Sampling Theory

F-test: ( 𝐹𝑐𝑎𝑙 - calculated value of 𝐹 )

(Test for equality of Population The table value for 𝑛1 − 1, 𝑛2 − 1 = 9,13

variances using F-distribution) degrees of freedom at 5% level of significance is
Suppose we want to test whether two 𝐹0.05 = 2.71.
independent samples 𝑥1 , 𝑥2 , … … … . . 𝑥𝑛 1 and
( 𝐹0.05 - tabulated value of 𝐹 for 5% level of
𝑦1 , 𝑦2 , … … … . . 𝑦𝑛 2 have been drawn from the
significance)
normal population with the same variance 𝜎 2 .
As 𝐹𝑐𝑎𝑙 < 𝐹0.05 , 𝐻0 is accepted.
Let
𝑛1 2 𝑛1
𝑖=1 𝑖
𝑥 −𝑥
𝑛 1𝑆2
𝑥 𝑖 −𝑥 2 Therefore, the difference is not significant.
𝑛 1 −1
1 ∵ 𝑆12 = 𝑖=1
𝑛 1 −1 𝑛1
𝐹= 𝑛2 2 = 𝑛 2𝑆2
𝑛2 Note:
𝑦 −𝑦 2 𝑦 𝑖 −𝑦 2
𝑖=1 𝑖
𝑛 2 −1 𝑛 2 −1 𝑆22 = 𝑖=1
The numerator value (2.5) is more than the
𝑛2
denominator value (1.55) and hence 𝐹0.05 was
𝑢2 computed for 𝑛1 − 1, 𝑛2 − 1 degrees of freedom,
⟹ 𝐹 = 𝑣2
otherwise, we have to take test statistic as
𝑛 2𝑆2
𝑛 𝑆2 𝑛 𝑆2 2
Where 𝑢2 = 𝑛 1−1
1
and 𝑣 2 = 𝑛 2−1
2
𝑛 2 −1
1 2 𝐹= 𝑛 1𝑆2
and then 𝐹0.05 has to be determined for
1
𝑛 1 −1
The test statistic 𝐹 defined above follows 𝑛2 − 1, 𝑛1 − 1 degrees of freedom.
F-distribution with 𝑛1 − 1, 𝑛2 − 1 degrees of
freedom considering the greater of 𝑢2 and 𝑣 2 always
in the numerator of 𝐹. 2. In one sample of 8 observations the sum of the
squares of deviations of the sample values from
Note: the sample mean was 84.4 and in another sample
Table value for 𝑛1 − 1, 𝑛2 − 1 degrees of of 10 observations, it was 102.6. Test whether
freedom is different from the value for 𝑛2 − 1, the difference in variance is significant at 5%
𝑛1−1 degrees of freedom. level using F-test.

Working procedure Soln: Given 𝑛1 = 8; 𝑛2 = 10;

1. Set the null hypothesis 𝐻0 : 𝜎𝑥2 = 𝜎𝑦2 =𝜎 2
8 2 10 2
𝑛1
𝑥 −𝑥
2 𝑖=1 𝑥𝑖 − 𝑥 = 84.4 ; 𝑖=1 𝑦𝑖 − 𝑦 = 102.6
𝑖=1 𝑖 𝑛 1𝑆21
𝑛 1 −1
𝑛 1 −1
2. Test statistic 𝐹 = 𝑛2 2 = 𝑛 2𝑆2
𝐻0 : 𝜎𝑥2 = 𝜎𝑦2
𝑦 −𝑦 2
𝑖=1 𝑖
𝑛 2 −1 𝑛 2 −1
Under the null hypothesis
3. Compare the calculated value of 𝐹 for the two
given samples with the tabulated value of 𝐹 for 𝑛1
𝑖=1 𝑖
𝑥 −𝑥
2
84.4
𝑛1 − 1, 𝑛2 − 1 degrees of freedom at any 𝑛 1 −1
7
𝐹= 𝑛2 2 = 102 .6 = 1.056 = 𝐹𝑐𝑎𝑙
required level of significance. 𝑦 −𝑦
𝑖=1 𝑖 9
𝑛 2 −1

