Estimation and Testing of Hypothesis

Introduction: Sometimes the area of investigation is very large. It becomes

difficult to study the entire population under investigation. In such case a
smaller group called a „sample‟ is taken from the population. The necessary
data is obtained from the sample and then the results are generalized to the
entire population.

Sampling distribution of sample mean:

Consider a bag containing three tickets with 1, 3 and 5. Suppose we draw two
tickets with replacements one after the other from them. We see that sample
mean (𝑥 ) changes as the sample changes and thus, we can consider sample
mean as the random variable. The mean and standard deviation of sample
mean are calculated as follows.
𝑥
Mean of 𝑥 = and
𝑛

(𝑥 ) 2 𝑥 2
S.D. of 𝑥 = −
𝑛 𝑛
Let us take the sample for the above data and find the mean and S.D.
Sample no. No. on the ticket Sample mean (𝑥 )2
in the sample
1 1,1 1 1
2 1,3 2 4
3 1,5 3 9
4 3,1 2 4
5 3,3 3 9
6 3,5 4 16
7 5,1 3 9
8 5,3 4 16
9 5,5 5 25
𝑥 = 27 (𝑥 )2 = 93
Thus, for the sample,
𝑥 27
Mean of 𝑥 = = =3
𝑛 9

(𝑥 ) 2 𝑥 2 93 27 2 93 2 31 4 2
S.D. of 𝑥 = − = − = − 3 − 9 = =
𝑛 𝑛 9 9 9 3 3 3

Next, consider the population 1, 3 and 5 and let us find the mean and S. D.
We have, n = 3, 𝑥 = 9 and 𝑥 2 = 35
𝑥 9
Population mean = = = = 3 and
𝑛 3

𝑥 2 𝑛 2
S.D. of Population = –
𝑛 𝑛
 35
= – (3)2
3

35
= – 9
3

35 − 27
= 3

8
= 3

4∗ 2
= 3

2
= 2∗
3

Note that men of 𝑥 = 3 = Men of the population.

2 1 2∗ 2 𝑆.𝐷.𝑜𝑓 𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑆.𝐷 .𝑜𝑓 𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛
S.D. of sample, 𝑥 = = ∗ = =
3 2 3 2 𝑠𝑎𝑚𝑝𝑙𝑒 𝑠𝑖𝑧𝑒

Thus, in general,

For population mean = 𝜇

Population S. D. = 𝜎 and
Sample size, = 𝑛 , we have,
1. Mean of sample mean (𝑥 ) = 𝜇
𝜎
2. S.D. of (𝑥 ) = 𝑛

If the population values follows Normal distribution with mean 𝜇 and

𝜎
S.D. = 𝜎 then distribution of 𝑥 IS ALSO Normal with mean 𝜇 and SD = 𝑛
Note:
1. For a distribution with very large population and sample size 𝑛 ≥ 30, the
distribution is considered as a Normal distribution by a “Central Limit
theorem”.
2. If we considered mean distribution as a Normal distribution than, s.n.v. z
can be written as,
𝑥 − 𝜇
𝑧= 𝜎
𝑛

Definition:
The distribution of 𝑥 is known as it sampling distribution.
The standard deviation of 𝑥 is known as its „Standard Error‟.

Note: We solve the following assuming that the 𝑥 behave as normal.

Ex.1. A large population has a mean height of 150 cms and a std de viation of
20 cm. A random sample of size 100 is taken from this population. Find
the probability that the sample mean will
(i) Exceed 151 cms
(ii) lie between 148 cm and 155 cm.
Assume that the population of height is normal and that the sampling is
with replacement.
(Given for s.n.v. Z area between (i) Z = 0 to Z = 0.5 is 0.1915
(ii) Z = 0 to Z = 1 is 0.3413
(iii) Z = 0 to Z = 2.5 is 0.4938 )
Soln: Let 𝑥 be the variate that follows normal distribution.
We have,
𝜇 = 150
𝜎 = 20
𝑛 = 100
For normal distribution, s.n.v. z is given by,
𝑥 −𝜇 𝑥 − 1570 𝑥 − 150
𝑧= 𝜎 = 20 = 2
𝑛 100

