How to Report Mann-Whitney U Test in APA Format

If you ran a Mann-Whitney U test in SPSS and now need to write up the results for your thesis or research paper, this guide shows you exactly how to report a Mann-Whitney U test in APA format. You will get the sentence template, a breakdown of which numbers to copy from the SPSS output, how to calculate and report the effect size, and the p-value rules APA requires.

Quick Answer: The APA Sentence Template

The Mann-Whitney U test is the non-parametric alternative to the independent-samples t-test. You use it when you compare two independent groups on an ordinal outcome (such as Likert-scale responses), or when your continuous outcome is not normally distributed within the groups. If you have not checked normality yet, see our guide on how to test normality in SPSS.

The APA 7th edition format for reporting the test (APA, 2019, p. 182) looks like this:

A Mann-Whitney U test indicated that test anxiety was significantly higher among female students (Mdn = 45, n = 15) than among male students (Mdn = 38, n = 15), z(n1 = 15, n2 = 15) = −2.10, p = .036, r = .38.

Three things to notice. First, the statistical symbols — z, p, r, n, and U — are italicized, while Mdn stays in regular type: it is an abbreviation (for median), not a symbol, and APA sets abbreviations in roman. Second, the p-value has no zero before the decimal point. Third, the report always includes an effect size — APA (2019, p. 88) asks for one with every significance test.

The Assumptions Box (Check Before Reporting)

The Mann-Whitney U test has few assumptions, but they are real — check them before you write up anything:

• Independent observations — every participant belongs to one group only, and no participant's score influences another's, within or between groups. Paired or repeated data need the Wilcoxon signed-rank test instead.
• At least ordinal outcome — the outcome can be meaningfully ranked (ordinal, interval, or ratio data all qualify).
• It tests distributions, not means — the test asks whether values in one group are generally larger than in the other (a difference in distributions / mean ranks).
• Similar shapes for median language — describing the result as a difference in medians (as the template does) is only strictly justified when the two groups' distributions have a broadly similar shape and spread. Check this on a boxplot or histograms; if the shapes differ a lot, report the result as a difference in mean ranks instead.

Which Numbers to Take From Your SPSS Output

In SPSS, run the test from Analyze → Nonparametric Tests → Legacy Dialogs → 2 Independent Samples. Move your outcome into the Test Variable List and your grouping variable into the Grouping Variable box. Then click Define Groups and type the two codes your grouping variable actually uses — for example, type 1 in the Group 1 box and 2 in the Group 2 box — and click Continue. (Without these codes the test cannot run; use whatever codes your own data uses, such as 1 = male, 2 = female.) Make sure Mann-Whitney U is ticked under Test Type, then click OK.

On current SPSS versions you can also run it from Analyze → Nonparametric Tests → Independent Samples, the newer dialog. The output looks different — a Hypothesis Test Summary plus a detailed table — and z is labelled Standardized Test Statistic there. The numbers you report are the same.

SPSS prints two tables you need:

1. Ranks table — for each group it gives N, the Mean Rank, and the Sum of Ranks. The mean ranks tell you the direction of the difference: the group with the higher mean rank generally has the higher scores. Note what this table does not contain: medians.
2. Test Statistics table — the Mann-Whitney U value, the Wilcoxon W (the sum of ranks for one group — you do not normally report it), the standardized z value, and the p-value under Asymp. Sig. (2-tailed). If you selected exact significance (via the Exact… button in the dialog; on some versions exact tests need IBM's Exact Tests module), the table also shows an Exact Sig. row — that is the p-value to copy if you report the exact version.

The medians come from somewhere else. The APA sentence reports each group's median, but neither table above prints one. Get them from Analyze → Descriptive Statistics → Explore: put your outcome in the Dependent List and your grouping variable in the Factor List, and click OK — the Descriptives table prints a Median row for each group.

APA (2019, p. 182) gives two accepted versions. If you use the exact test, report the U statistic: U(n1 = 11, n2 = 34) = 285.50, p = .008. If you use the asymptotic (large-sample) version, which is what SPSS reports as z, report that instead: z(n1 = 11, n2 = 34) = 2.85, p = .004. In practice, most students report z: the legacy dialog prints it in the Test Statistics table alongside U — just make sure the p-value you copy matches the version you are reporting (Asymp. Sig. for z, Exact Sig. for the exact U).

One thing that confuses almost everyone: the sign of z. SPSS's z is very often negative even when the second group scores higher — its sign follows the order in which SPSS ranked the groups, not the direction of your result. Never describe direction from the sign of z. Take the direction from the mean ranks (and the medians), as the template does.

