Wilcoxon Signed Rank Sum Test

input fleeting Statistics 3 requirements:

1. 2 populations
2. No need to be normality requirement but are either ordinal or interval.
3. Matched pairs ( not 相互独立)

Core: We ignore the the exact difference of numbers, but to only record the sign, for example: 65 45 32 23 they could be record as 4 3 2 1 And that’s called the sign test.

The Method

The method is same as the Sign Test , We eliminate all differences that are equal to 0. And we need to calculate the sum of ranks. ($T^{+}$ and $T^{-}$)

Classification

$n\le 30$ : small samples $n>30$ : big samples Mean: $E(T)=\frac{n(n+1)}{4}$ Standard deviation: ${\sigma }_T=\sqrt{\frac{n(n+1)(2n+1)}{24}}$ The standardized test statistic is $z=\frac{T-E(T)}{{\sigma }_T}$

==Related to: Wilcoxon Rank Sum Test