Which statistical test assesses differences in matched pairs when the outcome is ordinal?

Which statistical test assesses differences in matched pairs when the outcome is ordinal?

• A. Wilcoxon signed-rank test
• B. Mann-Whitney U test
• C. Student's t-test for paired samples
• D. Sign test

Answer: (A)

When outcomes are ordinal and you have matched pairs, you want a nonparametric approach that honors the order of the data and leverages how big the differences are. The Wilcoxon signed-rank test does just that: for each pair, you look at the difference, ignore any zero differences, take the absolute value and rank these magnitudes, then assign the original difference’s sign to each rank. Summing the signed ranks gives a statistic that reflects whether there’s a systematic tendency for one condition to yield higher values than the other. Because it uses the ranking of differences and their signs, it works well with ordinal data and paired design, without assuming normality. Mann-Whitney U is for two independent groups, not paired data, so it’s not appropriate here. The paired t-test requires the differences to be approximately normal and the data to be continuous, which isn’t suitable for ordinal outcomes. The sign test looks only at the direction of differences and ignores their magnitudes, making it less powerful than the Wilcoxon test.

When outcomes are ordinal and you have matched pairs, you want a nonparametric approach that honors the order of the data and leverages how big the differences are. The Wilcoxon signed-rank test does just that: for each pair, you look at the difference, ignore any zero differences, take the absolute value and rank these magnitudes, then assign the original difference’s sign to each rank. Summing the signed ranks gives a statistic that reflects whether there’s a systematic tendency for one condition to yield higher values than the other. Because it uses the ranking of differences and their signs, it works well with ordinal data and paired design, without assuming normality.

Mann-Whitney U is for two independent groups, not paired data, so it’s not appropriate here. The paired t-test requires the differences to be approximately normal and the data to be continuous, which isn’t suitable for ordinal outcomes. The sign test looks only at the direction of differences and ignores their magnitudes, making it less powerful than the Wilcoxon test.