POLYTECHNIC UNIVERSITY OF THE PHILIPPINES

COLLEGE OF ENGINEERING
DEPARTMENT OF INDUSTRIAL ENGINEERING

MODULE 1
TESTS OF HYPOTHESES FOR A SINGLE SAMPLE
1.0 Hypothesis Testing

1.1 Statistical Hypothesis

Many problems in Industrial Engineering require for us to

decide whether to accept or reject a statement about some
statistical parameter. The statement is called a hypothesis and
the decision-making procedure about the hypothesis is called
hypothesis testing. Many types of decision-making problems,
tests or experiments in Industrial Engineering can be
formulated as hypothesis-testing problems.

DEFNITION: A statistical hypothesis is a statement about

the parameters of one or more populations.

Suppose that we are interested in the burning rate of a solid

propellant used to power aircrew escape systems. Burning
rate is a random variable that can be described by a
probability distribution. Suppose further that we are
interested on the mean burning rate; specifically, we are
interested in deciding whether or not the mean burning rate is
50 centimeters per second. This can be expressed formally as

H 0 : = 50 cm / s
H 1 : 50 cm / s

The statement H 0 : = 50 cm / sis called the null

hypothesis, and the statement H 1 : 50 cm / s is called the
alternative hypothesis. In the situation above, the alternative
hypothesis specifies value of that could be greater or less
than 50 cm/s, it is called a two-sided alternative hypothesis.
In some situations, a one-sided alternative hypothesis is more
appropriate, as in

H 0 : = 50 cm / s H 0 : = 50 cm / s
H 1 : < 50 cm / s or H 1 : > 50 cm / s

NOTE: Hypotheses are always statements about the

population or distribution under study, not statements
about the sample.

QUESTION: What is the implication of the note above?

A procedure leading to a decision about a particular

hypothesis is called a test of hypothesis.

1.2 Tests of Statistical Hypotheses

To illustrate the general concept of a test of hypothesis,

consider the propellant burning example earlier. The null
hypothesis is that the mean burning rate is 50 cm/s, and the
alternate is that it is not equal to 50 cm/s. Converting them
into mathematical expressions, and subsequently into the test
of hypothesis format, we have:

H 0 : = 50 cm / s
H 1 : 50 cm / s

Suppose that a sample of n = 10 specimens and that the

sample mean burning rate x is observed. The sample mean
is an estimate of the true population mean . A value of the
sample mean x that falls close to the hypothesized value of
= 50 cm / s is evidence that the true mean is really 50
cm/s, which implies that such evidence supports the null
hypothesis H 0 . If the sample mean x though is significantly
different from 50 cm/s, it is evidence in support of the
alternative hypothesis H 1 .

NOTE: In the example above, the sample mean x is what

we refer to as a test statistic as we are testing x whether
its value is significantly close to or different from .

The sample mean x can take on many different values. This

means that if the sample mean x falls within a certain range
(called the acceptable region), then we will fail to reject the
null hypothesis and conclude that there is no significant
evidence to support that say that the actual mean is not
equal to 50. If the sample mean x falls outside this
acceptable range (called the critical region), then we will
reject the null hypothesis and conclude that there is
significant evidence to support that the true/actual mean
is not equal to 50. The boundaries between the critical
regions and the acceptable region are called the critical
values.
 Acceptable region

Critical regions

Since we are dealing with probabilities, we run the risk of

giving the wrong decision and conclusion. There are two
types of risks in terms of giving the wrong decision: the type
I or error and the type II or error.

DEFINITION: Type I error occurs when we reject a

statement as true but is actually correct; type II error
occurs when we accept a statement as true but is actually
false. As a short cut, type I is rejecting when true and type
II is accepting when false.

Some points regarding type I and type II errors:

1. The size of the critical region, and consequently the

probability of a type I error, can always be reduced by
appropriate selection of the critical values.
2. Type I and type II errors are related. A decrease in the
probability of one type of error always results in an
increase in the probability of the other, provided that the
sample size n does not change.
3. An increase in sample size will generally reduce both
and , provided that the critical values are held constant.
4. When the null hypothesis is false, increases as the true
value of the parameter approaches the value hypothesized
in the null hypothesis. The value of decreases as the
difference between the true mean and the hypothesized
value increases.