Problems
The table value for (7,9) degrees of freedom (d.f) at
1. Test the equality of standard deviations for the
data given below at 5% level of significance: 5% level of significance is 𝐹0.05 = 3.29
𝑛1 = 10; 𝑛2 = 14; 𝑆1 = 1.5; 𝑆2 = 1.2 As 𝐹𝑐𝑎𝑙 < 𝐹0.05 , 𝐻0 is accepted.
Soln: 𝐻0 : 𝜎𝑥2 = 𝜎𝑦2 = 𝜎 2 Therefore, the difference is not significant at 5%
level of significance.
𝑛1 𝑆12
𝑛 −1 2.5
𝐹= 1 2 = = 1.61 = 𝐹𝑐𝑎𝑙
𝑛2 𝑆2 1.55
𝑛2 − 1

Page 15
 Sampling Theory

3. Two random samples drawn from 2 normal 4. The daily wages in rupees of skilled workers in
populations are given below. Test whether the 2 two cities are as follows.
populations have the same variance. Size of sample S.D of wages
City
of workers in the samples
Sample A 28 30 32 33 31 29 34 City A 16 25
Sample B 29 30 30 24 27 28 - City B 13 32
Examine whether the samples have been drawn from Test at 5% level the equality of variances of the
normal population having the same variance. wage distribution in the two cities.

Soln: Given 𝑛1 = 7; 𝑛2 = 6; Soln: Given 𝑛1 = 16; 𝑛2 = 13; 𝑆1 = 25; 𝑆2 = 32

28+30+32+33+31+29+34 𝑛 1 𝑆12 16×25 2

𝑥= = 31 = = 666.67
7 𝑛 1 −1 15

29+30+30+24+27+28 𝑛 2 𝑆22 13×32 2

𝑦= 6
= 28 = = 1109.33
𝑛 2 −1 12
𝑛1
𝑥 𝑖 −𝑥 2
𝑆12 = 𝑖=1 We have to take 𝐹 as
𝑛1

𝑛1 𝑛2 𝑆22
∴ 𝑛1 𝑆12 = 𝑖=1 𝑥𝑖 − 𝑥 2
𝑛 −1
𝐹 = 2 2 = 1.66
⇒ 𝑛1 𝑆12 = −3 2
+ −1 2
+ 1 2
+ 2 2
+ 0 2
+ −2 2
+ 3 2 𝑛1 𝑆1
𝑛1 − 1
∴ 𝑛1 𝑆12 = 28
Table value for 𝑛2 − 1, 𝑛1 − 1 = 12,15 degrees
of freedom is 𝐹0.05 = 2.48 .
Similarly,
As 𝐹𝑐𝑎𝑙 < 𝐹0.05 , 𝐻0 is accepted.
𝑛2
𝑦 𝑖 −𝑦 2
𝑆22 = 𝑖=1
𝑛2

𝑛2 Similar problems for practice

𝑛2 𝑆22 = 𝑖=1 𝑦𝑖 − 𝑦 2

⇒ 𝑛2 𝑆22 = 1 2
+ 2 2
+ 2 2
+ −4 2
+ −1 2
+ 0 2 1. For two samples of sizes 8 and 12 the observed
variances are 0.064 and 0.024. Test the
∴ 𝑛2 𝑆22 = 26 hypothesis that the samples came from normal
populations with variances equal.
𝑛 1𝑆21
𝑛 1 −1 4.67
If we take 𝐹 = 𝑛 2𝑆2
then, 𝐹 = 5.2 2. In a sample of 8 observations the sum of the
2
𝑛 2 −1 squared deviations of items from the mean was
94.5. In another sample of 10 observations the
Numerator value is less than the denominator and
value was found to be 101.7. Test whether the
hence we have to take
difference is significant.
𝑛 2𝑆22
𝑛 2 −1 3. Two random samples drawn from two normal
𝐹= 𝑛 1𝑆21
𝑛 1 −1
populations are

26 A 63 65 68 69 71 72 - - - -
5
∴𝐹= 28 = 1.11 = 𝐹𝑐𝑎𝑙 B 63 62 65 66 69 69 70 71 72 73
6 Test whether the two populations have the same
Table value of for 𝐹 for 𝑛2 − 1, 𝑛1 − 1 = 5,6 variance.
degrees of freedom is 𝐹0.05 = 4.39 .
As 𝐹𝑐𝑎𝑙 < 𝐹0.05 , 𝐻0 is accepted.

Hence the two samples could have been drawn from

the population having the same variance.

