Kruskal-Wallis rank sum test
Given that we found in the last exercise that the homogeneity of variance assumption of linear modeling was violated, we may want to try an alternative.
One non-parametric alternative to ANOVA is the Kruskal-Wallis rank sum test. For those with some statistics knowledge, it is an extension of the Mann-Whitney U test for when there are more than two groups, like with our grade
variable. For us, the null hypothesis for this test would be that all of the int_rate
s have the same ranking by grade
.
The Kruskal-Wallis rank sum test can be conducted using the kruskal.test()
function, available in base R. Luckily for you, the use of this function is very similar to using lm()
or aov()
: you input a formula and a dataset, and a result is returned.
This exercise is part of the course
Experimental Design in R
Exercise instructions
- Use
kruskal.test()
to examine whetherint_rate
varies bygrade
when a non-parametric model is employed.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Conduct the Kruskal-Wallis rank sum test
kruskal.test(___,
data = ___)