Statistics for Management Unit 9

Unit 9 Testing of Hypothesis in Case of Large

and Small Samples
Structure:
9.1 Introduction
Objectives
Relevance
Assumptions
9.2 Testing Hypothesis
Null and Alternate hypothesis
Interpreting the level of significance
Hypothesis are accepted and not proved
9.3 Selecting a Significance Level
Preference of type I error
Preference of type II error
Determine appropriate distribution for the test of Mean
9.4 Two–tailed Tests and One–tailed Tests for Mean
Case study on Two–tailed and One-tailed tests
9.5 Classification of Test Statistics
Statistics used for testing of hypothesis
Test procedure
How to identify the right statistics for the test
9.6 Testing of Hypothesis in the Case of Small Samples
9.7 ‘t’ Distribution
Uses of ‘t’ test
9.8 Summary
9.9 Glossary
9.10 Terminal Questions
9.11 Answers
9.12 Case Study

9.1 Introduction
In the previous unit, estimation, we have studied about the estimation of the
parameter from the samples and the methods of estimation. In this unit,
Testing of hypothesis, we will study about hypothesis and the testing of
hypothesis. Estimation is about estimating the parameters and finding out

Manipal University Jaipur Page No. 359

Statistics for Management Unit 9

the confidence intervals. Hypothesis testing is the opinion about the

population parameter that may or may not be in the confidence interval
derived from the sample. Hypothesis testing is helpful in decision making.
Before starting this unit, let’s refresh the concepts we have studied on
estimation.
Hypothesis testing begins with an assumption, called hypothesis that we
make about a population parameter wherein we assume a certain value for
the population parameter. To test the validity of our assumption, we gather
sample data and determine the difference between the hypothesised value
and the actual value of the sample statistic. Then we judge whether the
difference is significant.
The smaller the difference, the greater the likelihood that our hypothesised
value for the parameter is correct. The larger the difference, the smaller the
likelihood that our hypothesised value for the parameter is correct.
Unfortunately, the difference between the hypothesised population
parameter and the actual statistic is more often; neither so large that we
automatically reject our hypothesis, nor so small that we just as quickly
accept it. So in hypothesis testing, as in most significant real-life decisions,
clear-cut solutions are the exception, not the rule.
Objectives:
After studying this unit, you should be able to:
 describe the basic concepts of testing hypothesis
 describe the different test statistics available
 identify the test for a given problem
 identify the type of errors
9.1.1 Relevance
Caselet
You need to be objective
The government in a certain country says that radiation levels in the area
surrounding a nuclear power plant are well below levels considered harmful.
Three people in the area died of leukemia. The local people immediately put
the blame on the radioactive fallout. Does the death of three people make
us assume that the government is wrong with its information and that we
make assumption or hypothesis, that radiation levels in the area are
Manipal University Jaipur Page No. 360
Statistics for Management Unit 9

abnormally high? Alternatively, do we accept that the deaths from leukemia

are random and not related to the nuclear power facility? You should not
accept or reject a hypothesis about a population parameter- in this case the
radiation levels in the surrounding area of the nuclear power plant, simply by
institution. You need to be objective in decision making. For this situation an
appropriate action would be to take samples of the incidence of leukaemia
cases over a reasonable period of time and use these to test the hypothesis.
The purpose of this unit is to find out how to use hypothesis testing to
determine if a claim is valid.
(Source: Derek L Waller published by Elsevier Inc Ed 2008)

9.1.2 Assumptions
Although hypothesis testing sounds like some formal statistical term and
completely unrelated to business decision making, in fact, managers
propose and test hypothesis all the time. For example, “if we drop the price
of this car model by Rs.1,500, we will sell 50,000 cars this year” is a
hypothesis. To test this hypothesis, total car sales till the end of the year
have to be counted.
Managerial hypothesis are based on intuition; the marketplace decides
whether the manager’s intuitions were correct. Hypothesis testing is about
making inferences about a population from only a small sample. The bottom
line in hypothesis testing is when we ask ourselves (and then decide)
whether a population, like this one, would be likely to produce a sample like
the one we are looking at.

9.2 Testing Hypothesis

9.2.1 Null and Alternate hypothesis
In hypothesis testing, we must state the assumed or hypothesised value of
the population parameter before we begin sampling. The assumption we
wish to test is called the null hypothesis and is symbolised by ’H0’.

Example 1
We want to test the hypothesis, that the population mean is equal to 500.
We would symbolise it as follows and read it as,
The null hypothesis is that the population mean = 500 which is written as,

 0 :   500

Manipal University Jaipur Page No. 361

Statistics for Management Unit 9

From agricultural and medical applications of statistics, the term ‘null

hypothesis’ is derived. In order to test the effectiveness of a new fertilisers
or drugs, the tested hypothesis (the null hypothesis) was that it had no
effect, that is, there was no difference between treated and untreated
samples. If we use a hypothesised value of a population mean in a problem,
we would represent it symbolically as ‘0’. This is read as - ‘The
hypothesised value of the population mean’.
If our sample results fail to support the null hypothesis, we must conclude
that something else is true. Whenever we reject the hypothesis, the
conclusion we do accept is called the alternative hypothesis and is
symbolised as ‘H1.
For the null hypothesis H0:  = 200, we will consider three alternative
hypotheses:
H1:   200 (population mean is not equal to 200)
H1:  > 200 (population mean greater than 200)
H1:  < 200 (population mean less than 200)
Example 2
If we want to test the success rate of a particular treatment, we make null
hypothesis for success rate ‘p’ (for the test value of 0.99) as:
 0 : p  0.99 and alternative hypothesis is among
 1 : p  0.99
 1 : p  0.99
 1 : p  0.99

