Chapter 9

Test of Hypothesis

Prepared By
Dr. Mohammad Bayezid Ali
Associate Professor
Department of Finance
Jagannath University, Dhaka.
Some Conceptual Issues in Hypothesis Testing
Hypothesis: A thesis is some thing that has been proven to be
true. A hypothesis is some thing that has not yet been proven
to be true. It is some statement about a population parameter
or about a population distribution. A hypothesis is one’s
assessment about the characteristics of the a population. A
statistical hypothesis is one’s believe about the real condition
of a statistical population or statistical universe.
Test of Hypothesis: The procedure which permit us to verify
whether there is a true or significant difference between
sample mean and population mean or whether to accept or
reject the hypothesis is known as test of hypothesis.
Methodology of Hypothesis Testing: Until and unless it is
verified deploying an appropriate statistical test, we should
always accept the null hypothesis that means prior to getting
test result we are to assume that the sample represents the
entire population.
 Types of Hypothesis
There are two types of hypothesis:
i. Null Hypothesis (Ho): A maintained hypothesis that
is held to be true unless sufficient evidence to the
contrary is obtained. For example: the hypothesis
which asserts that there is no significant difference
between sample mean and population mean is
known is known as null hypothesis.

ii. Alternative Hypothesis (Ha): A hypothesis against

which the null hypothesis is tested and which will be
held to be true if the null is held false. For example:
the hypothesis which asserts that there is a
significant difference between sample mean and
population mean is known as alternative hypothesis.
 Four Different Cases in Hypothesis Testing
Decision on Null States of Nature
Hypothesis (Ho) Null hypothesis Null hypothesis
(Ho) is true (Ho) is false
Fail to reject or Case= 1 Case= 2
Accept Ho Correct Decision Wrong Decision = Type
Pr obality (1   ) II Error; Pr obality 
Level of Confidence
Reject Ho Case= 3 Case= 4
Wrong Decision = Type Correct Decision
I Error; Pr obality  Pr obality (1   )
Level of Significance Power of the Test
 Concept Review
Type I Error: The rejection of a true null hypothesis
which is alternatively known as alpha ( ) risk.
Type II Error: The failure to reject false null hypothesis
which is alternatively known as beta (  ) risk.
Significance Level: The probability of rejecting a true
null hypothesis. In maximum cases, it is 5 percent.
Confidence Test: The probability of rejecting a false null
hypothesis. In maximum cases, it is 95 percent.
Sometimes it is called power of the test.
 Steps in Hypothesis Testing
1. Identification of Management queries: It deals with what the
management want to know.
2. Establishing Null Hypothesis (Ho): This stage deals with designing of
null hypothesis which is to be tested.
3. Determine the level of significance. Sometime, appropriate instruction
has been given or in the absence of any instruction, use 5 percent
level.
4. Selecting or choosing the appropriate test. For large sample size (i.e.
more than 30), z-test and for small sample size (i.e. less than or
equal to 30), t-test will be applied.
5. Finding the calculated value based on appropriate test formulas.
6. Finding the critical or theoretical or table value for given degrees of
freedom at pre-specified level of significance.
7. Comparing the calculated value with critical value. If calculated value is
greater than critical value, reject Ho. But, if calculated value is lesser
than critical value, accept Ho.
8. Explain the decision based on the result and draw conclusions on
whether to accept or reject the null hypothesis.
 Types of Test
1. Z-test
2. T-test
3.  -test
2

4. F-test

z-test and t-test are used for the same purpose and the
formula is also relatively identical. But the only difference is
that when the sample size is more than 30 then z-test will
be used but when the sample size is les than 30 then t-test
will be used.
 Decision Making Rules for Any Test

