MANN-WHITNEY U TEST
MANN-WHITNEY U TEST
Use when assumptions of independent two-sample t-test are violated (e.g., non-normal distribution).
Results: group rank differences instead of group mean differences.
STEPS FOR CARRYING OUT MANN-WHITNEY
Define hypotheses:
Null (Ho): No difference between groups.
Alternative (H1): Some difference exists.
Check assumptions.
Compare medians of two groups.
Calculate:
t-statistic & effect size
Mean ranks
U value
z value
p-value
Interpret results.
HYPOTHESES
Null Hypothesis: Ho: No difference between treatment and control.
Alternative Hypothesis: H1: Some difference between treatment and control.
DATASET OVERVIEW
Treatment (Yoga) Group: n = 16
Control Group: n = 16
Number of Prison Incidents recorded for both groups.
ASSUMPTIONS
Data violated normality assumptions for t-test.
Presence of extreme outliers.
MEDIAN AND DESCRIPTIVE STATISTICS
Treatment Median = 2, Control Median = 3
Treatment Mean = 3.625, Control Mean = 5.125
Sample Variance for Treatment = 16.25, Control = 39.32
CALCULATING MANN-WHITNEY
Use external tools (e.g., Mann-Whitney U Test Calculator) for calculations.
RESULTS SUMMARY
U-value = 114, Z-Score = -0.5088, p-value = 0.610
No significant difference found (U = 114, z = -0.51, p = .610).
If significant, include effect size with the statement.
EFFECT SIZE
Formula: r = z/√n
Interpret effect size:
r < 0.3: small effect
0.3 ≤ r < 0.5: medium effect
r ≥ 0.5: large effect
WRITING UP RESULTS
Example format:
"A Mann-Whitney U test was conducted to examine differences in prison incidents…"
Include effect size if significant.