To find: 𝑃(𝑥 𝑒𝑥𝑐𝑒𝑒𝑑𝑠 151 ) i.e. P(𝑥 > 151)

Now , when 𝑥 = 151, s.n.v. Z is given by,
151 − 150
𝑧= 2
1
=
2
= 0.5
∴ 𝑃(𝑥 > 151) = 𝑃(𝑧 > 0.5)
= Area to the right of z = 0.5
= 0.5 - Area(bet z = 0 to z = 0.5)
= 0.5 – 0.1915
= 0.3085
Next,
To find: 𝑃(148 < 𝑥 < 155)
Now , when 𝑥 = 148, s.n.v. Z is given by,
148 − 150
𝑧= 2
−2
= 2
= −1
Next, when 𝑥 = 155, s.n.v. Z is given by,
155 − 150
𝑧= 2
5
= 2
 = −2.5
∴ 𝑃(148 < 𝑥 < 155) = 𝑃(−1 < 𝑧 < 2.5)
= Area to the left of z = - 1 to z = 2.5
= Area ( z= -1 to z = 0) + Area ( z = 0 to z = 2.5)
= Area ( z = 0 to z = 1) + Area ( z = 0 to z = 2.5)
= 0.3413 + 0.4938
= 0.8351

Ex.2. The mean life of a large set of fluorescent tubes is 1570 hour with a
standard deviation of 150 hours. A sample of 100 tubes is drawn from it
with replacement. Find the probabilities that the mean life of these tubes
will-
(i) exceed 1600 hrs.
(ii) not exceed 1540 hrs.
(Given: For s.n.v. z, area between (i) z = 0 to z = 2 is 0.4772
(ii) z = 0 to z = 1.33 is 0.482)
Soln: Let 𝑥 be the variate that follows normal distribution.
We have,
𝜇 = 1570
𝜎 = 150
For normal distribution, s.n.v. z is given by,
𝑥 −𝜇 𝑥 − 1570 𝑥 − 1570
𝑧= 𝜎 = 150 = 15
𝑛 100

To find: 𝑃(𝑥 > 1600)

Now , when 𝑥 = 1600, s.n.v. Z is given by,
1600 −1570
𝑧= 15
30
=
15
= 2
∴ 𝑃(𝑥 > 1600) = 𝑃(𝑧 > 2)
= Area to the right of z = 2
= 0.5 - Area(bet z = 0 to z = 2)
= 0.5 – 0.4772
= 0.0228
Next,
To find: 𝑃(𝑥 < 1540)
Now , when 𝑥 = 1540, s.n.v. Z is given by,
1540 −1570
𝑧= 15
−30
= 15
 = −2
∴ 𝑃(𝑥 < 1540) = 𝑃(𝑧 < −2)
= Area to the left of z = -2
= Area to the right of z = 2
= 0.5 - Area(bet z = 0 to z = 2)
= 0.5 – 0.4772
= 0.0228
Sampling Distribution of Sample Proportion:
Suppose 15 % of the adults staying in a city wear glasses. A sample of 100
adults is taken and 14 % of them were found wearing glasses. Than the
proportion of adults who wear glasses in the city is 15/100. It is called
“Population Proportion” (of adults who wear glasses) and is denoted by P.
Similarly the proportion of adults wearing glasses from the sample is
14/100. It is known as “sample proportion” and is denoted by p.
Observe that, as sample varies, the sample proportion p varies and
therefore p can be considered to be a random variable.
Result: Let,
P = Population proportion
p = Sample proportion
n = Sample size
than
1. Mean of 𝑝 = 𝑃
𝑃𝑄
2. Std. deviation of 𝑝 = where 𝑄 = 1 − 𝑃
𝑛

Note:
1. Sample proportion (p) in general follows Binomial Distribution
2. For a large value of n it can be approximated by Normal Distribution
3. The distribution of p is called the “Sampling distribution” and the std
deviation of p is called the “Standard Error”.
4. Standard Normal Variate (s.n.v.) for p is given by,
𝑝−𝑃
𝑧=
𝑃𝑄
𝑛