Generally, the analyst controls the type I error probability

when he or she selects the critical values. Thus, it is usually
easy for the analyst to set the type I error probability at (or
near) any desired value. Since the analyst can directly control
the probability of rejecting H 0 , the decision to reject the
null hypothesis is considered a strong conclusion.

The probability of type II error, on the other hand, is not

constant and depends on the true value of the parameter and
the sample size selected by the analyst. Because the type II
error is a function of both the sample size and the extent to
which the null hypothesis is false, the decision to accept H 0
is usually considered a weak conclusion. Thus, instead of
saying we accept H 0 , the term fail to reject H 0 is used
as this implies that we have not found sufficient evidence to
reject H 0 . This implies that more data are possibly required.

DEFINITION: The power of a statistical test is the

probability of rejecting the null hypothesis when the
alternative hypothesis is true. It is computed as 1 .

1.3 One-Sided and Two-Sided Hypotheses

A test of any hypothesis such as

 H 0 : = 0
H1 : 0

is called a two-sided test, because it is important to detect

differences from the hypothesized value of the mean that lie
on either side. In this type of test, the critical region is split
into two parts, with equal probability placed in each tail of
the distribution of the test statistic.

Many hypothesis-testing problems involve a one-sided

alternative hypothesis, such as

H 0 : = 0 H 0 : = 0
H1 : < 0 or H1 : > 0

If the alternative hypothesis is H 1 : > 0 , the critical region

should lie in the upper tail of the distribution of the test
statistic; if the alternative hypothesis is H 1 : < 0 , the
critical region should lie in the lower tail of the distribution.

NOTE: In constructing hypotheses, always state the null

hypothesis as an equality.

1.4 General Procedure for Hypothesis Tests

1. Identify the parameter of interest

2. State the null and alternative hypotheses
3. Choose a significance level
4. Determine an appropriate test statistic
5. State the critical region for the statistic
6. Compute any necessary sample quantities, substitute
into the equation for the test statistic, and compute
that value.
 7. Decide whether or not H 0 should be rejected
8. State the appropriate conclusion based on your
decision in step 7

EXERCISES:

1. In each of the following situations, state whether it is a

correctly stated hypotheses testing or not. If not, state your
reason.
a. H 0 : = 25; H 1 : 25
b. H 0 : > 10; H1 : = 10
c. H 0 : x = 50; H1 : x 50
d. H 0 : p = 0.1; H1 : p = 0.5
e. H 0 : s = 30; H1 : s > 30

2. The proportion of adults living in a certain town in

Arizona who are college graduates is estimated to be p = 0.4.
To test this hypothesis, a random sample of 15 adults is
selected. If the number of college graduates is between 4 and
8, the hypothesis will be accepted; otherwise we will
conclude that p 0.4.
a. Find the type I error probability for this procedure,
assuming that p = 0.4.
b. Find the probability of committing a type II error if
the true proportion is really p = 0.2.

2.0 Tests on the Mean of a Normal Distribution, Variance Known

( )
2

2.1 Hypothesis Tests on the Mean

Suppose we wish to test the hypotheses

 H 0 : = 0
H1 : 0

where 0 is a specified constant. The test statistic to be used

for this type of problem is:

X 0
Z=

n

EXAMPLE: Aircrew escape systems are powered by a solid

propellant. The burning rate of this propellant is an important
product characteristic. Specifications require that the mean
burning rate must be 50 centimeters per second. We know
that the standard deviation of burning rate is = 2 cm/s. The
experimenter decides to specify a type I error probability or
significance level of = 0.05 and selects a random sample
of n = 25. The sample average of burning rate obtained is
x = 51.3 cm/s. What conclusions should be drawn?

Let use the 8-step procedure outlined earlier:

1. Identify the parameter of interest

The parameter of interest is , which is the mean

burning rate.

2. State the null and alternative hypotheses

H 0 : = 50
H 1 : 50
3. Choose a significance level
As stated, = 0.05 .

4. Determine an appropriate test statistic

X 0
The test statistic is Z =

n

5. State the critical region for the statistic

Reject H 0 if z 0 > 1.96 or if z0 < 1.96

6. Compute any necessary sample quantities,

substitute into the equation for the test statistic, and
compute that value.

