the power of Python's SciPy ecosystem (scipy.stats.kruskal) hosted on Google Cloud. This ensures computational stability and speed. We have validated our output against R to ensure that the H-statistic and P-values are accurate to 4 decimal places.

Beyond the numbers, visualization is key to understanding group differences. We provide:

• Raincloud Plots: State-of-the-art visualization for distribution density and raw data jitter.
• Box & Violin Plots: Standard tools for viewing interquartile ranges and density.
• Histograms: For raw frequency analysis.

If significant results are found, you can download a full PDF Report containing the statistical breakdown and all generated charts.

Input Requirements & Limitations

We have optimized this tool for speed and privacy. The limits are:

• Row Limit: Supports up to 5,000 rows per group.
• Column Limit: Supports up to 10 groups (columns).
• Total Data Points: Maximum 50,000 data points.

This ensures that the rendering of interactive JavaScript charts remains smooth on any device. Supports CSV import.

Comparison: Kruskal-Wallis vs. One-Way ANOVA

The One-Way ANOVA is a parametric test that compares the means of three or more groups. It is highly sensitive to outliers and requires data normality. The Kruskal-Wallis test compares the median ranks. If your P-value is significant in Kruskal-Wallis, it indicates that at least one group stochastically dominates another, requiring Post-Hoc testing to identify exactly which groups differ.