Ex.1. 20 % of a large consignment of apples are found to be bad. Find the

probability that at least 25% apples are bad in a sample of size 400 drawn
from it. (Given: for s.n.v. z, area between (i) z = 0 to z = 2.5 is 0.4938)
Soln: Given that, for bad apples,
𝑃 = 0.2
∴ 𝑄 = 1 −𝑃
= 0.8
𝑛 = 400
Note that the sample size is large and thus, we assume that the sample
proportion of p follows Normal distribution.
Note that, the mean of p = P = 0.2
𝑃∗𝑄 0.2 ∗ 0.8 0.16
∴ S.D. of P = Standard Error, 𝑆. 𝐸. = = = = 0.02
𝑛 400 400

Now,
When p = 25% = 0.25
Therefore, s.n.v. z is given by,
𝑝− 𝑃
𝑧=
𝑃 ∗𝑄
𝑛
0.25 − 0.2
= 0.02
0.05
= 0.02
= 2.5
∴ Prob(p>0.25) = P(z > 2.5)
= Area to the right of 2.5
= 0.5 – 0.4938
= 0.0062
Confidence interval and confidence limits:
For large proportion:
Let a sample size n be drawn from a large population with mean 𝜇 and
standard deviation 𝜎. Let 𝑥 denote the sample mean of the sample drawn.
Than s.n.v. z is given by ,
𝑥–𝜇
𝑧= 𝜎
𝑛

We know that area under the s.n.v. between 𝑧 = ±1.96 is 0.95

In other words,
Prob ( -1.96 < z < 1.96 ) = 0.95
𝑥–𝜇
i.e. 𝑃 −1.96 < 𝜎 < 1.96 = 0.95
𝑛
𝜎 𝜎
i.e. 𝑃 −1.96 ∗ < 𝑥 – 𝜇 < 1.96 ∗ = 0.95
𝑛 𝑛
𝜎 𝜎
i.e. 𝑃 −1.96 ∗ < 𝑥 – 𝜇 < 1.96 ∗ = 0.95
𝑛 𝑛
𝜎 𝜎
i.e. 𝑃 −𝑥 − 1.96 ∗ < – 𝜇 < − 𝑥 + 1.96 ∗ = 0.95
𝑛 𝑛
𝜎 𝜎
i.e. 𝑃 𝑥 − 1.96 ∗ < 𝜇 < 𝑥 + 1.96 ∗ = 0.95 (a<x<b then –a>-x>-b)
𝑛 𝑛
𝜎 𝜎
Hence between 𝑥 − 1.96 ∗ and 𝑥 , +1.96 ∗ , we find middle 95 % of the
𝑛 𝑛
distribution.
𝜎 𝜎
Here, 𝑥 − 1.96 ∗ and 𝑥 + 1.96 ∗ are called 95% Confidence limit for the
𝑛 𝑛
𝜎 𝜎
population mean ( 𝜇 ) and 𝑥 − 1.96 ∗ , 𝑥 + 1.96 ∗ is called 95%
𝑛 𝑛
Confidence interval.

On the similar lines we can calculate the 99 % confidence limit and confidence
interval interval. The procedure can be repeated for population proportion. The
different limits and intervals are written below in the tabular form as follows:
𝑥–𝜇
1) For Population mean, 𝑧 = 𝜎
𝑛

l.o.s (𝛼) Confidence limits Confidence interval

5 % (0.05) 𝑥 − 1.96 ∗
𝜎
and 𝑥 + 1.96 ∗
𝜎 𝜎 𝜎
𝑛 𝑛 𝑥 − 1.96 ∗ , 𝑥 + 1.96 ∗
𝑛 𝑛
1 % (0.01) 𝑥 − 2.58 ∗
𝜎
and 𝑥 + 2.58 ∗
𝜎 𝜎 𝜎
𝑛 𝑛 𝑥 − 2.58 ∗ , 𝑥 + 2.58 ∗
𝑛 𝑛
Certainity 𝑥 −3∗
𝜎
and 𝑥 + 3 ∗
𝜎 𝜎 𝜎
𝑛 𝑛 𝑥 −3∗ , 𝑥 +3 ∗
𝑛 𝑛
 𝑝 –𝑃
2) For Population mean, 𝑧 =
𝑃𝑄
𝑛