Given values: x = 51.3 and = 2

51.3 50
z0 = = 3.25
2
25

7. Decide whether or not H 0 should be rejected

Since z0 = 3.25 > 1.96 , reject H 0 .

8. State the appropriate conclusion based on your

decision in step 7

The mean burning rate is not equal to 50 centimeter per

second.
2.2 P-Values in Hypothesis Tests

The P-value approach has been widely in practice. The P-

value is the smallest level of significance that would lead to
rejection of the null hypothesis with the given data. It gives
the analyst an idea on whether the computed value of the test
statistic was just barely in the rejection region or whether it
was very far into the region.

For a normal distribution test, it is relatively easy to compute

for the P-value. If z0 is the computed value of the test
statistic, the P-value is

2[1 ( z 0 )] H1 : 0
P = 1 ( z0 ) H1 : > 0
( z 0 ) H1 : < 0

( z 0 ) is the standard normal cumulative distribution

function. Let us consider the earlier propellant example. The
computed value of the test statistic is z0 = 3.25 ; since the
alternative hypothesis is two-tailed, the P-value is

P = 2[1 (3.25)] = 0.0012

This implies that the null hypothesis would be rejected at any

level of significance 0.0012 .

EXERCISES:

3. A melting point test of n = 10 samples of a binder used in

manufacturing a rocket propellant resulted in
 x = 154.2 degrees Fahrenheit. Assume that the melting point
is normally distributed with = 1.5 degrees Fahrenheit.

a. Test H 0 : = 155 versus H1 : 155 using

= 0.01
b. What is the P-value for this test?
c. What is the -error if the true mean is = 150 ?

4. An engineer who is studying the tensile strength of a steel

alloy intended for use in golf club shafts knows that that
tensile strength is approximately normally distributed with
= 60 psi. A random sample of 12 specimens has a mean
tensile strength of x = 3,250 psi.

a. Test the hypothesis that the mean strength is 3,500

psi. Use = 0.01 .
b. What is the smallest level of significance you would
be willing to reject the null hypothesis?

5. Supercavitation is a propulsion technology for undersea

vehicles that can greatly increase their speed. It occurs above
approximately 50 meters per second, when pressure drops
sufficiently to allow the water to dissociate into water vapor,
forming a gas bubble behind the vehicle. When the gas
bubble completely encloses the vehicle, supercavitation is
said to occur. Eight tests were conducted on a scale model of
an undersea vehicle in a towing basin with the average
observed speed x = 102.2 meters per second. Assume that
speed is normally distributed with known standard deviation
= 4 meters per second.
 a. Test the hypothesis that the mean average observed
speed is less than 100 meters per second using
= 0.05 .
b. Compute the power of the test if the true mean
speed is as low as 95 meters per second.

6. Medical researchers have developed a new artificial heart

constructed primarily of titanium and plastic. The heart will
last and operate almost indefinitely once it is implanted in the
patients body, but the battery pack needs to be recharged
about every four hours. A random sample of 50 battery packs
is selected and subjected to a life test. The average life of
these batteries is 4.05 hours. Assume that battery life is
normally distributed with standard deviation equal to 0.2
hours.

Is there evidence to support the claim that mean battery life

exceeds 4 hours? Use = 0.05 ?

3.0 Tests on the Mean of a Normal Distribution, Variance

Unknown

Let us now consider the case of hypothesis testing on the

mean of a population with unknown variance. For this
particular situation, we will use s2 to estimate the value of
2 . In addition, instead of using the normal distribution, we
will use the t-distribution,

X
T=
S n
having n 1 degrees of freedom. The critical region is
dependent on the level of significance, i.e. , and the
alternative hypothesis, i.e. whether it is a one-sided test or a
two-sided test. To summarize:

H1 : 0 t 0 > t 2,n 1 or t 0 < t 2,n 1

H1 : > 0 t 0 > t ,n 1
H1 : < 0 t 0 < t ,n 1

EXAMPLE: The increased availability of light materials with

high strength has revolutionized the design and manufacture
of golf clubs, particularly drivers. Clubs with hollow heads
and very thin faces can result in much longer tee shots,
especially for players of modest skills. This is due partly to
the spring-like effect that the thin face imparts to the ball.
Firing a golf ball at the head of the club and measuring the
ratio of the outgoing velocity of the ball to the incoming
velocity can quantify this spring-like effect. The ratio of
velocities is called the coefficient of restitution of the club.
An experiment was performed in which 15 drivers produced
by a particular club maker were selected at random and their
coefficients of restitution measured. In the experiment, the
golf balls were fired from an air cannon so that the incoming
velocity and spin rate of the ball could be precisely
controlled. It is of interest to determine if there is evidence
(with = 0.05 ) to support a claim that the mean coefficient
of restitution exceed 0.82. The observations follow:

