When to Use the Mann-Whitney U Test (vs. t-Test)

May 2026·5 min read

When to use the Mann-Whitney U test comes down to one situation: you want to compare two independent groups, but the normality assumption for a t-test cannot be met. Mann-Whitney U is the non-parametric alternative — it compares rank distributions instead of means and works reliably with small samples and ordinal data. This guide covers the decision, the SPSS steps, and APA reporting with effect size r.

What Is Mann-Whitney U? The Non-Parametric t-Test

Mann-Whitney U tests whether the rank distributions of two independent groups differ significantly. Instead of comparing means, it ranks all observations from both groups together and checks whether one group’s ranks tend to be higher than the other’s.

This makes it suitable when the t-test’s normality assumption is violated, when data are ordinal, or when sample sizes are small and you cannot rely on the Central Limit Theorem.

Despite being non-parametric, Mann-Whitney U is not weaker than the t-test — it is simply more appropriate under different conditions.

Mann-Whitney U vs. t-Test: Side-by-Side Comparison

Both tests compare two independent groups. The choice depends on your data and sample size:

When Normality Is Violated — What to Do

Run Shapiro-Wilk first: Analyze → Descriptive Statistics → Explore → Plots → Normality plots with tests.

• If Shapiro-Wilk p > .05: normality not significantly violated → use the independent samples t-test.
• If Shapiro-Wilk p < .05 AND n < 30 per group: normality violated → use Mann-Whitney U.
• If Shapiro-Wilk p < .05 AND n ≥ 30 per group: use the t-test anyway (Central Limit Theorem), but inspect the Q-Q plot and histogram.

[TIP] With n > 100 per group, Shapiro-Wilk almost always returns significant even for trivial deviations. In this case, rely on the Q-Q plot and histogram rather than the Shapiro-Wilk p-value.

How to Run Mann-Whitney U in SPSS: Step-by-Step

SPSS path: Analyze → Nonparametric Tests → Legacy Dialogs → 2 Independent Samples → move your DV to Test Variable List, grouping variable to Grouping Variable → Define Groups → tick Mann-Whitney U → OK.

• In the output, read:
• Mann-Whitney U statistic
• Wilcoxon W statistic (sum of ranks for the smaller group)
• Z statistic (standardised test statistic)
• Asymptotic Significance (2-tailed) = your p-value

Report medians (not means) for each group.

Interpreting the Output and Calculating Effect Size r

The p-value tells you whether the rank distributions differ significantly. The U statistic alone is not meaningful without context.

• Effect size r = Z ÷ √N, where N = total sample size.
• Benchmarks: r = .10 small, .30 medium, .50 large.

Example: Z = −2.84, N = 121 → r = 2.84 ÷ √121 = 2.84 ÷ 11 = .26 (small-medium effect).

APA Reporting Format for Mann-Whitney U

• Significant:
• "Students who exercised regularly (Mdn = 20.0) showed significantly lower stress ranks than non-exercising students (Mdn = 25.0), U = 1324.5, z = −2.84, p = .005, r = .26."
• Non-significant:
• "No significant difference in stress ranks was found between exercising and non-exercising students (Mdn = 21.0 vs. Mdn = 23.0), U = 1654.0, z = −1.12, p = .263, r = .10."

[INFO] Report medians (Mdn), not means (M), when using Mann-Whitney U. The test ranks data rather than using means, so medians are the appropriate descriptive statistic.