Mann-Whitney’s U Test

  • Assumptions to check:

    • Data has to be at least ordinal scale

    • Group independence

  • Report design

    • Name the statistical package or program used in the analysis

  • Report statistics:

    •  the U-statistic (or z-statistic),  exact p-value, sample and group sizes 

    • Strongly recommended: report effect size 

  • Example

    • Sample size >30

      • Methods section:  To evaluate the hypothesis we used Mann-Whitney’s U test.  We used GraphPad Prism (RRID:SCR_002798) to perform the  analysis.

      • Results section: A Mann-Whitney U test was performed to evaluate whether the reaction time for the female group (n=35) had the same values as the reaction time for the male group (n=37). The results indicated that females had significantly greater reaction time than males with  z = -2.05, p = .040 with medium effect size r=0.48. 

    • Small sample size 

      • Methods section: To evaluate the hypothesis we used Mann-Whitney’s U test.  We used GraphPad Prism (RRID:SCR_002798) to perform the  analysis.

      • Results section: A Mann-Whitney U test was performed to evaluate whether the reaction time for the female group (n=10) had the same values as the reaction time for the male group (n=15). The results indicated that females had significantly greater reaction time than males with  U = 12, p = .040 with small effect size r=0.28.