Example 3
If we want to test if the attribute of educational qualification has any
influence on the income of an individual, we make null hypothesis as:
 0 : Educational qualification has no influence on the income of an
individual
and alternative hypothesis is
1 : Educational qualification has an influence on the income of the
individual

Manipal University Jaipur Page No. 362

Statistics for Management Unit 9

9.2.2 Interpreting the level of significance

The purpose of hypothesis testing is not to question the computed value of
the sample statistic but to make a judgment about the difference between
that sample statistic and a hypothesised value for population parameter.
The next step after stating the null and alternative hypothesis; is to decide
what criterion to be used for deciding whether to accept, or reject the null
hypothesis. If we assume the hypothesis is correct, then the significance
level will indicate the percentage of sample statistic that is outside certain
limits. (In estimation, the confidence level indicates the percentage of
sample statistic that falls within the defined confidence limits).
9.2.3 Hypothesis are accepted and not proved
Even if our sample statistic does fall in the non-shaded region, this does not
prove that our null hypothesis (H0) is true; it simply does not provide
statistical evidence to reject it. Why? It is because the only way in which the
hypothesis can be accepted with certainty is for us to know the population
parameter; unfortunately, this is not possible.
Therefore, whenever we say that we accept the null hypothesis, we actually
mean that there is no sufficient statistical evidence to reject it. Use of the
term accept, instead of do not reject, has become a standard practice. It
means that when sample data do not suggest us to reject a null hypothesis,
we believe that the hypothesis is true. Figure 9.1 depicts the non shaded
region that makes up 95 percent of the area under the curve.

Fig. 9.1: Acceptance and Rejection Region of Sample

Manipal University Jaipur Page No. 363

Statistics for Management Unit 9

9.3 Selecting a Significance Level

There is no single standard or universal level of significance for testing
hypothesis. In some instances, a 5% level of significance is used. In the
published results of research papers, researchers often test hypothesis at
the 1 percent level of significance. Hence, it is possible to test a hypothesis
at any level of significance. However, remember that our choice of the
minimum standard for an acceptable probability, or the significance level, is
also the risk we assume of rejecting a null hypothesis when it is true.
The higher the significance level we use for testing a hypothesis, the higher
the probability of rejecting a null hypothesis when it is true. 5% level of
significance implies we are ready to reject a true hypothesis in 5% of cases.
If the significance level is high then we would rarely accept the null
hypothesis when it is not true but, at the same time, often reject it when it is
true.
When testing a hypothesis we come across four possible situations. Table
9.1 depicts the four possible situations.
Table 9.1: Possible Situations when Testing a Hypothesis
Hypothesis is
True False
Test results says Accept Type II error
Reject

Type I error
The combinations are:
 If null hypothesis is true, and the test result make us to accept it, then
we have made a right decision.
 If null hypothesis is true, and the test result make us to reject it, then we
have made a wrong decision (Type I error). It is also known as
consumer’s risk, denoted by .
 If hypothesis is false, and the test results make us to accept it, then we
have made a wrong decision (Type II error). It is known as producer’s
risk, denoted by  .1 – is called power of the test.
 If hypothesis is false, and the test result make us to reject it – we have
made a right decision.
Manipal University Jaipur Page No. 364
Statistics for Management Unit 9

9.3.1 Preference of type I error

For example, making a type I error (rejecting a null hypothesis when it is
true) involves the time and trouble of reworking a batch of chemicals that
should have been accepted. At the same time, making a type II error
(accepting a null hypothesis when it is false) means taking a chance that an
entire group of users of this chemical compound will be poisoned.
Obviously, the management of this company will prefer a type I error to a
type II error and, as a result, will set very high levels of significance in its
testing to get low ’s.
9.3.2 Preference of type II error
For example, making a type I error involves disassembling an entire engine
at the factory, but making a type II error involves relatively inexpensive
warranty repairs by the dealers. In this case, the manufacturer is more likely
to prefer a type II error and will set lower significance levels in its testing.
9.3.3 Determine appropriate distribution for the test of Mean
After deciding what level of significance to use, our next task in hypothesis
testing is to determine the appropriate probability distribution. We have a
choice between the normal distribution, and the ‘t’ distribution.
The rules for choosing the appropriate distribution are similar to those we
encountered in the unit on estimation. Later in this unit, we shall examine
the distributions appropriate for testing hypothesis about proportions.
Table 9.2 depicts when to use the normal and ‘t’ distributions in making tests
of means.

Table 9.2: Conditions for Using the Normal and ‘t’ Distributions in
Testing Hypothesis about Means
When the Population When the Population
Standard Deviation is Standard Deviation is
known not known
Sample size ‘n’ is larger Normal distribution, Normal distribution,
than 30. z–table z–table
Sample size ‘n’ is 30 or Normal distribution, ‘t’ distribution, ‘t’ table
less and we assume the z–table
population is normal or
approximately so.

Manipal University Jaipur Page No. 365

Statistics for Management Unit 9

One more rule has to be kept in mind, when testing the hypothesised values
of a mean. As in estimation, use the finite population multiplier whenever the
population is finite in size, sampling is done without replacement, and the
sample is more than five percent of the population.

