1. Learn
  2. /
  3. Courses
  4. /
  5. A/B Testing in R

Exercise

Mann-Whitney U analysis

A Mann-Whitney U test is a non-parametric assessment of the median of two groups, making no assumption about the distribution. Similar to the t-test, a Mann-Whitney U test can be assessed by finding the Cohen's d effect size and running a power analysis on the resulting values.

A dataset of subjects playing a video game includes a column indicating the number of rounds of the game they have played and two versions of the game which each subject participated in one of. You are interested in whether there are a different number of rounds played in the two versions. The distributions are not normal, indicating the Mann-Whitney U test should be used to assess the group differences.

The gamedata dataset, as well as the pwr and effectsize packages have been loaded for you.

Instructions 1/3

undefined XP
    1
    2
    3
  • Run a two-tailed Mann-Whitney U test on the sum_gamerounds data grouped by the version of the gamedata dataset.