0.8411 0.8191 0.8182 0.8125 0.8750

0.8580 0.8532 0.8483 0.8276 0.7983
0.8042 0.8730 0.8282 0.8359 0.8660
SOLUTION:

1. The parameter of interest is the mean coefficient of

restitution.

H 0 : = 0.82
2. H : > 0.82
1

3. = 0.05

4. The appropriate test statistic is

X
T=
S n

5. Critical region: t0 > t0.05 ,14 = 1.761

6. Compute:

0.83725 0.82
t0 = = 2.72
0.02456 15

7. Decision: reject the null hypothesis since computed value

is within the critical region.

8. The mean coefficient of restitution exceeds 0.82.

4.0 Hypothesis Tests on the Variance and Standard Deviation of a
Normal Population

Sometimes, hypothesis tests on the population variance or

standard deviation are needed. Suppose we wish to test the
hypothesis that the variance of a normal population 2
equals a certain value, e.g. 02 , or equivalently, that the
standard deviation is equal to 0 . To test

H 0 : 2 = 02
H 1 : 2 02

the following test statistic will be used:

=
2 (n 1)S 2
0 2
0

which follows a chi-square distribution with n 1 degrees of

freedom. The critical regions are summarized below:

H 1 : 2 02 02 > 2 2, n 1 or 02 < 12 2, n 1
H 1 : 2 > 02 02 > 2 , n 1
H 1 : 2 < 02 02 < 2 , n 1

EXAMPLE: An automatic filling machine is used to fill

bottles with liquid detergent. A random sample of 20 bottles
results in a sample variance of fill volume of
s 2 = 0.0153 (fluid ounces)2. If the variance of fill volume
exceeds 0.01(fluid ounces)2, an unacceptable proportion of
bottles will be underfilled or overfilled. Is there evidence in
the sample data to suggest that the manufacturer has a
problem with underfilled or overfilled bottles? Use = 0.05 ,
and assume that the fill volume has a normal distribution.

SOLUTION:

1. The parameter of interest is the population variance 2

H 0 : 2 = 0.01
2.
H 1 : 2 > 0.01

3. = 0.05

4. The appropriate test statistic is

=
2 (n 1)S 2
0
0
2

5. Critical region: 0 > 0.05,19 = 30.14

2 2

6. Computation:

19(0.0153)
02 = = 29.07
0.01

7. Since the computed test statistic is not within the critical

region, we fail to reject the null hypothesis.

8. There is no strong evidence to conclude that the variance

of fill volume exceeds 0.01.
5.0 Tests on a Population Proportion

It is often necessary to test hypotheses on a population

proportion. For example, suppose that a random sample of
size n has been taken from a large population and that X
observations in this sample belong to a class of interest. Then
P = X n is a point estimator of the proportion of the
population p that belongs to this class.

NOTE: The parameters n and p are actually the parameters of

a binomial distribution. Fortunately, we can use the normal
distribution to approximate the binomial distribution for tests
on large samples.

For this test type, the hypotheses are:

H 0 : p = p0
H 1 : p p0

and the test statistic is given as:

p p 0
Z0 =
p0 (1 p0 )
n

with the critical regions given as:

z 0 > z 2 or z 0 < z 2

EXAMPLE: A semiconductor manufacturer produces

controllers used in automobile engine applications. The
customer requires that the process fallout or fraction
defective at a critical manufacturing step not exceed 0.05 and
that the manufacturer demonstrate process capability at this
level of quality using = 0.05 . The semiconductor
manufacturer takes a random sample of 200 devices and finds
that four of them are defective. Can the manufacturer
demonstrate process capability for the customer?

SOLUTION:

1. The parameter of interest is the process fraction defective

p.