9.4 Two–tailed Tests and One–tailed Tests for Mean

A two-tailed test of a hypothesis will reject the null hypothesis, if the sample
mean is significantly higher than or lower than the hypothesised population
mean. Thus, in a two-tailed test, rejection region is split in two parts under
the distribution curve.
A two-tailed test is appropriate when:
The null hypothesis is  = 0 (where 0 is some specified value)
The alternative hypothesis is   0.
One-tail test is appropriate in each of the following cases:
1 :    0
1 :    0

9.4.1 Case study on Two-tailed and One-tailed tests

Case Study
Let’s assume that a manufacturer of light bulbs wants to produce bulbs
with a mean life of:
   0  1000 hours

If the lifetime is shorter, he/she will lose customers to his/her competitors;

if the lifetime is longer, he/she will have a very high production cost
because the filaments will be excessively thick.
In order to see whether the production process is working properly, a
sample of the output is taken to test the hypothesis,
 0 ;   1000
A two-tailed test is used because he/she does not want to deviate
significantly from 1,000 hours in either direction, the appropriate
alternative hypothesis is:
1 :   1000
Therefore, the null hypothesis is rejected, if the mean life of bulbs in the
sample is either too far above 1,000 hours or too far below 1,000 hours.

Manipal University Jaipur Page No. 366

Statistics for Management Unit 9

However, there are situations in which a two-tailed test is not appropriate,

and we must use a one-tailed test.

Case Study (contd.)

Consider the case of a wholesaler who buys light bulbs from the
manufacturer discussed earlier. The wholesaler buys bulbs in bulk and
does not want to accept a lot of bulbs unless their mean life is at least
1,000 hours. As each shipment arrives, the wholesaler tests a sample to
decide whether the shipment should be accepted. The company will
reject the shipment only if the wholesaler feels that the mean life is below
1,000 hours. If the bulbs are better than expected (with a mean life above
1,000 hours), the shipment will not be rejected because the longer life
comes at no extra cost.
So, the wholesaler’s hypothesis is:

Ho:  = 1,000 and H1:  < 1,000 hours.

The wholesaler rejects ‘H0’ only if the mean life of the sampled bulbs is
significantly below 1,000 hours. Figure 9.2 depicts why this test is called
a left-tailed test (or a lower-tailed test).

Fig. 9.2: Left-tailed Test

In general, a left tailed (lower-tailed) test is used if the hypothesis is

Ho:  = o and H1:  < o. In such a situation, sample evidence with the
sample mean significantly below the hypothesised population leads us to
reject the null hypothesis in favour of the alternative hypothesis. Stated
differently, the rejection region is in the lower tail (left tail) of the distribution
of the sample mean, and that is why we call this a lower-tailed test.

Manipal University Jaipur Page No. 367

Statistics for Management Unit 9

A left-tailed test is one of two kinds of one-tailed tests. The other kind of
one-tailed test is a right-tailed test (or an upper-tailed test). An upper-tailed
test is used when the hypothesis is H1:  > o. Only values of the sample
mean that are significantly above the hypothesised population mean will
cause us to reject the null hypothesis in favour of the alternative hypothesis.
Figure 9.3 depicts an upper-tailed test where the rejection region is in the
upper tail of the distribution of the sample mean.

Fig. 9.3: Right-tailed Test

Tests for proportion and other parameters are similarly discussed; rejection
regions are similarly identified with reference to the given level of
significance and appropriate distribution.
In each example of hypothesis testing, when we accept a null hypothesis on
the basis of sample information, we are really saying that there is no
statistical evidence to reject it. We are not saying that the null hypothesis is
true. The only way to prove a null hypothesis is to know the exact value of
the population parameter or the population distribution and that is not
possible with just sampling. Thus, we accept the null hypothesis and behave
as if it is true simply because we can find no evidence to reject it.

Example 4
The hypothesis to be tested is Ho:  = 100, against the alternative
hypothesis H1:   100, with sample size n = 20, population standard
deviation σ =2.5. Here sample size is smaller than 30 but population
standard deviation is given; hence to test the hypothesis the probability
distribution used is ‘normal distribution’.

Manipal University Jaipur Page No. 368

Statistics for Management Unit 9

Example 5
The hypothesis to be tested is Ho:  = 10 against the alternative
hypothesis H1:  > 10, with sample size n = 20, population standard
deviation is not known. Here sample size is smaller than 20 but
population standard deviation is not given, hence to test the hypothesis
the probability distribution used is ‘t- distribution’.

Self Assessment Questions

1. For the following cases; specify which probability distribution to use in
hypothesis testing:
i. H0:  = 27, H1:   27, X = 33, sample  = 4, n = 25
ii. H0:  = 98.6, H1:  > 98.6, X = 99.1,  = 1.5, n = 50
iii. H0:  = 3.5, H1:  < 3.5, X = 2.8, sample  = 0.6, n = 18
iv. H0:  = 57, H1:  > 57, X = 65, sample  = 12, n = 42

9.5 Classification of Test Statistics

9.5.1 Statistics used for testing of hypothesis
Table 9.3a: Statistics for Testing the Hypothesis on Proportion; Large Sample Case
Test Description
Test Statistics Notes
No. of Test
1 Test for 

specified p p p = Sample proportion
proportion –
Z p = Hypothised value of
pq population proportion
infinite
population n q = 1 – p, n= sample size
2 Test for 
 p = Sample proportion
specified ( p  p)
proportion – Z p = Hypothised value of
finite  pq  Nn population proportion
   
population  n   N 1  q = 1 – p,n = Sample size
N = Population size
3 Test for 

difference in
  p1 = first sample proportion
( p1  p 2 )
proportions of Z 

two samples –     p 2 = second sample

p1 q 1 p 2 q 2
different  proportion
population n1 n2 n1 = first sample size
n2 = second sample size