l.o.s (𝛼) Confidence limits Confidence interval

5 % (0.05) 𝑃𝑄 𝑃𝑄 𝑃𝑄 𝑃𝑄
𝑝 − 1.96 ∗ and 𝑝 + 1.96 ∗ 𝑝 − 1.96 ∗ , 𝑝 + 1.96 ∗
𝑛 𝑛 𝑛 𝑛

1 % (0.01) 𝑃𝑄 𝑃𝑄 𝑃𝑄 𝑃𝑄
𝑝 − 2.58 ∗ and 𝑝 + 2.58 ∗ 𝑝 − 2.58 ∗ , 𝑝 + 2.58 ∗
𝑛 𝑛 𝑛 𝑛

Certainity 𝑃𝑄 𝑃𝑄 𝑃𝑄 𝑃𝑄
𝑝 −3∗ and 𝑝 + 3 ∗ 𝑝 −3 ∗ , 𝑝+3 ∗
𝑛 𝑛 𝑛 𝑛

3) For some problems on sample mean, Population mean 𝜇 may be given. We

need to take this 𝜇 as 𝑥
4) For some problems on sample proportion, Population proportion P should be
taken as sample proportion p. (i.e. 𝑃 = 𝑝 )

Examples: (problems based on Population mean)

Estimation of point: (Population M ean)
Ex.1. A survey of 36 married people yield a mean age of at the time of their
marriage as 26 years with a S.D. of 2.4 years. Find 95 5 Confidence limit for
the age at the time of marriage.
Soln: Given 𝑛 = 36 , 𝜎 = 2.4 years, 𝑥 = 26 years and 𝛼 = 5 %
Now, Confidence limits for population mean at 95 % are given by,
𝜎 𝜎
𝑥 − 1.96 ∗ and 𝑥 + 1.96 ∗
𝑛 𝑛
2.4 2.4
26 − 1.96 ∗ and 26 + 1.96 ∗
36 36
2.4 2.4
26 − 1.96 ∗ 6 and 26 + 1.96 ∗ 6
26 − 1.96 ∗ 0.4 and 26 + 1.96 ∗ 0.4
26 − 0.784 and 26 + 0.784
25.216 and 26.784
Thus the 95 % confidence limits are 25.216 and 26.784

Ex.2. From a random sample of 64 plastic containers manufactured by a

process, the average weight is found to be 50 gms. if the standard deviation
of the weight of the containers produced by the machine is 2 gms. Find the
limits of weights within which the average weight of the containers
manufactured by the process lie with 95 % confidence.
Soln: We proceed as follows.
Given 𝑛 = 64 ,
𝜎 = 2 gms,
𝑥 = 50 gms and 𝛼 = 5 %
Now, Confidence limits for population mean at 95 % are given by,
𝜎 𝜎
𝑥 − 1.96 ∗ 𝑛 and 𝑥 + 1.96 ∗ 𝑛
 2 2
50 − 1.96 ∗ and 50 + 1.96 ∗
64 64
2 2
50 − 1.96 ∗ 8 and 50 + 1.96 ∗ 8
50 − 0.49 and 50 + 0.49
49.51 and 50.49
Thus the 95 % confidence limits are 49.51 and 50.49

Estimation of interval: (Population M ean)

Ex.1. A sample of 50 bulbs from a large consignments showed a mean life of
52 hours with a standard deviation 4 hours. Find the confidence interval
within which the mean life of the bulbs lie almost certainly.
Soln: We Proceeds as follows:
Given that:
𝑛 = 50
𝑥 = 52 (𝜇)
𝜎=4
To find : Confidence interval for certainty.
Confidence interval for certainty is given by,
𝜎 𝜎
𝑥−3∗ 𝑛 , 𝑥+ 3∗ 𝑛
4 4
= 52 − 3 ∗ , 52 + 3 ∗
50 50
4 4
= 52 − 3 ∗ , 52 + 3 ∗ 7.07
7.07
= 52 − 1.7 , 52 + 1.7
= 50.3 , 53.7
Thus the Confidence interval for certainty is 50.3 , 53.7

