612 STATISTICS AND PROBABILITY
Table 63.2
12 3 456
Class Class z-value for Area from Area for Expected
mid-point boundaries,x class boundary 0 toz class frequency
4.009 −2.56 0.4948
4.011 0.0255 13
4.013 −1.87 0.4693
4.015 0.0863 43
4.017 −1.19 0.3830
4.019 0.1880 94
4.021 −0.51 0.1950
4.023 0.2628 131
4.025 0.17 0.0678
4.027 0.2345 117
4.029 0.85 0.3023
4.031 0.1347 67
4.033 1.53 0.4370
4.035 0.0494 25
4.037 2.21 0.4864
Total: 490
To determine theχ^2 -value
Theχ^2 -value is calculated using a tabular method
as shown below.
Diameter Observed Expected,
of rivets frequency,o frequency,e
4.011 12 13
4.015 47 43
4.019 86 94
4.023 123 131
4.027 107 117
4.031 97 67
4.035 28 25
o−e (o−e)^2
(o−e)^2
e
− 1 1 0.0769
4 16 0.3721
− 8 64 0.6809
− 8 64 0.4885
− 10 100 0.8547
30 900 13.4328
3 9 0.3600
χ^2 =
∑
{
(o−e)^2
e
}
= 16. 2659
To test the significance of theχ^2 -value
The number of degrees of freedom is given by
N− 1 −M, whereMis the number of estimated
parameters in the population. Both the mean and the
standard deviation of the population are based on
the sample value,M=2, henceν= 7 − 1 − 2 =4.
From Table 63.1, theχ^2 p-value corresponding toχ^20. 95
andν 4 is 9.49.Hence the null hypothesis that the
diameters of the rivets are normally distributed
is rejected.Forχ^20. 10 ,ν 4 , theχ^2 p-value is 1.06, hence
the fit is not ‘too good’. Since the null hypothesis
is rejected, the second significance test need not be
carried out.
Now try the following exercise.
Exercise 227 Further problems on fitting
data to theoretical distributions
- Test the null hypothesis that the observed data
given below fits a binomial distribution of the
form 250(0. 6 + 0 .4)^7 at a level of significance
of 0.05.
Observed
frequency 8 27 62 79 45 24 5 0
Is the fit of the data ‘too good’ at a level of
confidence of 90%?
⎡ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣
Expected frequencies:
7, 33, 65, 73, 48, 19, 4, 0;
χ^2 -value= 3 .62,χ^20. 95 ,
ν 7 = 14 .1, hence hypothesis
accepted.χ^20. 10 ,ν 7 = 2 .83,
hence data is not ‘too good’
⎤ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