Manipal University Jaipur Page No. 369

Statistics for Management Unit 9

4 Test between 

proportion –
p1 = first sample proportion
when 

populations  
p 2 = second sample
are similar (p1  p 2 ) proportion
Z n1 = first sample size
with respect to
 1 1 
a given p 0 q 0    n2 = second sample size
attribute  n1 n 2   
n p  n 2 p2
p0  1 1
n1  n 2
and q0 = 1- p0
Table 9.3b: Statistics for Testing the Hypothesis on Mean; Large Sample
Case
Test Description
Test Statistics Notes
No. of Test
5 Test for  = Population mean
specified
mean –
X = Sample mean
infinite (X  ) p = Population S.D
Z
population, p In case p is not known, we
n>30 and use s in its place calculating
n
population
variance(s)  ( X  X )2
s  i
known n 1
6 Test for  = Population mean
specified
mean –
X = Sample mean
( X  ) p = Population S.D
finite Z
p In case p is not known, we
population,
Nn
n>30 and   use s in its place calculating
population
n  N 1 
variance(s)  ( X  X )2
s  i
known n 1
7 Test for X 1 and X 2 are sample mean
difference in (X1  X 2 )
Z for the first and second
means–
  p1 2  p 2 2  samples respectively.
different   
 n1 n2  n1= first sample size
population,  
n>30 and n2 = second sample size

Manipal University Jaipur Page No. 370

Statistics for Management Unit 9

population In case  p1 and  p 2 are not

variance(s)
known known, we use  s1 and
 s 2 respectively in their
places calculating

 ( X  X )2
1i 1
 s1 = n 1
1

(X  X )2
s2 = 2i 2
n 1
2

8 Test for X 1 and X 2 are sample mean

difference in
for the first and second
means–
samples respectively.
same
population, n1= first sample size
n>30 and n2 = second sample size
population In case  p is not
variance(s)
known (X 1  X 2 ) known, we use  s12
Z
in its place calculating
2 1 1 
 p    2 2 2 2
 n1 n 2  N1 (1  d1 )  N 2 ( 2  d 2 )
 s12 
N1  N 2

d1 = X1  X12
d2 = X 2  X12

N1 X1  N 2 X 2
X 12 
N1  N 2

9.5.2 Test procedure

Having calculated appropriate z-statistic or t-statistic, to reject or accept the
null hypothesis, it is necessary to identify the rejection region with reference
to the given level of significance. If the calculated statistic is in the rejection
region, we accept the alternative hypothesis against the null hypothesis at

Manipal University Jaipur Page No. 371

Statistics for Management Unit 9

that level of significance. Otherwise, we accept null hypothesis at given level

of significance. Table 9.4 depicts the rejection region, normally denoted by
‘R’.
Table 9.4: Kinds of Tests

Kind of test z- statistic t- statistic

Two tail test R: |z| > |ztable| R: |t| > |ttable|
Lower tail test R: z < ztable R: t < ttable
Upper tail test R: z > ztable R: t > ttable

Figure 9.4 depicts the hypothesis testing procedure.

Fig. 9.4: Hypothesis Testing Procedure

Manipal University Jaipur Page No. 372

Statistics for Management Unit 9

9.5.3 How to identify the right statistics for the test

Figure 9.5 depicts the step by step procedure to identify the right statistics
for the test.

Fig. 9.5: Identification of Right Statistics for the Test

Self Assessment Questions

2. i) Null hypothesis states that there is a significant difference between
observed and hypothetical values. (True/False)

Manipal University Jaipur Page No. 373

Statistics for Management Unit 9

ii) 1% level of significance means we are ready to reject a true

hypothesis in 99% of cases. (True/False)
iii) If the Null hypothesis H0:  = X or H0: p = ps or H0: 1 = 2 or H0: p1
= p2 then it is two-tailed test. (True/False)
iv) If the calculated value of a statistic is not in the rejection region R,
then Ho is accepted. (True/False)
v) 1 -  is called power of the test. (True/False)
vi) If n1 = 300, n2 = 500, 1 = 50, 2 = 60, 1 = 10, 2 = 12 are results of
two samples taken from two cities A and B then we test for between
means under different population. (True/False)
vii) If n < 30, then we do not apply z test unless, population S.D is
known. (True/False)

Solved problem 1
XYZ press hypothesis is that the average life of its latest web-offset press is
14,500 hours. They know the standard deviation of the press life is 2,100
hours. From a sample of 25 presses, the company finds a sample mean of
13,000 hours. At a significance level of 0.01, should the company conclude
that the average life of the presses is less than the hypothesised 14,500
hours?
Solution
The procedure is described here:
1. Null hypothesis H0:  = 14,500
Alternate hypothesis H1:  < 14,500 (one-tailed test)
2. Level of significance  = 0.01  Ztab = - 2.33 and R:z< -2.33
(X  )
3. Test statistics Z
p
n

4. Given  = 14,500, X = 13,000, p = 2,100, n = 25

Note: Although n < 25, population standard deviation is given, therefore
it becomes Z test.

Manipal University Jaipur Page No. 374

Statistics for Management Unit 9

p 2100 2100
    420
n 25 5

13000  14500
Z cal    3.57
420
5. Conclusion: Since Zcal (-3.57) < Ztab (-2.33) and is in the rejection region,
H0 is rejected. In other words, we accept that the average life of the
press is significantly lesser than 14,500 hrs at 1% level of significance.

Solved problem 2
Theatre owners in India know that a hit movie ran for an average of 84 days,
with a standard deviation of 10 days in each city the movie was screened. A
particular movie distributor was interested in comparing the popularity of the
movie in his/her region with that of the population. The distributor randomly
chose 75 theatres in the region and found a popular movie ran for 81.5
days.
1) State appropriate hypothesis for testing whether there was a significant
difference between theatres in the distributor’s region and the
population.
2) At 1% significance level, test this hypothesis.

