Introduction
Non-parametric tests don't assume a specific distribution. They're useful when assumptions aren't met.
Wilcoxon Test
# One-sample
wilcox.test(x, mu = 0)
# Two-sample
wilcox.test(group1, group2)
# Paired
wilcox.test(before, after, paired = TRUE)
Mann-Whitney U Test
# Two independent samples
wilcox.test(value ~ group, data = df)
Kruskal-Wallis Test
# Non-parametric ANOVA
kruskal.test(value ~ group, data = df)
Friedman Test
# Repeated measures
friedman.test(value ~ group | subject, data = df)
Spearman Correlation
# Non-parametric correlation
cor.test(x, y, method = "spearman")
Summary
Use non-parametric tests when data doesn't meet parametric assumptions. They are more robust.