Page 16
 Sampling Theory

Examples on Fitting theoretical Expected

Observed
𝑥 𝑃 𝑥 frequencies
distribution to a given collection of 200 𝑃 𝑥
frequencies
observed data: 0 𝑃 0 = 0.6065 121.3 123
1 𝑃 1 = 0.3033 60.66 59
Example (1): 2 𝑃 2 = 0.0758 15.16 14
Suppose five unbiased coins are tossed and 3 𝑃 3 = 0.0126 2.52 3
numbers of heads are noted. The experiment is 4 𝑃 4 = 0.0027 0.54 1
repeated 64 times and the following distribution is Total 1 N~200 N=200
obtained.

No. of heads 0 1 2 3 4 5 Total 𝝌𝟐 -test to test the goodness of fit:

Frequencies 3 6 24 26 4 1 N=64
𝝌𝟐 − 𝑪𝒉𝒊 𝒔𝒒𝒖𝒂𝒓𝒆
Let us try to fit a binomial distribution to this data. Let 𝑜1 , 𝑜2 , 𝑜3 , … … . . 𝑜𝑛 be the observed
frequencies and 𝑒1 , 𝑒2 , 𝑒3 , … … . . 𝑒𝑛 be the
𝑃 𝑥 = 𝑛𝐶𝑥 𝑝 𝑥 𝑞𝑛−𝑥 corresponding expected frequencies such that
𝑛 𝑛
𝑖=1 𝑜𝑖 = 𝑁 = 𝑖=1 𝑒𝑖 where 𝑁 is the number of
Here 𝑛 = 5 ∴ 𝑃 𝑥 = 5𝐶𝑥 𝑝 𝑥 𝑞𝑛−𝑥 members in the population.
1
As coins are unbiased, we have 𝑝 = 2 = 𝑞 Suppose we intend to test the null hypothesis-

Expected 𝐻0 : The theoretical frequency distribution is a good

Observed fit to the observed frequency distribution
𝑥 𝑃 𝑥 frequencies
frequencies
64 𝑃 𝑥
1 1 Against the alternative hypothesis
0 𝑃 0 = 64 × =2 3
32 32
𝐻1 : The theoretical frequency distribution is not a
5 5
1 𝑃 1 = 64 × = 10 6 good fit to the observed frequency distribution.
32 32
10 10
2 𝑃 2 = 64 × = 20 24 To test 𝐻0 against 𝐻1 , Chi-square test of goodness of
32 32
fit is applied.
10 10
3 𝑃 3 = 64 × = 20 26
32 32 Here, the test statistic is
5 5
4 𝑃 4 = 64 × = 10 4 𝑛
32 32 𝑜𝑖 − 𝑒𝑖 2
2
1 1 𝜒 =
5 𝑃 5 = 64 × =2 1 𝑒𝑖
32 32 𝑖=1
Total 1 64 64
Under 𝐻0 this is a chi-square variate with 𝑛 − 𝑐
degrees of freedom where 𝑛 is the number of terms
Example (2): in the 𝜒 2 (after pooling the frequencies which are
Let us fit a Poisson distribution to the less than 5 with the adjacent ones-Refer example)
following data. and c is the number of constraints.

The theoretical frequencies 𝑒𝑖 are computed

𝑥 0 1 2 3 4 Total 𝑛 𝑛
𝑓 123 59 14 3 1 200 such that 𝑖=1 𝑜𝑖 = 𝑁 = 𝑖=1 𝑒𝑖 . This is one
constraint. Apart from this, if any parameter is
estimated from the oserved distribution, every such
𝑓𝑖 𝑥 𝑖 0+59+28+9+4 100
Mean: 𝜇 = = = 200 = 0.5 estimation would be a constraint. Thus, the value of
𝑓𝑖 200
𝑐 would be one more than the number of parameters
𝑒 −𝜇 𝜇 𝑥 𝑒 −0.5 0.5 𝑥 estimated from the observed distribution.
𝑃 𝑥 = =
𝑥! 𝑥!