1. Calculated /Estimates Value>Table / Theoritical

/Critical Value→ Reject the Null Hypothesis
2. Calculated /Estimates Value<Table / Theoritical
/Critical Value→ Accept the Null Hypothesis
 Purposes of z-test or t-test
Both the z-test or t-test are used in the
following areas:
1. To test the difference between sample
mean ( x ) and population mean (  ) .
2. To test the difference between two sample
mean ( x1andx2 ) .
3. To test the significance of the coefficient of
correlation (  ) .
Problem:1 To test the difference between sample
mean and population mean (z-test).
The mean lifetime of a sample of 100 light tubes
produced by a company is found to be 1570 hours with
standard deviation of 80 hours. The management of the
light tubes producing company wants to know whether
the mean life time of the tubes produced by the
company is 1600 hours.
Answer: Given, n 100
X 1570 Hours
 1600hours
s 80 Hours
 Problem: 1 Solution
1. Management Queries: Whether the mean life
time of bulbs is 1600 hours.
2. Null Hypothesis (Ho): There is no significance
difference between sample mean and population
mean.
3. Suppose the level of significance is ( ) 0.05.
4. As the sample size is more than 30 , we have to
apply z-test.
 Problem: 1 Solution cont..
x   1570  1600
5. The calculated value is z
s

80
 3.73

n 1 100  1
6. At 5 percent level of significance, the table value or
critical value is 1.96.
7. As the calculated value falls outside the critical or
acceptable region, so the test fails to accept the null
hypothesis
Critical Critical
Value=-1.96 Value=+1.96

Unlikely 0.475 Unlikely

0.475
outcomes: outcomes:
p = 0.025 p = 0.025

1600 hours
 Problem: 1 Solution cont..

8. Failing to accept null hypothesis (Ho)

implies that there is statistically significant
difference between sample mean and
population mean of the life time of tube
bulbs.
 Exercise for Practice
A sample of 900 has a mean 3.47. can it be
reasonably regarded as a sample drawn
from a normal population with mean 3 and
standard deviation 2.7.
Problem: 2 To test the difference between sample
mean and population mean (t-test).
10 oil gallons are taken at random from an atomic filling
machine. The mean weight of the gallon is 15.8 kg and an
standard deviation of 0.50 kg. Does the sample mean
weight differs significantly from the expected weight of
16.0 kg.
Ans: Given,
n 10
X 15.8kg
 16.0kg
s 0.50kg
 Problem: 2 Solution
• Management Queries: Whether the mean weight
of the oil gallons is 16.0 kg.
• Null Hypothesis (Ho): There is no significant
difference between sample mean (i.e.15.8 kg)
and population mean (i.e.16.0 kg).
• Suppose the level of significance is ( ) 0.05.
• As the sample size is less than 30 , we have to
apply t-test.
 Problem: 2 Solution cont..
x   15.8  16.0
5. The calculated value is t   1.2
s 0.50
n 1 10  1
6. At 5 percent level of significance and (n-1)=(10-1)
i.e. 9 degrees of freedom, the table value or critical
value is  2.26.
7. As the calculated value falls within the critical or
acceptable region, so the test accepts the null
hypothesis. Calculated
Value= -1.2
Critical
Critical Value=+2.26
Value=-2.26

Unlikely outcomes: Unlikely outcomes:

p = 0.025 p = 0.025
0.475 0.475

16kg
 Problem: 2 Solution cont..

8. Accepting null hypothesis (Ho) implies that

there is no statistically significant difference
between sample mean (i.e. 15.8 kg) and
population mean (i.e. 16.0 kg) of weight of
oil gallons.
 Exercise for Practice
A sample of 25 has a mean 35. Can we think
that this sample is appropriate of the
population with mean 37 and standard
deviation 15.
 Problem: 3 To test the difference between two
sample mean (z-test).
A group of 40 sales men and a group of 50 sales girls
are subjected to intelligence test. the mean score is 74
and 78 respectively and the two groups with standard
deviation is 8 and 7. Test whether there is any difference
between the intelligence of the two groups.
Ans: Given,
n1 40 n2 50
X 1 74 X 2 78
s1 8 s2 7
 Problem: 3 Solution
1. Management Queries: Whether there is any
significant difference between the intelligence of
two sample group.
2. Null Hypothesis (Ho): There is no significant
difference between two sample mean (i.e. 74 and
78).
3. Suppose the level of significance is ( ) 0.05.
4. As the sample size for both of the group is more
than 30 , we have to apply z-test.a
 Problem: 3 Solution cont..
x1  x2 74  78
z   2.49
5. The calculated value is s1
2
s2
2
8 2
7 2
 
