Search

# Kruskal-Wallis Test for Multiple Samples Help

The Kruskal-Wallis test determines if there is a difference between the medians of different samples. The median is used in this test since it is a better measure of the central tendency of the data than the average for non-normal data. In addition, pairwise comparisons are given to determine which medians are significantly different. The example below shows how to do this test using the SPC for Excel software.

Example (from Statistics and Data Analysis, by Ajit Tamhane and Dorothy Dunlop, Prentice-Hall, 2000)

An experiment was done to compare four different methods of teaching the concept of percentages to sixth graders. 28 classes were randomly assigned to the four methods. A 45 item test was given to all classes. The average test scores are given below. We want to use the Kruskal-Wallis test to determine if there are significant differences in the four methods.

#### Kruskal-Wallis Test for Multiple Samples Output

The output from the Kruskal-Wallis test is shown below.

The data are reprinted at the top of the output.

If p value ? alpha, then there are differences between the pairs.

The pairwise comparisons using the studentized range distribution are listed at the bottom of the output. If there is a significant difference, Yes is printed in red.

Note:

Uses the chi-square distribution to determine the p value.

Scroll to Top