Exercise:
1. In a study of television viewing habits, in order to obtained an interval
estimates of the average number of hours per week that teenager spend
watching television programmes, a random sample of 100 teenage children is
taken. The sample investigation revealed that a mean of 9.2 hours with S.D.
of 3.2 hours. Obtained the desired interval of estimates with confidence
coefficient of 0.99. (8.3744, 10.0256)
2. For a given sample of 200 items drawn from a large population, the mean is
65 and the S.D. is 8. Find the 95 % Confidence limits for the population
mean. (63.8913 and 66.1087)
3. A hospital is collecting data regarding the number of days spent by a patient
in the hospital for typhoid. A sample shows the following:
No. of days spent (x) 3 4 5 6 7 8 9
No. of patients (y) 8 10 15 20 20 15 12
Find the limits within which the mean number of days required to be spent in
the hospital by a patient for typhoid lies almost certainty. (5.745 and 6.795)
Estimation of Point: (Sample proportion)
Ex.1. A random sample of 100 balls slected from a large consignment of tennis
balls gave 10 % bad balls. Find 99 % confidence limits for the percentage of
bad balls in the consignment.
Soln: We Proceeds as follows:
To find: the 99% confidence limits to the percentage of persons who
would buy the product.
Given that:
𝑛 = 100
𝑝 = 10% = 0.1
𝑃 = 𝑝 = 0.1
⟹ 𝑄 = 1 − 𝑃 = 0.9 and 𝛼 = 1 %
99% Confidence limits are given by,
𝑃𝑄 𝑃𝑄
𝑝 − 2.57 ∗ and 𝑝 − 2.58 ∗
𝑛 𝑛

0.1∗0.9 0.1∗0.9
0.1 − 2.58 ∗ and 0.1 + 2.58 ∗
100 100
0.1 − 2.58 ∗ 0.03 and 0.1 + 2.58 ∗ 0.03
0.1 − 0.0744 and 0.1 + 0.0744
0.0226 and 0.1744
i.e. 0.0226 and 0.1744
Thus, 99% confidence limits to the percentage of persons are:
2.26 and 17.44

Exercise:
1. In a sample of 1000 T.V. viewer, 340 watch a particular programme. Find
99 % confidence limits for the percentage of all viewers who watch the
programme. (30.135% and 37.865%)
2. A department store wants to determine the percentage of shoppers who
leaves only after having actually bought at least one item. A random sample
of 900 shoppers leaving the store, showed that 750 had brought something
ranging from a couple of saop bars to a complete bedroom furniture set.
What is 99% confidence interval for the true percentage of buyers?
(82%, 84%)
3. A sample of 900 days is taken from a meterological records of a certain
districts and 100 of them are found to be foggy. Determine the probable
limits for the percentage of foggy days in the district. (7.89% and 14.135).
Testing of Hypothesis:

Hypothesis: Hypothesis is the statement which is believe to be true

and an argument generally is based upon.

Null Hypothesis and Alternative Hypothesis:

Null Hypothesis: Null Hypothesis is a assumption which is believe to be true. It
is denoted by 𝐻0 .
Alternative Hypothesis: A hypothesis which is considered as true in the
absence of null hypothesis is called alternative hypothesis. It is denoted by 𝐻1 .
Note:
1. Generally, a statistical hypothesis is referred to as a Null hypothesis.
Example: 𝐻0 = 𝜇 = 𝑅𝑠. 5000/−.
2. Hypothesis other than Null hypothesis is called as Alternative hypothesis.
Example: 𝐻1 = 𝜇 ≠ 𝑅𝑠. 5000/−
3. Alternative hypothesis is the assumption which can be considered to be the
alternative to the null hypothesis.
4. If the Null Hypothesis is not true then what is true is called as Alternative
hypothesis.