Page 17
 Sampling Theory

Note (1): 𝒙 𝑷 𝒙
𝒆𝒊
𝒐𝒊
2
𝜒 -test is one tailed (Right tailed). 180.𝑷 𝒙
0 0.6722 121 130
i.e. if 𝜒 2 𝑐𝑎𝑙 ≤ 𝜒 2 𝑡𝑎𝑏𝑙𝑒 then 𝐻0 is accepted 1 0.2666 48 32
2 0.0555 10 15
otherwise 𝐻0 is rejected.
3 0.0055 1 2
11 18
Note (2): 4 0 0 1
The chi-square test of goodness of fit is applicable 5 0 0 0
subject to the following conditions. Tot 1 180 180

1. The observations should be independent.

2. The total frequency N should be large. Here, the last three theoretical (expected)
3. The theoretical frequencies 𝑒𝑖 should be 5 or frequencies are less than 5. Therefore, they are
more. If any 𝑒𝑖 is less than 5, it should be pooled pooled with the adjacent ones such that, finally all
with the adjacent frequency. the frequencies are 5 or more.
4. If any parameter is estimated from the observed
After pooling, we have
distribution, corresponding to every such
estimation, one degree of freedom should be 𝒐𝒊 𝒆𝒊 𝒐𝒊 − 𝒆𝒊 𝟐 𝒐𝒊 − 𝒆𝒊 𝟐
lessened. 𝒆𝒊
130 121 81 0.6694
Problems 32 48 256 5.3333
1. The following data relates to the number of 18 11 49 4.4545
mistakes in each page of a book containing 180 Total 10.4572
pages.
No. of
5 or The test statistic is
mistakes per 0 1 2 3 4 Total
more
page
No. of pages 130 32 15 2 1 0 180 𝑛 2
2
𝑜𝑖 − 𝑒𝑖
Test whether the Poisson distribution is a good 𝜒 = = 10.4572 = 𝜒 2 𝑐𝑎𝑙
𝑒𝑖
fit to this observed distribution. 𝑖=1

Soln: 𝑛 = 3 (after pooling)

𝐻0 : Poisson distribution is a good fit to the observed
distribution 𝑐=2
𝑛 𝑛
(1 for 𝑖=1 𝑜𝑖 =𝑁= 𝑖=1 𝑒𝑖 which is common in
The alternative hypothesis is
all cases and 1 for estimating the parameter 𝜇 from
𝐻1 : Poisson distribution is not a good fit to the the observed distribution).
observed distribution

To test 𝐻0 , we fit a Poisson distribution to the data. Thus,

𝜒 2 is a chi-square variate with 𝑛 − 𝑐 = 3 − 2 = 1
𝑒 −𝜇 𝜇 𝑥 degrees of freedom.
𝑃 𝑥 =
𝑥!
Now, from chi-square distribution table, the value of
Here the parameter 𝜇 is estimated by finding the
𝜒 at 5% level of significance is 𝜒 2 0.05 = 3.84
mean from the observed distribution.

𝑓. 𝑥 72 As, 𝜒 2 𝑐𝑎𝑙 > 𝜒 2 0.05 , 𝐻0 is rejected.

𝜇= = = 0.4
𝑁 180
Conclusion:
𝑒 −0.4 0.4 𝑥
Poisson distribution is not a good fit to the observed
∴ 𝑃 𝑥 =
𝑥! distribution.

Page 18
 Sampling Theory

2. To an observed frequency distribution, binomial Thus,

distribution is fitted after estimating 𝑝 from the 𝜒 2 is a chi-square variate with 𝑛 − 𝑐 = 6 − 2 = 4
observed data. The observed and theoretical degrees of freedom.
frequencies are given below
𝑥𝑖 0 1 2 3 4 5 6 7 Tot Now, from chi-square distribution table, the value of
𝑜𝑖 3 3 17 31 28 11 1 2 96 𝜒 at 5% level of significance is 𝜒 2 0.05 = 9.49
𝑒𝑖 1 7 19 27 24 13 4 1 96
Test whether binomial distribution is a good fit.
As, 𝜒 2 𝑐𝑎𝑙 < 𝜒 2 0.05 , 𝐻0 is accepted.
Soln:

𝐻0 : Binomial distribution is a good fit. Conclusion:

