Friedman Test

Non-parameter Statistic

Application

1. The problem objective is to compare 2 or more populations
2. The data are either ordinal or interval, but not normal
3. The samples generated from a randomized block experiment（随机区组检验）

Hypothesis

$H_0$: The locations of all $k$ populations are the same. $H_1$: At least two population locations differ.

Test Statistic

The test statistic of Friedman, is denoted by F. Here, $b=$ number of blocks.

We can use this method to test whether 2 populations differ.

Use the Friedman Test to test whether the difference exists, but if we want to know the position difference. We have to use the Sign Test Here is whether to choose Kruskal Wallis Test or Friedman Test