H 0 : p = 0.05
2. H : p < 0.05
1

3. = 0.05

4. The appropriate test statistic is

p p 0
Z0 =
p0 (1 p0 )
n

5. Critical region: z 0 < z 0.05 = 1.645

6. Computation:

4
0.05
200
z0 = = 1.95
0.05(0.95)
200

7. Decision: Since the test statistic is within the critical

region, we reject the null hypothesis.
 8. Conclusion: The process is capable.

6.0 Testing for Goodness of Fit

The hypothesis-testing procedures that we have discussed in

previous sections are designed for problems in which the
population or probability distribution is known and the
hypothesis involve the parameters of the distribution.
However, there are instances when we do not know the
underlying distribution of the population, and we wish to
test the hypothesis that a particular distribution will
suffice as a population model, which is known as testing
for goodness of fit.

The test procedure requires a random sample of size n from

the population whose probability distribution is unknown.
These n observations are arranged in a frequency histogram,
having k bins or class intervals.

Let Oi be the observed frequency in the ith class interval.

From the hypothesized probability distribution, we compute
the expected frequency in that same ith class interval, which
is denoted by Ei. The test statistic is:

=
2
k
(Oi Ei )2
0
i =1 Ei

with k p 1 degrees of freedom, where p represents the

number of parameters of the hypothesized distribution
estimated by sample statistics. We reject the hypothesis that
the distribution of the population is the hypothesized
distribution if 02 > 2 ,k p 1 .
EXAMPLES:

Poisson Distribution

The number of defects in printed circuit boards is

hypothesized to follow a Poisson distribution. A random
sample of n = 60 printed boards has been collected, and the
following number of defects observed.

Number of Defects Observed Frequency

0 32
1 15
2 9
3 4

The mean of the assumed Poisson distribution in this

example is unknown and must be estimated from the sample
data. The estimate of the mean number of defects per board is
the sample average:

(32 0 + 15 1 + 9 2 + 4 3) = 0.75
60

From the Poisson distribution with parameter 0.75, we may

compute pi , the theoretical, hypothesized probability
associated with the ith class interval. The values of pi are as
follows:
 e 0.75 (0.75)
0

p1 = P( X = 0 ) = = 0.472
0!
e 0.75 (0.75)
1

p2 = P( X = 1) = = 0.354
1!
e 0.75 (0.75)
2

p3 = P( X = 2 ) = = 0.133
2!
3
p 4 = P ( X 3) = 1 pi = 1 (0.472 + 0.354 + 0.133) = 0.041
i =1
The expected frequencies are computed by multiplying the
sample size n = 60 time the respective probabilities pi ,

Ei = npi

# OF DEFECTS PROBABILITY EXP. FREQ.

0 0.472 28.32
1 0.354 21.24
2 0.133 7.98
3 (or more) 0.041 2.46

If you will notice, the expected frequency for 3 or more

defects is less than 3. Though there is no general
agreement regarding the minimum value of expected
frequencies, values of 3 to 5 are widely used as minimal. If
the expected frequency is less than the set minimum, the cell
with the expected frequency below the minimum is combined
with the cell preceding it. For this example, we combine the
expected frequency of 3 or more defects with that of 2
defects. The revised table is as follows:
# OF DEFECTS OBS. FREQ. EXP. FREQ.
0 32 28.32
1 15 21.24
2 (or more) 13 10.44

The chi-square test statistic will have k p 1 = 3 1 1 = 1

degree of freedom, because the mean of the Poisson
distribution was estimated from the data.

1. The variable of interest is the form of the distribution of

defects in printed circuit boards.

2. H0: The form of the distribution of defects is Poisson.

H1: The form of the distribution of defects is not Poisson.

3. = 0.05

4. The test statistic to be used is:

=
2
k
(Oi Ei )2
0
i =1 Ei

5. Critical Region: 02 > 02.05,1 = 3.84

6. Computation:

=
2 (32 28.32) (15 21.24 ) (13 10.44)
2

+
2

+
2

= 2.94
0
28.32 21.24 10.44

7. Since 02 = 2.94 , we fail to reject the null hypothesis.

 8. There is no sufficient evidence to prove that the data does
not follow a Poisson distribution.

EXERCISE:

7. Consider the following frequency table of observations on

the random variable X.