Binomial distribution is a good fit.
𝒙 𝒐𝒊 𝒆𝒊
0 3 1 3. 10000 digits are randomly chosen from a
1 3 06 7 08
telephone directory and the following data is
2 17 19 obtained.
3 31 27 Digit Frequency
4 28 24 0 926
5 11 13 1 1207
6 1 4 2 1097
7 2 03 1 05 3 1066
Tot 96 96 4 1275
5 833
6 1007
The frequencies are pooled in such a way that none
7 872
of the theoretical frequencies is less than 5. 8 864
However, observed frequencies may be less than 5. 9 853
Total 10000
After pooling, we have Test whether there is equi-distribution in the
telephone director at 1% level of significance.
𝒐𝒊 𝒆𝒊 𝒐𝒊 − 𝒆𝒊 𝟐 𝒐𝒊 − 𝒆𝒊 𝟐
𝒆𝒊 Soln:
6 8 4 0.5
17 19 4 0.2105 𝐻0 : The digits are equi-distributed in the telephone
31 27 16 0.5926 directory.
28 24 16 0.6667
11 13 4 0.3077 The expected (theoretical) frequencies corresponding
3 5 4 0.2 to each of the digits should be equal.
Total 2.4775 10000
i.e. 𝑒𝑖 = 10
= 1000
𝒐𝒊 𝒆𝒊 𝒐𝒊 − 𝒆𝒊 𝟐 𝒐𝒊 − 𝒆𝒊 𝟐
The test statistic is 𝒆𝒊
926 1000 5476 5.476
𝑛 2
𝑜𝑖 − 𝑒𝑖 1207 1000 42849 42.849
𝜒2 = = 2.4775 = 𝜒 2 𝑐𝑎𝑙 1097 1000 9409 9.409
𝑒𝑖
𝑖=1 1066 1000 4356 4.356
1275 1000 75625 75.625
𝑛 = 6 (after pooling) 833 1000 27889 27.889
1007 1000 49 0.049
𝑐=2 872 1000 16384 16.384
𝑛 𝑛
(1 for 𝑖=1 𝑜𝑖 =𝑁= 𝑖=1 𝑒𝑖 which is common in 864 1000 18496 18.496
all cases and 1 for estimating the parameter 𝑝 from 853 1000 21609 21.609
the observed distribution). Total 222.142

Page 19
 Sampling Theory

The test statistic is The test statistic is

𝑛 2 𝑛 2
2
𝑜𝑖 − 𝑒𝑖 2 2
𝑜𝑖 − 𝑒𝑖
𝜒 = = 222.142 = 𝜒 𝑐𝑎𝑙 𝜒 = = 4.72 = 𝜒 2 𝑐𝑎𝑙
𝑒𝑖 𝑒𝑖
𝑖=1 𝑖=1
𝑛 = 10
𝑐=1 𝑛=4
𝑛 𝑛
(1 for 𝑖=1 𝑜𝑖 =𝑁= 𝑖=1 𝑒𝑖 ). 𝑐=1
𝑛 𝑛
Thus,
(1 for 𝑖=1 𝑜𝑖 =𝑁= 𝑖=1 𝑒𝑖 ).
𝜒 2 is a chi-square variate with 𝑛 − 𝑐 = 10 − 1 = 9 Thus,
degrees of freedom. 𝜒 2 is a chi-square variate with 𝑛 − 𝑐 = 4 − 1 = 3
Now, from chi-square distribution table, the value of degrees of freedom.
𝜒 at 1% level of significance is 𝜒 2 0.01 = 21.67
Now, from chi-square distribution table, the value of
As, 𝜒 2 𝑐𝑎𝑙 > 𝜒 2 0.01 , 𝐻0 is rejected. 𝜒 at 5% level of significance is 𝜒 2 0.05 = 7.81
Conclusion:
In the telephone dictionary, the digits are not equi- As, 𝜒 2 𝑐𝑎𝑙 < 𝜒 2 0.05 , 𝐻0 is accepted.
distributed.
Conclusion:
4. According to a theory in Genetics, the The result of the experiment supports theory.
proportion of beans of four types A, B, C and D
in a generation should be 9:3:3:1. In an 5. In order to test whether a die is biased, it is
experiment, among 1600 beans, the frequency of thrown 72 times and the results are tabulated as
beans of each of the above four types were 882, follows:
313, 287 and 118 respectively. Does the result Result of throw 1 2 3 4 5 6 Tot
support the theory? Number of
8 14 15 9 13 13 72
throws
Soln:
What is your conclusion?
𝐻0 : The result of the experiment supports the theory.
Soln:
Under 𝐻0 , the expected frequencies should be in the 𝐻0 : The die is unbiased.
ratio 9:3:3:1.
Under 𝐻0 , all the sides of the die are equiprobable.
1600 ×9
∴ 𝑒1 = = 900 Therefore, their frequencies should be equal.
9+3+3+1
So, the theoretical frequencies are
𝑒2 =
1600 ×3
= 300 72
9+3+3+1 𝑒𝑖 = = 12
6
1600 ×3
𝒙𝒊 𝒐𝒊 𝒆𝒊 𝒐𝒊 − 𝒆𝒊 𝟐 𝒐𝒊 − 𝒆𝒊 𝟐
𝑒3 = 9+3+3+1
= 300 𝒆𝒊
1 8 12 16 1.3333
1600 ×1 2 14 12 4 0.3333
𝑒4 = = 100
9+3+3+1 3 15 12 9 0.75
4 9 12 9 0.75
𝒐𝒊 𝒆𝒊 𝟐 𝟐 5 13 12 1 0.0833
𝒐𝒊 − 𝒆𝒊 𝒐𝒊 − 𝒆𝒊
𝒆𝒊 6 13 12 1 0.0833
882 900 324 0.36 Total 3.33
313 300 169 0.56 The test statistic is
287 300 169 0.56 𝑛 2
118 100 324 3.24 2
𝑜𝑖 − 𝑒𝑖
𝜒 = = 3.33 = 𝜒 2 𝑐𝑎𝑙
Total 4.72 𝑒𝑖
𝑖=1

