3) Calculation of W:
2
2
W
=
b
S
There is a 5% probability of not having a normal distribution if W is lower than W95 given in table C:
Table C :
N
W95
15
0.881
20
0.905
25
0.918
30
0.927
35
0.934
40
0.940
45
0.945
50
0.947
5 - NORMAL DISTRIBUTION TEST: POPULATION OVER 50 MEASUREMENTS (CHI-SQUARED
TEST)
1) Distribute into classes of at least 4 or 5 measurements
2) Calculate the mean and standard deviation
mean:
∑
x
i
=
X
---------- -
N
standard deviation:
2
∑
(
)
x
–
X
i
σ
=
--------------------------- -
N 1
–
3) Calculate for each class limit li:
(
)
l
–
X
i
u
=
---------------- -
i
σ
4) Calculate:
2
(
'
)
n
–
n
2
i
i
∑
χ
=
----------------------- -
'
n
i
where:
n = number of measurements in class i
n' = theoretical number of measurements for a normal distribution
'
[
( ) F u
(
n
=
N F u
–
i
i
i 1
–
DELTA4000/5000/4D/5D
) ]
6159938010-07
English
23 / 200