Solution
The procedure is explained in the form of steps:
1. Null hypothesis Ho:  = 84
Alternate hypothesis H1:   84 (two-tailed test)
2. Level of significance  = 0.01  Ztab = 2.58 and R:|z| > 2.58
3. Test statistics
(X  )
Z
p
n

4. Given  = 84, X = 81.5, p = 10, n = 75

p 10
Therefore,   1.1547
n 75

Manipal University Jaipur Page No. 375

Statistics for Management Unit 9

84  81.5
Z cal   2.165
1.1547
5. Conclusion: Since Zcal (2.165) < Ztab (2.58), and not in the rejection
region, H0 is accepted at 1% level of significance.

Solved Problem 3
A ketchup manufacturer is in the process of deciding whether to produce a
new extra spicy brand of ketchup. In a survey of 6000 households, the
company’s market research team found that, 355 households would buy the
extra spicy brand. A more extensive study carried out 2 years ago showed
that 5% of the households would buy the brand then. At 2% level of
significance, should the company conclude that there is an increased
interest in the extra spicy flavour?
Solution
The procedure is explained in the following steps:
1. Null hypothesis Ho: p = 0.05
Alternate hypothesis H1: p > 0.05 (one-tailed test)
2. Level of significance  = 0.01  Ztab = 2.05 and R:z > 2.05
3. Test statistics

p p
Z
pq
n

4. Given p = 0.05, p = 355/6000 = 0.0592, n = 6000, q = 1 – p =1- 0.05=
0.95
( 0.0592  0.05 )
Z cal   3.29
0.05  0.95
.
6000

5. Conclusion: Since Zcal (3.29) > Ztab (2.05), and found in the rejection
region, Ho is rejected and it is accepted that there is an increase of the
proportion of population having an interest in the new flavour.

Manipal University Jaipur Page No. 376

Statistics for Management Unit 9

Solved Problem 4
Microsoft estimated that out of 10,000 potential software buyers, 35% wait
to purchase the new OS Windows Vista, until an upgrade has been
released. After an advertising campaign to reassure the public was
released, Microsoft surveyed 3000 buyers and found 950 who are still
skeptical. At 5% level of significance, can the company conclude that the
population of skeptical people had decreased?
Solution
The procedure is explained in the following steps:
1. Null hypothesis Ho: p = .35
Alternate hypothesis H1: p < 0.35
2. Level of significance  = 0.05  Ztab = - 1.645 and R: z < -1.645
3. Test statistics

( p  p)
Z
 pq  Nn
   
 n   N 1 

4. Given p = 950/3000 = 19/60 = 0.317, p = 0.35, q = 1-p = 1- 0.35 = 0.65,

N=10,000, n = 3000

 pq  Nn  0.35  0.65   10000  3000 

        = 0.0073
 n   N 1   3000   10000  1 
( 0.317  0.35 )
Z cal    4.52
.0073
5. Conclusion: Since Zcal (-4.52) < Ztab (-1.645) and is in the rejection
region, Ho is rejected. At 5% level of significance, we conclude that the
proportion of skeptical people has significantly decreased.

Solved problem 5
A machine is designed to pack 200ml of a medicine with a standard
deviation of 5ml. A sample of 100 bottles when measured had a mean
content of 201.3ml. Test whether the machine is functioning properly (use
5% level of significance).

Manipal University Jaipur Page No. 377

Statistics for Management Unit 9

Solution
The procedure is explained in the following steps:
1. Null hypothesis Ho:  = 200
Alternate hypothesis H1:   200 (two-tailed test)
2. Level of significance  = 0.05  Ztab = 1.96 and R: |Z| > 1.96.
3. Test statistics
(X  )
Z
p
n

4. Given  = 200, X = 201.3, p = 5, n = 100

p 5
   0.5
n 100

201 .3  200
Z cal   13 / 5  2.60
0 .5
5. Conclusion: Since Zcal (2.60) > Ztab (1.96) and Zcal is in the rejection
region, Ho is rejected. Hence at 5% level of significance, we reject null
hypothesis and conclude that the machine is not functioning properly.

9.6 Testing of Hypothesis in the Case of Small Samples

Till now, we have studied about the testing of hypothesis when sample size
is large using normal distribution. However, if the sample size is small, then
the distributions of the statistics are far from normal and hence normal test
cannot be applied. Hence to deal with small samples, tests of significance
known as exact sample tests have been developed. For all practical
purposes the sample is termed as small if n  30.
The basic fundamental assumptions in all exact sample tests are:
 the parent population from which the sample drawn is normally
distributed
 Sample/samples is/are drawn at random
 They are independent of each other

Manipal University Jaipur Page No. 378

Statistics for Management Unit 9

It should be noted that the methods and theory of small samples are
applicable to large samples, but the reverse is not true.

9.7 ‘t’ Distribution

The ‘t’ distribution was developed by W.S.Gossett in the name ‘student’.
Therefore, it is known as student’s ’t’ distribution. The properties of ‘t’
distribution are:
1. ‘t’ distribution is a continuous probability distribution
2. “t” statistic is defined as:
( X  )
t
s
n
(X  X) 2
Where, s 
n 1
3. The probability density function is given by:

v  1/ 2
 t2 
f ( t )  C 1  
 v
where,
C = Constant required to make the area under the curve equal to unity.
 = n – 1, Degree of Freedom.
4. The value of ‘t’ ranges from -  to + 
5. “” is called the parameter of the distribution
6. It is symmetrical about mean
7. Its mean is zero
8. Variance of the distribution is greater than one
9. It has larger areas at the tails compared to normal distribution and
lower height at the mean.
10. It tends to a normal distribution as n  ∞.
9.7.1 Uses of ‘t’ test
The ‘t’ test is used:
 To test a specified value
Manipal University Jaipur Page No. 379
Statistics for Management Unit 9

 To test the differences between values (independent sample)

 As a paired ‘t’–test (dependent sample)
 To construct confidence interval for the estimates
Table 9.5 depicts the description of test in the case of small samples where
the population standard deviation is not known.
Table 9.5: Description of Test in the Case of Small Samples