Page 20
 Sampling Theory

𝑛=6
𝑐=1
𝑛 𝑛 The test statistic is
(1 for 𝑖=1 𝑜𝑖 =𝑁= 𝑖=1 𝑒𝑖 ). 𝑛 2
2
𝑜𝑖 − 𝑒𝑖
Thus, 𝜒 = = 3.084 = 𝜒 2 𝑐𝑎𝑙
𝑒𝑖
𝑖=1
𝜒 2 is a chi-square variate with 𝑛 − 𝑐 = 6 − 1 = 5
𝑛=4
degrees of freedom.
𝑐=1
Now, from chi-square distribution table, the value of 𝑛 𝑛
(1 for 𝑖=1 𝑜𝑖 =𝑁= 𝑖=1 𝑒𝑖 . Note that neither m
𝜒 at 5% level of significance is 𝜒 2 0.05 = 11.07
nor p is estimated).
2 2
As, 𝜒 𝑐𝑎𝑙 <𝜒 0.05 , 𝐻0 is accepted. Thus,
Conclusion: 𝜒 2 is a chi-square variate with 𝑛 − 𝑐 = 4 − 1 = 3
The die is unbiased. degrees of freedom.
Now, from chi-square distribution table, the value of
𝜒 at 5% level of significance is 𝜒 2 0.05 = 7.81
6. A survey of 64 families with 3 children each is
conducted and the number of male children in As, 𝜒 2 𝑐𝑎𝑙 < 𝜒 2 0.05 , 𝐻0 is accepted.
each family is noted. The result are tabulated as
Conclusion:
follows:
Male and female children are equiprobable.
Male children 0 1 2 3 Total
Families 6 19 29 10 64
Apply chi-square test of goodness of fit to test
Similar problems for practice
whether male and female children are
equiprobable.
1. Among 64 offsprings of a certain cross between
Soln: Guinea pigs 34 were red, 10 were black and 20
were white. According to the genetic model
𝐻0 : Male and female children are equiprobable.
these numbers should be in the ratio 9:3:4. Are
(Probability of male child is 0.5)
the data consistent with the model at 5% level?
Under 𝐻0 , to the given data, binomial distribution
Hint: 𝐻0 : Data are consistent.
can be fitted. (m=3 and p=0.5) 9 3
𝑒1 = 64 × 16 = 36; 𝑒2 = 64 × 16 = 12
4
𝑃 𝑥 = 𝑚𝐶𝑥 𝑝 𝑥 𝑞𝑚−𝑥 = 3𝐶𝑥 0.5 𝑥
0.5 3−𝑥
𝑒3 = 64 × 16 = 16; 𝜒 2 𝑐𝑎𝑙 = 1.44
𝜒 2 0.05 = 5.99 at 𝑛 − 𝑐 = 3 − 1 = 2 d.f
Expected
𝑥 𝑃 𝑥 frequencies As, 𝜒 2 𝑐𝑎𝑙 < 𝜒 2 0.05 , 𝐻0 is accepted.
64 𝑃 𝑥
0 𝑃 0 = 0.125 8 2. The following table gives the number of train
1 𝑃 1 = 0.375 24 accidents in a country that occurred during the
2 𝑃 2 = 0.375 24 various days of the week. Find whether the
3 𝑃 3 = 0.125 8 accidents are uniformly distributed over the
Total 1 64 week.
Hint: 𝐻0 : Accidents are uniformly .
If the accidents are to be uniformly distributed it is
𝒐𝒊 𝒆𝒊 𝟐 𝟐
𝒐𝒊 − 𝒆𝒊 𝒐𝒊 − 𝒆𝒊 126
expected that = 18 accidents happen per day.
7
𝒆𝒊
6 8 4 0.5 ∴ 𝑒𝑖 = 18 for all the days of the week.
19 24 25 1.042 𝜒 2 𝑐𝑎𝑙 = 9.78
29 24 25 1.042 𝜒 2 0.05 = 12.59 at 𝑛 − 𝑐 = 7 − 1 = 6 d.f
10 8 4 0.5 As, 𝜒 2 𝑐𝑎𝑙 < 𝜒 2 0.05 , 𝐻0 is accepted.
Total 3.084

