Statistics 102 Unit 7.

1: Statistical Hypothesis Testing

Concept Description
Statistical Introductory concepts:
hypothesis - If we obtain evidence from the sample that is
inconsistent with the stated hypothesis, we decide to
reject a hypothesis
- If we have insufficient evidence to show that a sample is
inconsistent with the state hypothesis, we accept a
hypothesis
Concerns of - A researcher seeks to determine whether the children in a
hypothesis particular shelter has an attention span that is less than two
testing can hours
revolve on the (What we are concerned here is the mean)
mean ( μ ¿ and
the value of - A surveyor seeks to check whether or not 45% of
proportion (p) commercial establishments in area at a safe distance away
from the earthquake danger zone
(What we are concerned here is the proportion)
Statistical Claim or assertion about a parameter or some parameters of one or
hypothesis more population
Examples of - Let μ be the mean number of text messages from mobile
hypothesis phones. Such that μ=87 is the mean number of text
messages from mobile phones (Example 1)

- Let p be the proportion of students that did not attend the

general assembly. Such that p = 0.08 is the proportion of
students that did not attend the general assembly. (Example
2)
Example 1 In example 1, we can infer two hypotheses:
- μ=87--- a claim of equality assumed as the correct claim
null hypothesis
- μ ≠ 87--- a competing claim alternative hypothesis
Example 2 - p=0.15
- p ≠ 0.15
Null hypothesis Denoted by H o

Claim about a characteristic of a population that is initially assumed

as the true statement
Alternative Denoted by H a
hypothesis
Competing claim or statement to the null hypothesis
Points to - We set up the null hypothesis with the hope that it will be
remember in rejected
hypothesis - The null hypothesis will not be rejected if the sample data
testing does not provide strong evidence there is no evidence to
believe otherwise
- There are only two possible decisions: rejection of null
hypothesis vs the acceptance of null hypothesis
- A test procedure can only show strong evidence for the
alternative hypothesis by rejecting the null hypothesis
- A test procedure is specified in terms of a particular test
statistic and reject rule
- Test statistic used to test hypotheses about parameters is a

1|Page
 point estimator of the parameter
Form of null H o : population parameter=hypothesized value
hypothesis - Can also be represented as:
- H o : population parameter ≥hypothesized value
- H o : population parameter ≤hypothesized value
Forms of - H a : population parameter >hypothesized value
alternative - H a : population parameter <hypothesized value
hypothesis
- H a : population parameter ≠ hypothesized value
Rejection or - Set or collection of unlikely values for the real statistic
critical region -
Sample problem A manufacturer of car tires wants to test whether their machine still
1 produces car tires with a mean diameter of 16 inches.

Answer:
- H o : μ=16
- H a : μ ≠16

Exercises 1. Write the null and alternative hypothesis of each situation

(1) A company that produces a particular brand of snacks claims that each box of snack
pack they produce has a mean of 15 grams. Consumers would not want to purchase an item
with less than the advertised weight.

Solution:

(2) An airline claims that the percentage of their flights prone to delays weekly due to air
traffic and other technical issues does not exceed by 28% of their total flights. People who
fly with the airline would be unhappy if the percentage of delayed flights is greater than
the airline claims.

Solution:

Concept Description
Errors in hypothesis Introductory concepts:
testing - There is always a chance that testing the sample data in
lieu of the whole population would lead us to a wrong
decision or conclusion about a parameter that we are
studying.
Type I error Error of rejecting a null hypothesis when it is true

2|Page
Type II error - Error of failing to reject a null hypothesis when it is
false
Sample problem A manufacturer of car tires wants to test whether their
machine still produces car tires with a mean diameter of 16
inches.

Answer:
- H o : μ=16
- H a : μ ≠16
Type I error in the - Error rejecting a null hypothesis when it is true
example - To reject the true null hypothesis is to believe that the
machine produces car tires with non-identical diameters
when it actually produces identical ones (16)
Type II error in the - Error of not rejecting a false null hypothesis
example - Accept the claim that the machine produces car tires
with a mean diameter of 16 when it does not
α (level of significance) Probability of a Type I error
β Probability of a Type II error
Points to remember - It is ideal to get a result of α ∧β that are both equal to 0
in hypothesis testing but since we are basing our decision from a sample
rather than a census this would be impossible to
obtain
- A standard test procedure gives the researcher direct
control over the value of α but not in β
- Reducing the probability of Type I error increases the
risk of a Type II error
- Using a smaller value of a increases the value of B
(inversely related)
Assuming that a = - The probability of rejecting a true null hypothesis is five
0.05 percent (5 out of 100)
Assuming that a = - Results to a test procedure with a one percent
0.01 probability of a Type-I error ( 1 out of 100)