Test Description
Test Statistics Notes
No. of Test
1 Test for X is the sample
specified mean
value – infinite
 = Hypothised
population ( X  )
t value of
with s population mean
d.f. = n -1, n
Population  ( X  X )2
variance s  i
n 1
unknown
2 Test for ( X  )
specified t
value – finite s N  n
  N = Population
population n  N 1 size
with d.f.= n-1,
Population
variance
unknown
3 Test between X1  X 2 X1 = first sample
values – t
mean
independent 2
(X  X )   (X  X )
2 1 1
samples with 1i 1 2i 2    X 2 = second
n n 2 n n 
d.f= n1 + n2 – 1 2  1 2 sample mean
2, X1  X 2 n1 and n 2 are
t
Population sizes of first and
( n  1) 2s  ( n  1) 2s 1 1
variances not 1 1 2 2   second sample
known but n n 2  n  n  respectively.
1 2  1 2
assumed to
be equal

Manipal University Jaipur Page No. 380

Statistics for Management Unit 9

4 Paired “t – D = Mean of
D
test t
(dependent  diff , where difference
n = sample size
samples) with n
d.f= n -1

 diff 

2
 D 2i  D .n
n 1

Solved problem 6
A random sample of 10 bags of fertilisers is found to have the following
weight (kg):
45, 49, 50, 49, 44, 52, 48, 45, 46, 45
Test at 5% level of significance whether the average packing weight can be
taken as 50 kg.
Solution: Table 9.6 depicts the frequency table for solved
problem 6.
Table 9.6: Frequency Table
Xi
X i X =  X i X = 
2

X i  47.3 Xi  47.32

45 -2.3 5.29
49 1.7 2.89
50 2.7 7.29
49 1.7 2.89
44 -3.3 10.89
52 4.7 22.09
48 0.7 0.49
45 -2.3 5.29
46 -1.3 1.69
45 -2.3 5.29

 X i =473 
∑ X i  X =64.1 
2

Manipal University Jaipur Page No. 381

Statistics for Management Unit 9

Sample mean is given by the formula

n

X i
473
X i 1
  47.3
n 10
The sample variance is given by,
n

 (X i  X) 2
64.1
s 2  i 1
  7.12
n 1 10  1
 s  2.6687

The steps are described as follows:

1. Null hypothesis Ho:   50
Alternate hypothesis H1:   50 (two tailed test)
2. Level of significance 5% and degrees of freedom (d.f.)= 9
ttab = 2.262 and the rejection region is R:|t| > 2.262
3. Test statistics
( X  )
t 
s
n
s 2.6687
Given X  47.3 ,  = 50,  s  2.6687 , = = 0.8439
n 10
47.3  50.0
t cal    3.19
0.8439
4. Conclusion: Since |tcal (-3.19)| > |ttab (2.262)|, tcal is in the rejection
region, H0 is rejected. Therefore we conclude, at 5% level of significance
the mean weight of fertiliser bags is not 50 kg.
Solved problem 7
For example, in the above problem, out of 1000 bags packed in a day, a
random sample of 10 was selected and the readings were as given in
solved problem 6. Test whether the population average weight is 50 kg.
Solution
The steps are described as follows:
1. Null hypothesis H0:   50

Manipal University Jaipur Page No. 382

Statistics for Management Unit 9

Alternate hypothesis H1:   50 (two tailed test)

2. Level of significance 5% and degrees of freedom (d.f.)= 9
ttab = 2.262 and the rejection region is R:|t| > 2.262.
3. Test statistics

( X  )
t
s N n
 
n  N  1 
s
Given n = 10, N = 1000, X  47.3 , = 0.8439
n

( X  ) ( 47.3  50)
t cal  
  3.2138
s N n  1000  10 
  0.8439  
n  N  1   1000  1 
4. Conclusion: Since |tcal (-3.2138)| > |ttab (2.262)|, tcal is in the rejection
region and thus Ho is rejected.

Solved Problem 8
Average tensile strength of nine samples of paper is found to be 15.8 units
and variance is 10.3. Can we say at 1% level of significance that it is a
random sample drawn from a population whose mean tensile strength is
17.5?
Solution
The steps are described as follows:
1. Null hypothesis H0:   17 .5
Alternate hypothesis H1:   17 .5
2. Level of significance 1% and degrees of freedom (d.f.) =n-1= 9-1=8
ttab = 3.36 and R:|t| > 3.36
3. Test statistics

Manipal University Jaipur Page No. 383

Statistics for Management Unit 9

( X  )
t 
s
n
X  15 .8 ,  = 17.5,  s = 10.3 ,
2
Given n=9

s 3.2084
  1.0698
n 9
15.8  17.5
t cal    1.5891
1.0698
4. Conclusion: Since Itcal (-1.5891)| < Ittab (3.36)I, Ho is accepted
 It can be considered as a random sample at 1% level of
significance.
Solved Problem 9
A sales manager wants to know whether a special promotional campaign is
a success. Table 9.7 depicts the data. Test at 5% level of significance,
whether it is a success?
Table 9.7: Sales Data Before and After the Campaign
Retail Outlets 1 2 3 4 5 6
Sales before campaign 50 48 31 42 28 53
Sales after campaign 56 55 30 45 29 58

Solution
Table 9.7a depicts the frequency table calculated for the sales data before
and after the campaign.
Table 9.7a: Frequency Table for the Sales Data Before and After the Campaign
Before (Xi) After (Yi) D = After – Before Di 2
Di = Yi - Xi
Campaign
50 56 6 36
48 55 7 49
31 30 -1 1
42 45 3 9
28 29 1 1
53 58 5 25
∑ Di = 21 ∑ Di 2 =121

Manipal University Jaipur Page No. 384

Statistics for Management Unit 9

 D i 21
Mean of Differences or D    3.5
n 6

 diff 

2
 D 2i  D .n
n 1

121  2 .6
 3.5
 diff  =3.08
6 1

The steps are described as follows:

1. Null hypothesis Ho: D  0
Alternate hypothesis H1: D  0 (one tailed test)
2. Level of significance 5% and d.f.= 5
ttab = 2.02 and R: t > 2.02.
3. Test statistics
D 3.5
t cal    2.78
 diff 3.08
n 6
4. Conclusion: Since tcal (2.78) > ttab (2.02) and is in the rejection region,
Ho is rejected. So, we accept the alternative hypothesis that there is a
significant difference in sales after the campaign.