Testing of Hypothesis and error in decision making:

Testing of Hypothesis:
Testing of hypothesis is the process of accepting or rejecting a null hypothesis
based on a chosen sample statistics.Any statistical hypothesis is based on the
information supplied by the sample data. The procedure of testing such
hypothesis does not guarantee that all decisions are accurate. There is always
a chance of some error. There are two types of errors.
Type I error: The process of rejecting a true hypothesis is called type I error
Type II error: The process of accepting a false hypothesis is called type II error.
The above error can be represented in a tabular form as follows:
Null Hypothesis 𝐻0 is
True Not true
Decision Reject 𝐻0 Type I error No error
Do not Reject 𝐻0 No error Type II error
Note:
1. Type I error is also called as 𝛼 error.
2. Type II error is also called as 𝛽 eoor.
Critical region:
The area of the normal curve corresponding to the type I error, is Critical region
(also known as Region of rejection)
Critical region is the region which corresponds to the test statistics for which
the hypothesis H0 is to be rejected if the sample point falls in the region. In
other words critical region is the region of rejection of null hypothesis.

Acceptance region:
The region that falls outside the region of critical region is called „Region of
acceptance‟
Level of significance:
Level of significance is define as the maximum probability of, with which we
would be willing to risk a Type I error.
[Alternatively, it is also defined as a test designed so that P(Type I Error) ≤ ∝,
then ∝ is called level of significance.] It is denoted by 𝛼.
Since ∝ is the probability value, it is always less than 1. Generally we take 𝛼
between1% to 5%. By default we take ∝ as 5% (0.05).

Steps involved in testing of hypothesis:

1. State 𝐻0 and 𝐻1
2. Specify the level of significance.
(If 𝑙. 𝑜. 𝑠. is not mentioned then we take it as 5 %)
3. Mentioned the decision criterion
4. Compute the value of test statistics.
5. Draw the conclusion

Decision criteria based on 𝑯𝟏 and :

Rejection of 𝐻1 can be easily done from the following table. The following table
takes into consideration for 𝛼 = 1% and 𝛼 = 5 %.
Alternative Decision criterion: Reject 𝐻0 iff
Hypothesis 𝐻1 𝑙. 𝑜. 𝑠. = 0.05 𝑙. 𝑜. 𝑠. = 0.01 𝑙. 𝑜. 𝑠. = 𝛼
(𝑖. 𝑒. , 𝛼 = 5%) (𝑖. 𝑒. , 𝛼 = 1%)
> 𝑧 > 1.64 𝑧 > 2.33 𝑧 > 𝑧𝛼
< 𝑧 < −1.64 𝑧 < −2.33 𝑧 < 𝑧𝛼
≠ 𝑧 < −1.96 𝑜𝑟 𝑧 < −2.58 𝑜𝑟 𝑧 > 𝑧𝛼 or
𝑧 > 1.96 𝑧 > 2.58 𝑧 > 𝑧𝛼

Note:
 1) Decision criterion is based on:
i) 𝐻1
ii) 𝑙. 𝑜. 𝑠. 𝛼 and
iii) Large Sample size n.
2) However, for a large samples, it will depend on 𝐻1 and 𝛼.
3) Here, z is the computed value of test statistics based on selected sample
/proportion.

Problems based on Sample mean:

Examples:
Ex. 1. The mean lifetime of a sample of 100 fluorescent light bulbs produced
by a company is found to be 1570 hours with a standard deviation of 120
hours. Test the hypothesis that the mean lifetime of a bulbs produced by
the company is 1600 hours against the alternative hypothesis that it is less
than 1600 hours at 5% l.o.s.
Soln: We proceed as follows:
Step 1.: 𝐻0 = the mean life of fluorescent bulbs is 1600 hours.
𝐻1 = the mean life of fluorescent bulbs is less than 1600 hours.
(i.e. 𝐻0 ∶ 𝜇 = 1600 hours
𝐻1 ∶ 𝜇 < 1600 hours )
Step 2.: 𝛼 = 0.05 (i.e. 5 %)
Step 3.: Based on 𝐻1 and 𝛼 and large value of n = 100, we take decision:
Reject 𝐻0 iff 𝑧 < −1.64
Step 4.: For a given sample, given that,
𝑥 = 1570
𝜎 = 120
𝑛 = 100
And for population, population mean, 𝜇 = 1600
Test statistics z is given by,
𝑥 −𝜇
𝑧= 𝜎
𝑛
1570 − 1600
= 120
100
−30
= 120
10
−30
= 12
= − 2.5
Step 5: Observe that the computed value of z (i.e. z = - 2.5 < -1.64.
Thus we reject H0.

