Lesson 8.

1: Basic Concepts in Testing Statistical

Hypotheses
Learning Activity
Perform as indicated:

1. Answer TRUE or FALSE the following:

a) A hypothesis test for which the type I error occurs with probability 0.02 has
probability of type II error equal to 0.98. FALSE
Explanation: There are two types of errors that we can make when doing hypothesis
testing- type I error and type II error. Type I error is the error committed when we decided
to reject the null hypothesis when in reality the null hypothesis (Ho) is true while Type II
error is the error committed when we decide to not reject the null hypothesis (Ho) when
in reality the null hypothesis is false or the alternative hypothesis (Ha) is true. Based on
the definition, the event that a type I error occurs is independent to the event of
committing type II error. Thus, if a hypothesis test for which the type I error occurs with
probability 0.02, we cannot say that a hypothesis test has a probability of Type II error
equal to 0.98.
However, if a hypothesis test for which type I error occurs with probability of 0.02, we
can say that there is a 0.98 probability that your decision on the Ho is correct.

b) If a test indicates that the null hypothesis can be rejected at the 0.05 level of
significance, then the test will also reject it at the 0.10 level of significance. TRUE
Explanation: The statement is true because the level of significance, 𝛼, is the maximum
probability of committing a type I error that a researcher is willing to commit. Thus, if the
null hypothesis is rejected at 𝛼 = 0.05 level of significance and we use the same data set
to perform the test at higher level of significance (greater than 0.05) for this case 0.10,
then the null hypothesis will also be rejected at 𝛼 = 0.10.
Another case is that if Ho is not rejected at 𝛼 = 0.05, it is possible to reject Ho using the
same data set by increasing 𝛼 to 0.10. However, we are doing so at a higher risk of
committing a type I error. Note that as your level of significance increases, the higher the
risk of committing type I error (that is probability of rejecting the Ho when in fact the Ho is
true).
c) We tend to reject 𝑯𝒐 for small 𝒑-values. TRUE
Explanation: Yes, we tend to reject Ho for small p-values. As you can read in your
learning guide, the p-value is often referred to as the “observed level of significance.”
Consequently, if the p-value is greater than our chosen level of significance, 𝛼, then we
do not reject the null hypothesis (i.e., the p-value is “large” so we can take Ho to be true).
On the other hand, if the p-value is less than or equal to our chosen level of significance,
then we reject the null hypothesis (i.e. the p-value is “small” so we can take Ho to be
false).

d) The null and alternative hypotheses are interchangeable, that is their roles in
statistical hypothesis testing are arbitrary. FALSE
Explanation: In performing a hypothesis testing we first identify and state the statistical
hypothesis to be tested- the null and alternative hypotheses. The null hypothesis, Ho, is
the statistical hypothesis which the researcher doubts to be true while the alternative
hypothesis (Ha) is the operational statement of the theory that the researcher believes
to be true (the research hypothesis) and wishes to prove and is a contradiction of the
null hypothesis.
In a test of hypothesis, we are going to test the null hypothesis against the alternative
hypothesis. Thus, these two hypotheses must be nonoverlapping statements about the
population. One and only one of the hypotheses can be true. If Ho is true then Ha must
be false; likewise, if Ha is true then Ho must be false. Thus, the null and alternative
hypotheses are not interchangeable, that is their roles in statistical hypothesis testing are
not arbitrary, these are specific and clear.

e) If the decision is to reject the null hypothesis, then it is impossible to commit a type
II error. TRUE
Explanation: Please refer to your learning guide Table 14. As you can see in the table,
if the decision is to reject the null hypothesis, it is impossible to commit a type II error. It
is only possible to commit a type II error if your decision is to retain or not reject the null.
And based on the table, if you decide to reject the null hypothesis, the possible type of
error that you can make is type I error, that is error of rejecting the Ho when in fact Ho is
true.
2. For each alternative hypothesis given below, state the null hypothesis.