Self Assessment Questions

3. i) ‘t’ distribution is __________ probability distribution.
ii) ‘t’ distribution’s parameter is __________.
iii) The mean and variance of the ‘t’ distribution are ________ and
________.

Manipal University Jaipur Page No. 385

Statistics for Management Unit 9

Activity:
1. A random sample of 200 tins of vanaspathi has a mean weight 4.97
kgs and a standard deviation of 0.2kgs. Test at 1% level of
significance, that the tins have 5 kgs. vanaspathi
2. A random sample of 100 rods drawn from a lot of rods has a mean
length 32.7cms. and a standard deviation of 1.3cms. Can it be
concluded that the lot has a mean of 32 cms?
Solution
1. H0 : µ = 5kg
H1: µ ≠ 5kg
Level of significance  = 0.01  Ztab = 2.58 and R: |Z| > 2.58
Test statistics
( X  )
Z
s
n
Given  = 5, X = 4.97,  s = 0.2, n = 200
4.97  5
Z cal    2.12
0 .2
200
Conclusion: Since IZcalI < Ztab, we accept H0 at 1% level of
significance and conclude that the tins have 5 kgs of vanaspathi.
2. H0 : µ = 32
H1: µ ≠ 32
Level of significance  = 0.05  Ztab = 1.96 and R: |Z| > 1.96
Test statistics
( X  )
Z
s
n
Given  = 32, X = 32.7,  s = 1.3, n = 100
32.7  32
Z cal   5.38
1.3
100
Conclusion: Since Zcal > Ztab, we reject H0 at 5% level of significance
and conclude that the lot does not have a mean of 32 cms.
Manipal University Jaipur Page No. 386
Statistics for Management Unit 9

9.8 Summary
Let us recapitulate the important concepts discussed in this unit:
 Hypothesis testing is the opinion about the population parameter that
may or may not be in the confidence interval derived from the sample.
 In hypothesis testing, we must state the assumed or hypothesised value
of the population parameter before we begin sampling. The assumption
we wish to test is called the null hypothesis and is symbolised by ’H0’.
 If our sample results fail to support the null hypothesis, we must
conclude that something else is true. Whenever we reject the
hypothesis, the conclusion we do accept is called the alternative
hypothesis and is symbolised as ‘H1.
 If null hypothesis is true, and the test result make us to accept it, then
we have made a right decision.
 If null hypothesis is true, and the test result make us to reject it, then we
have made a wrong decision (Type I error). It is also known as
consumer’s risk, denoted by .
 If hypothesis is false, and the test results make us to accept it, then we
have made a wrong decision (Type II error). It is known as producer’s
risk, denoted by  .1 – is called power of the test.
 If hypothesis is false, and the test result make us to reject it , we have
made a right decision.
 ‘t’ tests can be used for sample size (n  30) and samples whose
population standard deviations are not known.

9.9 Glossary
Level of significance: The smallest probability at which the null hypothesis
would be rejected (Type I error). Usually, if the significance level is less than
a number such as 0.05 (5%), the null hypothesis would be rejected in favour
of the alternative; the chance of getting a sample like the one being
analysed if the null hypothesis were true. A small significance level would
imply that getting such a sample was highly unlikely, suggesting that the null
hypothesis is probably not true; also called the P-value of the test.
Null distribution: The distribution of the test statistic assuming the null
hypothesis is true.

Manipal University Jaipur Page No. 387

Statistics for Management Unit 9

One-tailed test: A test in which the alternative hypothesis specifies that the
population parameter is strictly greater, or strictly lesser, than a specified
value. A test in which the alternative hypothesis specifies that the parameter
is on "one side" of the null hypothesis value; a test in which H1 contains >
or <.
P-value: The value that indicates how unusual a computed test statistic
compared with what would be expected under the null hypothesis. A small
value indicates that the null hypothesis should be rejected at any
significance level above the calculated value. For example, if the P value
equals 0.0246, we would reject the null hypothesis at the 5% significance
level, but would not reject it at the 1% significance level.
Two-tailed test: The rejection region in a two-tailed test is split between the
two tails of the distribution.
Type I error: Rejecting a true null hypothesis. The probability of a type I
error is indicated by alpha (α).
Type II error: Not rejecting a false null hypothesis. The probability of a type
II error is indicated by beta (β).
Z test for a population mean: Tests a hypothesis pertaining to the
population mean by using a z-test statistic to evaluate the magnitude of
difference between the sample mean.
Z test for a population proportion: Tests a hypothesis pertaining to the
population proportion by using a z-test statistic to evaluate the magnitude of
the difference between sample proportion and hypothesised population
proportion.

9.10 Terminal Questions

1. Twenty households out of 1000 were using Brand ‘A’ toothpaste. The
company increased the price of the brand. In a survey, they found that
only 12 households out of 1000 are using it now. Can we conclude at
5% level of significance that proportion of users has decreased?
2. A drill drills holes with standard deviation of depth 0.03cms. It is adjusted
to drill holes of depth 5.5cm. For 50 holes drilled, the mean depth is
5.503cm. Test at 5% level of significance whether the adjustment is
correct.

