Which test is non-parametric for comparing two related (dependent) samples?

Which test is non-parametric for comparing two related (dependent) samples?

• A. Wilcoxon
• B. Mann-Whitney U
• C. Kruskal-Wallis
• D. Friedman

Answer: (A)

When comparing two related (paired) samples without assuming normal distribution, the Wilcoxon signed-rank test is used. It focuses on the differences within each pair rather than raw values. By taking each pair’s difference, ranking the absolute differences, and then reattaching the sign, the test assesses whether the median difference between the paired observations is zero. This reliance on ranks rather than on a specific distribution makes it non-parametric and appropriate for ordinal data or when the differences aren’t normally distributed. Two related samples contrasts to two independent samples (where the Mann-Whitney U test is used), and contrasts to more than two groups (Kruskal-Wallis for independent samples; Friedman for more than two related samples).

When comparing two related (paired) samples without assuming normal distribution, the Wilcoxon signed-rank test is used. It focuses on the differences within each pair rather than raw values. By taking each pair’s difference, ranking the absolute differences, and then reattaching the sign, the test assesses whether the median difference between the paired observations is zero. This reliance on ranks rather than on a specific distribution makes it non-parametric and appropriate for ordinal data or when the differences aren’t normally distributed.

Two related samples contrasts to two independent samples (where the Mann-Whitney U test is used), and contrasts to more than two groups (Kruskal-Wallis for independent samples; Friedman for more than two related samples).