n1 n2 40 50
6. At 5 percent level of significance, the table value or
critical value is  1.96.
7. As the calculated value falls outside the critical or
acceptable region, so the test rejects the null
hypothesis. Table
Value=-1.2

Critical
Critical Value= +1.96
Value=-1.96

Unlikely outcomes: Unlikely outcomes:

p = 0.025 p = 0.025
0.475 0.475
 Problem: 3 Solution cont..

8. So, the two groups of 40 salesmen and 50

sales girls drawn from the population with
mean 74 and 78 and standard deviation 8
and 7 respectively regard significantly at 5
percent level of significance
 Problem: 3 To test the difference between
two sample mean (t-test).
Two groups consisting of 16 and 14 regional sales
managers were subjected to IQ test selected randomly
from two divisions. The mean score were 107 and 112 and
standard deviation 10 and 8 respectively. Test whether
there is any significant difference between the IQ scores of
two groups.
Given:
n1 16 n2 14
X 1 107 X 2 112
s1 10 s2 8
 Problem: 3 Solution
1. Management Queries: Whether there is any
significant difference between the IQ’s of two
sample group.
2. Null Hypothesis (Ho): There is no significant
difference between two sample mean (i.e. 107
and 112).
3. Suppose the level of significance is ( ) 0.05.
4. As the sample size for both of the group is less
than 30 , we have to apply t-test.
 Problem: 4 Solution cont..
x1  x2
5. The calculated value is t  where
1 1
S 
n1 n2
2 2
(n1  1) s1  (n2  1) s2 (16  1)(10) 2  (14  1)(8) 2
S  9.13
n1  n2  2 16  14  2

107  112
t   1.5
1 1
9.13 
16 14
6. At 5 percent level of significance and (n1  n2  2) (16  14  2) 28
Degrees of freedom the table value is  2.048.
 Problem: 4 Solution cont..
7. As the calculated value falls within the critical
or acceptable region, so the test accepts the
null hypothesis.
8. Accepting the null hypothesis implies that two
groups of 16 and 14 from the population with
mean 107 & 112 and standard deviation 10 &
8 is not regard significantly at 5 percent level
of significance and 28 degrees of freedom.
 Problem: 4 To test the significance of
the coefficient of correlation
The correlation coefficient between sales and
profit during the last 11 years (1995 – 2005)
is 0.90. Test whether the relationship is
valid.
Given:  0.90
n 11
 Problem: 5 Solution
1. Management Queries: Whether the relationship
between sales and profit with correlation
coefficient of 0.90 is valid.
2. Null Hypothesis (Ho): There is no significant
association between sales and profit with
correlation coefficient value of 0.90.
3. Suppose the level of significance is ( ) 0.05.
4. As the sample size is less than 30 , we have to
apply t-test.
 Problem: 5 Solution cont..
5. The calculated value is

r n 2 0.90 11  2
t  6.2
2 2
1 r 1  (0.90)
6. At 5 percent level of significance and (n-2) =9 degrees of
freedom the table value is  2.262.
7. As the calculated value falls outside the critical or
acceptable region, so the test rejects the null
hypothesis.
 Problem: 5 Solution cont..
8. Rejecting the null hypothesis implies that
there is significant associationship between
sales and profit. Therefore, the degree of
correlation coefficient is valid.