Statistics 102 Unit 7.2: Test of Hypothesis on the

Population Mean
Concept Description
Test Introductory concepts:
concernin - After we have formulated the null and alternative hypotheses and
g means identified an appropriate level of significance, we have to test the
with hypotheses by using an appropriate test statistic and
known determining the critical region on which we base our decision to
variances reject or not reject the null hypothesis.
A sample A z-test is used when the standard deviation (σ ) is known
which has x−hypothesized value
an z−
approxima
- σ
tely √n
normal - x : sample mean
distributio - z : z-score
n - σ : population standard deviation
- n : sample size

3|Page
Critical x−μ0
region of a z−
sample
- σ
which has √n
an
approxima
tely
normal
distributio
n
(transform
ation of a
z-value)
Alternative Alternative hypothesis Reject null hypothesis if
hypotheses μ< μ0 z ≤−z a
for an α
μ> μ0 z ≥ za
level of
significanc μ ≠ μo z ≤−z a ∨z ≥ z a
e: 2 2

Determini
ng the
dividing
lines or
critical
values of
the
random
variable z
that
correspon
d to x

4|Page
Critical
values for
a 0.05
level of
significanc
e for the
one-tailed
test

( -1.645 or
1.645)

5|Page
Critical
values for
a 0.05
level of
significanc
e for the
two-tailed
test

( -1.96 or
1.96)

Summary-
table

Sample - The manufacturer of a particular brand of call,m nned tuna claims

problem that each can has an average weight of 150 grams with a standard
on deviation of 8 grams. Suppose a random sample of 80 cans is tested
hypothesis and found to have an average weight of 145 grams, test the
-testing hypothesis that μ=150 grams against the alternative hypothesis
μ ≠150 grams with a 0.01 level of significance.
Solution - State the null and alternative hypothesis:

H o : μ=150 g
H a : μ ≠150 g

- Use a 0.01 level of significance as identified in the situation

α =0.01
- Since the population standard deviation is given, we use a z-test
and establish the critical region for a=0.01

We reject H 0 when z ≤−z a ∨z > z a ; z ≤−z 0.01 ∨z> z 0.01 ;

2 2 2 2

6|Page
 z ≤−2.58∨z >2.58
- Solve for z given the test statistic formula:
x−μ0
z=
- σ
√n
- z = (145-150)/ (8/√ 80) = -5.59

- Since z = -5.59 ≤ 2.58, we decide to reject the null hypothesis and

conclude that the average weight of the particular brand of canned
tuna is not equal to 150 grams.

Test x−hypothesized value

concernin t=
- s
g means
with √n
unknown - t: t-score
variances - x : sample mean
and the - s: sample standard deviation
sample - n: sample size
size is
small (n <
30)
t-score -
value of a random variable of a t-distribution with (n-1) degrees of
freedom
x−μ0
Transformation of t used to determine the critical region in terms of t-
t= values
s -
√n
For an a- Alternative hypothesis Reject null hypothesis if
level of μ< μ0 t ≤−t a
significanc
e we can
μ> μ0 t ≥ ta
determine μ ≠ μo t ≤−t a ∨t ≥ t a
the 2 2
dividing
lines or
critical
values that
correspon
d to x
Sample The mean waiting time for the processing of a personal loan in a certain
problem bank is 90 minutes with a standard deviation of 25 minutes. The bank
personnel are testing a new kind of processing method. If a random
sample of 15 clients recorded a mean processing time of 78 minutes with a
standard deviation of 12 minutes using a new method, test the claim that
the population mean is now less than 90 minutes. Use a 0.05 level of
significance and assume the population to be approximately normal.

7|Page
Statistics 102 Unit 7.3: Test of Hypotheses on the
Population Proportion
Concept Description
Test concerning - Testing hypotheses given a sample data that can help us
proportions with decide whether a value of a proportion in the form of a
small N percent or a probability is less than, greater than or equal to
a certain constant
Applications: - Proportion of the voters who favor a certain politician in a
municipality
- Product manufacturer would want to know the percentage of
his products which are defective
p-value or tail Lowest level of significance for which we reject the null hypothesis
probability
Testing the - Base on the binomial probability table
significance on - Corresponding p-value of the sample data
p-values
Test concerning - Based on the normal curve approximation
with a large N
z-value for x−n p0
testing in a test - z=
which concerns √ n p 0 (1− p0 )
on a large N - z: z-value
- x: number of units with the desired characteristic
- n: sample size
- p0 : proportion
Identifying the Alternative hypothesis Reject null hypothesis if
critical region p< p0 z ≤−z a
with a level of
significance
p> p0 z ≥ za
p ≠ po z ≤−z a ∨z ≥ z a
2 2

8|Page