The Wilcoxon signed rank test assumes only a continuous and symmetric distribution with mean =median =
If we have a sample
then we find
and rank them from smallest to largest.
The null hypothesis is![]()
The test statistic is
the sum of the ranks of those
with
positive.
The alternative hypothesis may be stated as one of those below, with associated rejection region for a level
test, where
and
are obtained from tables.
|
|
|
|
|
|
|
|
Either |
Example: A manufacturer of electric irons, wishing to test the accuracy of the thermostat control at the 500 degree fahrenheit setting, obtains actual temperatures at that setting for fifteen irons. They are
494.6, 510.8, 487.5, 493.2, 502.6, 485, 495.9, 498.2, 501.6, 497.3, 492.0, 504.3, 499.2, 493.5, 505.8
Assuming a symmetric distribution for the temperature, we can apply the Wilcoxon signed rank test. Subtracting 500 from each gives
-5.4, 10.8, -12.5, -6.8, 2.6, -15, -4.1, -1.8, 1.6, -2.7, -8.0, 4.3,-0.8, -6.5,5.8
The ranks are obtained by ordering these from smallest to largest, obtaining,
|
Absolute Value |
0.8 |
1.6 |
1.8 |
2.6 |
2.7 |
4.1 |
4.3 |
5.6 |
5.8 |
6.8 |
6.8 |
8 |
10.8 |
12.5 |
15 |
|
Rank |
1 |
2 |
3 |
4 |
5 |
6 |
7 |
8 |
9 |
10 |
11 |
12 |
13 |
14 |
15 |
|
Sign |
- |
+ |
- |
+ |
- |
- |
+ |
- |
+ |
- |
- |
- |
+ |
- |
- |
Thus
From the Wilcoxon tables,
when
is true so
is rejected if either
or
neither of which apply here so
is not rejected. There is no evidence at this level to suppose the thermostat is defective.