Values 0 1 2 3 4
Obs. Freq. 24 30 31 11 4

Based on these 100 observations, is a Poisson distribution

with a mean of 1.2 an appropriate model? Perform a
goodness-of-fit procedure with = 0.05 .

8. Define X as the number of underfilled bottles from a filling

operation in a carton of 24 bottles. Sixty cartons are inspected
and the following observations on X are recorded:

Values 0 1 2 3
Frequency 39 23 12 1
Based on these 75 observations, is a binomial distribution an
appropriate model? Perform a goodness-of-fit test with
= 0.05 .

7.0 Contingency Table Tests

Many times, the n elements of a sample from a population

may be classified according to two different criteria. It is then
of interest to know whether the two methods of classification
are statistically independent or not. For example, we may
want to consider the population of graduating engineering
students and determine whether the starting salary is
independent of academic disciplines. (of course we know IEs
 have the highest salaries among the different engineering
disciplines )

Assume that the first method of classification has r levels and

that the second method has c levels. Let Oij be the observed
frequency for level i of the first classification method and
level j of the second classification method. The data, in
general, would appear as shown below. Such a table is called
an r c contingency table.

Columns
1 2 c
1 O11 O12 O1c
2 O21 O22 O2 c
Rows
r Or1 Or 2 Orc

We are interested in testing the hypothesis that the row-and-

column methods of classification are independent. If we
reject this hypothesis, we conclude there is some interaction
between the two criteria of classification. The exact test
procedures are somewhat difficult to obtain, but an
approximate test statistic is valid for large n.

Let pij be the probability that a randomly selected element

falls in the ijth cell, given that the two classifications are
independent. Then pij = ui v j where ui is the probability that a
randomly selected element falls in row class i and v j is the
probability that a randomly selected element falls in column
class j. Assuming independence, the estimators of ui and v j
are
 1 c
ui = Oij
n j =1
1 r
v j = Oij
n i =1

Therefore, the expected frequency of each cell is

1 c r
Eij = nui v j = Oij Oij
n j =1 i =1

Thus, for a large sample size n, the statistic is:

=
2
r c (Oij Eij )
2

0
i =1 j =1 Eij

has an approximate chi-square distribution with (r 1)(c 1)

degrees of freedom. We reject the hypothesis of
independence if the observed value of the test statistic is
greater than the critical value, i.e. 02 > 2 ,( r 1)(c1) .

EXAMPLE: A company has to choose among three pension

plans. Management wishes to know whether the preference
for plans is independent of job classification and wants to use
= 0.05 . The opinions of a random sample of 500
employees are shown in the table preceding this text.

OBSERVED DATA
Pension Plan
Job Classification 1 2 3 Total
Salaried Workers 160 140 40 340
Hourly Workers 40 60 60 160
Total 200 200 100 500
 EXPECTED FREQUENCIES
Pension Plan
Job Classification 1 2 3 Total
Salaried Workers 136 136 68 340
Hourly Workers 64 64 32 160
Total 200 200 100 500

1. The variable of interest is employee preference among

pension plans

2. H0: Preference is independent of salaried versus hourly job

classification.
H1: Preference is not independent of salaried versus hourly
job classification.

3. = 0.05

4. The test statistic is

02 =
r c (O
ij Eij )
2

i =1 j =1 Eij

5. Since r = 2 and c = 3, the degrees of freedom for chi-

square are (r 1)(c 1) = (1)(2 ) = 2 , thus rejecting the null
hypothesis if 02 > 02.05, 2 = 5.99

6. Computations:
 =
2
r c (O
ij Eij )
2

0
i =1 j =1 Eij

=
(160 136) (140 136) (40 68) (40 64)
2

+
2

+
2

+
2

136 136 68 64

+
(60 64) (60 32)
2

+
2

64 32
= 49.63

7. Since 02 = 49.63 > 02.05, 2 = 5.99 , we reject the null

hypothesis.

8. The preference for pension plans is not independent of job

classification.

EXERCISE:

9. Grades in a statistics course and an operations research

course taken simultaneously were as follows for a group
of students:

OR Grade
Stat Grade A B C Other
A 25 6 17 13
B 17 16 15 6
C 18 4 18 10
Other 10 8 11 20
Are the grades in statistics and operations research related?
Use = 0.01 in reaching your conclusion.