How to report nonparametric tests in APA 7
For a Mann–Whitney test, report the test name, U statistic, standardized z when available, p value, and an effect size such as r. A Wilcoxon signed-rank result follows the same structure but uses the statistic label supplied by the analysis, commonly W. These Latin-letter statistical symbols are italicized. If your software uses a different label—such as T—match the software and define it clearly.
The U, W, and z statistics retain leading zeros because their scales are not bounded at 1. Effect-size r and p omit the leading zero. Use two decimals for the test statistic, z, and r, and three decimals for p. Values below .001 should appear as p < .001, never p = .000. A common mistake is reporting only a probability without the test statistic. Another is omitting the direction sign for z or failing to explain how ranks map to groups. Medians and interquartile ranges are often useful context for rank-based analyses and may be described before this inferential sentence.
Worked example
With 30 observations per group and no tied ranks, a two-sided normal approximation without continuity correction gave U = 287.00, standardized z = −2.41, p = .016, and effect-size r = |z| / √60 = .31.
Mann–Whitney U = 287.00, z = −2.41, p = .016, r = .31.
Common questions
How do I report a Wilcoxon signed-rank test?
It follows the same structure but uses the statistic label your analysis supplies, commonly W. If your software uses a different label, such as T, match it and define it clearly.
Which values keep the leading zero?
U, W and z keep it because they are not bounded at 1. Effect-size r and p omit it.
How many decimals should I use?
Two decimals for the test statistic, z and r, and three for p. Write p < .001, never p = .000.
What mistakes should I avoid?
Reporting only a probability without the test statistic, leaving out the sign of z, or not explaining how the ranks map to groups.
Should I also report medians?
Medians and interquartile ranges are often useful context for rank-based analyses and can be described before the inferential sentence.