Worked Example: From SPSS Output to APA Sentence

Suppose a researcher wants to compare test anxiety scores between two independent groups of students — 15 male and 15 female students (a hypothetical example to show the reporting steps). The anxiety scores are skewed, so a t-test is not appropriate. The SPSS output gives the following hypothetical results: Mann-Whitney U = 62.00, z = −2.10, Asymp. Sig. (2-tailed) = .036, with medians of 38 for males and 45 for females (medians obtained from Explore, as described above).

Step 1: check significance. The p-value (.036) is below the conventional alpha level of .05, so the difference is statistically significant.

Step 2: calculate the effect size. For the Mann-Whitney U test, the standard effect size is r = z / √N (Rosenthal, 1991), where N is the total sample size (both groups combined):

r = −2.10 / √30 = −2.10 / 5.48 = −0.38

Report the absolute value, r = .38. Following Cohen's (1988) conventional benchmarks for r — .10 small, .30 medium, .50 large — this is a medium-sized effect. Treat these benchmarks as rough guidance rather than strict cutoffs; different fields may read the same number differently. (An alternative effect size, the rank-biserial correlation, also exists — but r = z / √N is the version most students use because it needs only the numbers SPSS already prints.)

A note on checking z by hand: SPSS automatically corrects z for ties when it calculates it. With data that contain many tied values — Likert scales above all — the tie correction noticeably changes z and the p-value, so a z you compute from U with the textbook formula may not match your SPSS output exactly. That difference is expected; report the SPSS values.

Step 3: write the sentence.

A Mann-Whitney U test indicated that test anxiety was significantly higher among female students (Mdn = 45, n = 15) than among male students (Mdn = 38, n = 15), z(n1 = 15, n2 = 15) = −2.10, p = .036, r = .38 (medium effect).

Note that the sentence says females scored higher even though z is negative — the direction comes from the medians and mean ranks, exactly as explained above.

p-Value Rules You Must Follow

• Report the exact p-value to three decimal places: p = .036, never p = 0.036 and never p = .04.
• If SPSS prints .000, write p < .001 (this covers any p-value below .0005).
• Never write p = .000 — a p-value is never exactly zero.
• The conventional significance threshold is alpha = .05: values below it are described as "statistically significant".

A Note on Medians vs. Mean Ranks

Strictly speaking, the Mann-Whitney U test compares the distributions of the two groups — SPSS reports this as mean ranks. Many researchers describe the result as a difference in medians, which is only strictly correct when the two groups' distributions have a similar shape. In practice, reporting the medians alongside the test (as in the template above) is standard and accepted — just know that the test itself is a test of distributions, not of medians, and check the shapes (boxplot or histograms by group) before you use median language.

Frequently Asked Questions

Do I report U or z?

Both are accepted by APA (2019, p. 182): U for the exact test, z for the asymptotic version. SPSS prints both, so report the one that matches how you ran the test; most published reports use z.

How do I report a non-significant result?

Use the same template, with honest wording: "A Mann-Whitney U test indicated that the two groups did not differ significantly, z(n1 = 15, n2 = 15) = −0.82, p = .412, r = .15." Still report the effect size — APA asks for it regardless of significance.

Can I use this test with Likert-scale data?

Yes. Ordinal data is one of the main use cases of the Mann-Whitney U test — for example, comparing 5-point Likert responses between two independent groups. Expect many tied values; SPSS's tie-corrected z handles them, which is another reason to report the SPSS values rather than hand-computed ones.

What is the paired equivalent of this test?

If your two measurements come from the same people (for example, before and after scores), the paired equivalent is the Wilcoxon signed-rank test. Our guide on how to report a paired-samples t-test in APA format covers the reporting logic for paired comparisons, which follows the same APA pattern.

Do the groups need equal sample sizes?

No. The test handles unequal group sizes; just report n for each group separately so the reader can see them.

References

American Psychological Association. (2020). Publication manual of the American Psychological Association (7th ed.).

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum Associates.

Rosenthal, R. (1991). Meta-analytic procedures for social research (Rev. ed.). Sage.

---

Need help with your thesis or SPSS analysis? WhatsApp Farzana at +92 326 2047345.

Struggling with SPSS for your thesis? Get the step-by-step guide SPSS Made Simple — 30-page ebook + Excel templates pack. Buy it here: https://whop.com/researchmadesimple/spss-made-simple-a-step-by-step-guide-for-psychology-social-science-students/