Session Ready
Exercise

Rank sum tests

Another class of non-parametric hypothesis tests are called rank sum tests. Ranks are the positions of numeric values from smallest to largest. Think of them as positions in running events: whoever has the fastest (smallest) time is rank 1, second fastest is rank 2, and so on.

By calculating on the ranks of data instead of the actual values, you can avoid making assumptions about the distribution of the test statistic. It's most robust in the same way that a median is more robust than a mean.

Two commonly used rank-based tests are the Wilcoxon-Mann-Whitney test, which is like a non-parametric t-test, and the Kruskal-Wallis test, which is like a non-parametric ANOVA.

late_shipments is available.

Instructions 1/2
undefined XP
  • 1
    • Using the late_shipments dataset, run a Wilcoxon-Mann-Whitney test on the weight in kilograms versus whether or not the shipment was late.
    • 2
      • Using the late_shipments dataset, run a Kruskal-Wallace test on the weight in kilograms versus the shipment mode.