What is a nonparametric test?

• A. A test that requires large samples.
• B. A test that assumes a normal distribution and uses means.
• C. A test that does not assume a normal distribution; used with ordinal data or non-normal data or small samples.
• D. A test that only compares variances.

Answer: (C)

Nonparametric tests are distribution-free procedures that don’t rely on the data following a normal distribution. They’re designed for situations where the data may be ordinal, not normally distributed, or when you have a small sample. Because they don’t depend on means and standard deviations in the same way as parametric tests, they use ranks or medians instead, which makes them robust to outliers and violations of normality. This flexibility is what makes them appropriate for analyzing data that don’t meet parametric assumptions or when the measurement scale is not interval/ratio.

For example, tests like the Mann-Whitney U, Wilcoxon signed-rank, and Kruskal-Wallis are nonparametric and rely on ranks rather than raw values, and Spearman correlation uses rank-based association instead of Pearson’s correlation. In contrast, a description that emphasizes normal distributions and means fits parametric tests, not nonparametric ones, and a description focused solely on comparing variances misses the main use of nonparametric methods for differences in central tendency or ranks.