Manipal University Jaipur Page No. 388

Statistics for Management Unit 9

3. Out of 80 batteries produced by a process I, three were found to be

defective. In another sample of 130 produced by process II, two were
found to be defective. Test whether the proportion of defectives in two
processes differs, using 1% level of significance.
4. The table 9.8 depicts the data related to mean weight of a product. Test
whether there is a significant difference in means of the plants.
Table 9.8: Mean Weight of a Product
Plant A Plant B
Size 300 200
Mean 75.4 74.3
Variance 65.6 57.8
5. A machine is set to produce particular characteristics with mean 21.3
and S.D 0.4. A random sample of 625 observations has 21.33 as mean.
Test whether the sample mean differ significantly from population mean.
6. Out 10,000 pumpkins harvested, 1000 were randomly selected. 8%
were found to be rotten. The grower claims that only 7% are rotten. In
this claim tenable? Test at 5% level of significance.
7. A group of seven–week–old chickens reared on a high protein diet
weigh 12, 15, 11, 16, 14, 14 and 16 ounces. In another group, 5 chicken
received low protein diet and weigh 8, 10, 14, 10, and 13. Test whether
there is significant increase in weight due to high protein, use 5% level
of significance.
8. Table 9.9 depicts the strength test results of two yarns. Is there a
significant difference in the mean? Test at 5% level of significance.

Table 9.9: Strength Results of the Two Yarns

Sample Size Mean Sample Variance
Type A 4 52 42
Type B 9 42 56

9. The table 9.10 depicts the results related to the memory capacity of 10
students before and after training. Test at 5% level of significance
whether training is effective.

Manipal University Jaipur Page No. 389

Statistics for Management Unit 9

Table 9.10: Memory Capacity of 10 Students

Roll No 1 2 3 4 5 6 7 8 9 1
Before Training 1 14 11 8 7 1 3 0 5 6
After Training 1 16 10 7 5 1 10 2 3 8

9.11 Answers

Self Assessment Questions

1. i) Normal distribution
ii) Normal distribution
iii) ‘t’ distribution
iv) Normal distribution
v) Normal distribution
2. i) False
ii) False
iii) True
iv) True
v) True
vi) True
vii) True
3. i) Continuous
ii) Degrees of freedom
iii) Zero, greater than one

Terminal Questions
1. Zcal = 1.9457, H0 accepted
2. Zcal = 0.71, H0 accepted
3. Zcal = 0.50, H0 accepted
4. Zcal = 1.54, H0 accepted
5. Zcal = 18.75, H0 rejected
6. Zcal = 1.30, H0 accepted
7. tcal = 2.397, H0 is rejected
8. tcal = 2.21, H0 is rejected
9. tcal = 1.365, H0 is rejected

Manipal University Jaipur Page No. 390

Statistics for Management Unit 9

9.12 Case Study

Automatic Teller Machines
Situation
Banks in SriLanka are closed for 3.5 days from Friday afternoon to Monday
morning. In this case, banks need to have a reasonable estimate of how
much cash should be available in their ATMs. For BNP-Colombo branches
in the southeast of SriLanka, it estimates that for this 3.5 day period the
demand from its customers from hose branches with a single ATM machine
is $ 2,200 with a population standard deviation of $205. A random sample of
the withdrawal from 36 of its branches indicates a sample average
withdrawal of $ 2,235.
Discussion Questions:
i) Using the concept of critical values, at the 5% significance level does
this data indicate that the mean withdrawal from the machine it
different from $ 2,200?
ii) Re-examine question 1 using the p-value approach. Are your
conclusions the same? Explain your conclusions?
iii) What are the confidence limits at 5% significance? How do these
values corroborate your answers to question 1 and 2?
iv) Using the concept of critical values, at the 1% significance level does
this data indicate that the mean life of the population of the batteries is
different from $2,200?
v) Re-examine question 4 using the p-value approach. Are your
conclusions the same? Explain your conclusions?
vi) What are the confidence limits at 1% significance? How do these
values corroborate your answers to Question 4 and 5?
vii) Here we have used the test for a difference. Why is the bank
interested in the difference rather than a one–tail test, either left or
right hand?

References:
 Bevington, P.R. & Robinson, D.K. Data Reduction and Error Analysis for
the Physical Sciences (3rd Edition), (Paperback).
 Cowan, G. Statistical Data Analysis (Oxford Science Publications),
(Paperback).
Manipal University Jaipur Page No. 391
Statistics for Management Unit 9

 Devore, J.L. Probability and Statistics for Engineering and the Sciences,
Enhanced Review Edition. (Hardcover - Jan. 29, 2008).
 James, F. Statistical Methods in Experimental Physics (2nd Edition).
(Hardcover - Nov. 29, 2006).
 Levin, R.I. & Rubin, D.S. (2008) Statistics for Management, Seventh
Edition, PHI Learning Private Limited.
 Lyons, l. Statistics for Nuclear and Particle Physicists. (Paperback,
1989).
 Mandel, J. The Statistical Analysis of Experimental Data, (Paperback).
 Meyer, S.L. Data Analysis for Scientists and Engineers, (Paperback).
 Morris, H., Schervish, M.J. & Degroot, Probability and Statistics
[PROBABILITY & STATISTICS 3 -OS] (Paperback - Jan. 31, 2002).
 Press, W.H., Teukolsky, S.A., Vetterling, W.T. & Flannery, B.P.
Numerical Recipes: The Art of Scientific Computing, 3rd Edition.

Manipal University Jaipur Page No. 392