Page 21
 Sampling Theory

3. Five coins are tossed 320 times. The number of After pooling, we have
heads observed is given below. Examine
whether the coin is unbiased. 𝒐𝒊 𝒆𝒊
273 236.4
Hint: 𝐻0 : The coin is unbiased (p=1/2) .
70 118.2
𝑃 𝑥 = 5𝐶𝑥 0.5 𝑥 0.5 5−𝑥 30 29.5
1
𝑒1 = 320 × 32 = 10; 𝑒2 = 50 17 5.9
𝑒3 = 100; 𝑒4 = 100; 𝑒5 = 50; 𝑒6 = 10 𝜒 2 𝑐𝑎𝑙 = 46.3
𝜒 2 𝑐𝑎𝑙 = 12.5 𝜒 2 0.05 = 5.99 at 𝑛 − 𝑐 = 4 − 2 = 2 d.f
𝜒 2 0.05 = 11.07 at 𝑛 − 𝑐 = 6 − 1 = 5 d.f 𝑛 =4; 𝑐 =2
As, 𝜒 2 𝑐𝑎𝑙 > 𝜒 2 0.05 , 𝐻0 is rejected. 𝑛 𝑛
(1 for 𝑖=1 𝑜𝑖 =𝑁= 𝑖=1 𝑒𝑖 which is common in
4. A survey of 320 families with 5 children each all cases and 1 for estimating the parameter 𝜇 from
revealed the following information. the observed distribution).
No. of boys 5 4 3 2 1 0
No. of girls 0 1 2 3 4 5 As, 𝜒 2 𝑐𝑎𝑙 > 𝜒 2 0.05 , 𝐻0 is rejected.
No. of
14 56 110 88 40 12
families
Is the result consistent with the hypothesis that
male and female births are equally probable?
Hint: 𝐻0 : Male and female births are equiprobable.
(Probability of male child is 0.5)
𝑃 𝑥 = 5𝐶𝑥 0.5 𝑥 0.5 5−𝑥
𝑒1 = 10; 𝑒2 = 50; 𝑒3 = 100
𝑒4 = 100; 𝑒5 = 50; 𝑒6 = 10
𝒙𝒊 𝒐𝒊 𝒆𝒊
No. of male births
0 12 10
1 40 50
2 88 100
3 110 100
4 56 50
5 14 10
𝜒 2 𝑐𝑎𝑙 = 7.16
𝜒 2 0.05 = 11.07 at 𝑛 − 𝑐 = 6 − 1 = 5 d.f
As, 𝜒 2 𝑐𝑎𝑙 < 𝜒 2 0.05 , 𝐻0 is accepted.

5. Fit a poisson distribution for the following data

and test the goodness of fit.
x 0 1 2 3 4 5 6 Tot
f 273 70 30 7 7 2 1 390
Hint: 𝐻0 : Poisson distribution is a good fit.
𝒙𝒊 𝒐𝒊 𝒆𝒊
0 273 236.4
1 70 118.2
2 30 29.5 5.9 after
3 7 4.9 adding
4 7 17 0.6 0.3 so
5 2 0.1 that total
6 1 0 becomes
390
total 390 389.7

Page 22