Ex.2. An ambulance service claims that it takes on an average 8.9 minutes for
an ambulance to reach its destination in emergency calls. To check on this
claim, the agency which licenses ambulance service has timed them on 50
emergency calls, getting a mean of 9.3 minutes with a standard deviation
of 1.6 minutes. What can they conclude at 5 % l.o.s.
Soln: We proceed as follows:
Step 1.: 𝐻0 = An ambulance can reached its destination in 8.9 minutes.
𝐻1 = An ambulance cannot reached its destination in 9.8 minutes.
(i.e. 𝐻0 ∶ 𝜇 = 8.9 min
𝐻1 ∶ 𝜇 ≠ 8.9 min )
Step 2.: 𝛼 = 0.05 (i.e. 5 %)
Step 3.: Based on 𝐻1 and 𝛼 and large value of n = 50, we take decision:
Reject 𝐻0 iff 𝑧 < −1.96 or 𝑧 > 1.96
Step 4.: For a given sample, given that,
𝑥 = 9.3
𝜎 = 1.6
𝑛 = 50
And for population, population mean, 𝜇 = 8.9
Test statistics z is given by,
𝑥 −𝜇
𝑧= 𝜎
𝑛
9.3 − 8.9
= 1 .6
50
= 1.77
Step 5: Observe that the computed value of z (i.e. z = 1.77 not > 1.96.
Thus we do not reject H0.

Testing of hypothesis:(Sample proportion)

Ex.1. A coin is tossed at random 400 times and head turns up 240 times. Can
the coin be regarded unbiased? Use 5 % level of significance.
A certain coin showed up head on 240 occasions in 400 tosses. Test the
claim that the coin is unbiased.
Soln: We Proceeds as follows:
Step1: H0 = Selected coin is unbiased.
(The head or tail occurs in the ratio1:1
1
i.e. 𝑃 = 2 )
1
H1 = Selected coin is biased i.e. 𝑃 ≠ 2.
Step 2: Level of significance, 𝛼 = 0.05 = 5%
Step 3: Decision criteria: Based on H1 and 𝛼 (and a large value of n= 400)
we arrive at a decision criterion, reject H 0:
if z <-1.96 or z > 1.96
Step 4: Given that
1
𝑃 = 2 = 0.5
1
⟹ 𝑄 = 1−𝑃 = = 0.5
2
n = 400
240
𝑝 = 400 = 0.6
Test statistics z is given by,
 𝑝 −𝑃
𝑧= 𝑃𝑄
𝑛
0.6 − 0.5
= 0 .5∗0.5
400
0.1
= 0 .5
20
0.1 ∗ 20
= 0.5
=4
Step 5: Observe that computed value of z = 4 >1.96.
Thus we reject H0.

Ex.2. In a sample of 400 residents of a locality, 232 are men. Test the
hypothesis that sex ratio is 1:1 at 1 % l.o.s.
Soln: We Proceeds as follows:
Step1: H0 = Sex ratio is 1:1.
(The man and woman occurs in the ratio1:1
1
i.e. 𝑃 = )
2
1
H1 = Sex ratio is not 1:1 i.e. 𝑃 ≠ 2.
Step 2: Level of significance, 𝛼 = 0.01 = 5%
Step 3: Decision criteria: Based on H1 and 𝛼 (and a large value of n= 400)
we arrive at a decision criterion, reject H 0:
if z < - 2.58 or z > 2.58
Step 4: Given that
1
𝑃 = 2 = 0.5 (P = proportion of men in the population)
1
⟹ 𝑄 = 1 − 𝑃 = = 0.5
2
n = 400
232
𝑝 = 400 = 0.58 (p = proportion of men in the population)
Test statistics z is given by,
𝑝 −𝑃
𝑧 = 𝑃𝑄
𝑛
0.58 − 0.5
= 0.5 ∗0 .5
400
0.08
= 0 .5
20
0.08 ∗ 20
=
0.5
= 3.2
Step 5: Observe that computed value of z = 3.2 > 2.58.
Thus we reject H0.
References:
1. Gupta S.P. , Statistical Methods, Sultan Chand and sons.
2. Gupta C.B., Fundamentals of Statistics, Himalaya Publishing House
3. Shah R. J., Statistical Methods.