Wilcoxon rank sum large sample example
A subject was taught using two different pedagogies. Sample of students from each method then wrote a common exam. These were the results:
|
Method A |
70 |
90 |
82 |
64 |
86 |
77 |
84 |
|
|
79 |
82 |
89 |
73 |
81 |
83 |
66 |
|
Method B |
86 |
78 |
90 |
82 |
65 |
87 |
|
|
|
80 |
88 |
95 |
85 |
76 |
94 |
|
No assumptions about the normality of the data were made. Is there any significant difference in average grades between the two pedagogies? Test at a 5% level of significance.
Here is the Megastat output:
|
Wilcoxon - Mann/Whitney
Test |
|
|
||
|
|
|
|
|
|
|
|
n |
sum of ranks |
|
|
|
|
12 |
190 |
Method B |
|
|
|
14 |
161 |
Method A |
|
|
|
26 |
351 |
total |
|
|
|
|
|
|
|
|
|
|
162.00 |
expected value |
|
|
|
|
19.44 |
standard deviation |
|
|
|
|
1.41 |
z |
|
|
|
|
.1572 |
p-value (two-tailed) |
|
It should be noted that there is a bug in the Megastat macro. The correct Z value is (190 – 162)/19.44 = 1.44.