Note:
The null hypothesis is usually expressed as a hypothesis of equality or as a hypothesis of no
effect. A common type of null hypothesis is a simple expression stating a hypothesized value for
the unknown parameter. (Note: The null hypothesis always has an equality sign or symbol (𝜇 =
𝜇𝑜 𝑜𝑟 𝑝 = 𝑝𝑜 )).
On the other hand, a common type of alternative hypothesis is a composite expression about
the direction of the unknown parameter’s value from the hypothesized value, such as:
(i) the unknown parameter is greater than its hypothesized value; (𝜇 > 𝜇𝑜 𝑜𝑟 𝑝 > 𝑝𝑜 ) or
(ii) the unknown parameter is less than its hypothesized value; (𝜇 < 𝜇𝑜 𝑜𝑟 𝑝 < 𝑝𝑜 ) or simply
(iii) the unknown parameter is not equal to its hypothesized value (𝜇 ≠ 𝜇𝑜 𝑜𝑟 𝑝 ≠ 𝑝𝑜 ).

a) The mean age of junior engineering students at the Visayas State University is not 20
years.
Ho: 𝜇 = 20 𝑦𝑒𝑎𝑟𝑠 , The mean age of junior engineering students at the Visayas State
University is 20 years.

b) The proportion of engineers having a mean monthly income of less than PhP 50,000 is
less than 1/2 in Region 8.

1
Ho: 𝑝 = , The proportion of engineers having a mean monthly income of less than
2
PhP 50, 000 is ½ or 0.50 in Region 8.
1
Note: Though our alternative hypothesis here is 𝑝 < 2 , you cannot say that your null
1
hypothesis would be 𝑝 ≥ 2 which is the complete opposite of the null. Based on our definition
on the null hypothesis, it is usually expressed as a hypothesis of equality or as a
1
hypothesis of no effect. That is why our null hypothesis would be simply 𝑝 = 2 . We just used
the simple expression.

c) The mean height of the buildings of 30 or more stories in a large city is more than 700
feet.

Ho: 𝜇 = 700 𝑓𝑡. , The mean height of the buildings of 30 or more stories in a large city
is 700 feet.
Note: Though our alternative hypothesis here is 𝜇 > 700 𝑓𝑡. , you cannot say that your null
hypothesis would be 𝜇 ≤ 700 𝑓𝑡 which is the complete opposite of the null. Based on our
definition on the null hypothesis, it is usually expressed as a hypothesis of equality or as
a hypothesis of no effect. That is why our null hypothesis would be simply 𝜇 = 700 𝑓𝑡. We
just used the simple expression.
 d) The new method is better than the present method.

Ho: The new method is just the same with the present method.
or you can say:
Ho: The new method is not better than the present method.

e) The use of a new raw material as binder improves the strength of the construction material.

Ho: The use of a new raw material as binder does not improve the strength of the
construction material.

Refers to the population standard

deviation since it is not an
Lesson 8.2: Tests for the Population Mean estimation from the sample data.
That is why we can say that in this
problem, population standard
Learning Activity The hypothesized population
deviation is known, and thus we
mean, 𝜇𝑜
will use the test statistic Z.
Perform as indicated:
1. An electrical firm manufactures light bulbs that have a lifetime that is approximately normally
distributed with a mean of 800 hours and a standard deviation of 40 hours. Test the hypothesis
that 𝜇 = 800 hours against the alternative 𝜇 ≠ 800 hours if a random sample of 30 bulbs has
a mean life of 788 hours. Use a 0.05 level of significance.
Solution: The null and alternative hypotheses
were already given.
a) The null and alternative hypotheses:
Ho: 𝜇 = 800, The average lifetime of the manufactured light bulbs is 800 hours.
Ha: 𝜇 ≠ 800, The average lifetime of the manufactured light bulbs is not 800 hours.
b) Desired level of significance: 𝛼 = 0.05
c) Test Statistic: Z, since population variance, 𝜎 2 is known.

Given: 𝑋~𝑁(800, 402 ) 𝜇𝑜 = 800, 𝜎 = 40, 𝑥̅ = 788 ℎ𝑟𝑠. n=30

𝑥̅ − 𝜇𝑜 788 − 800 −12
𝑧𝑐 = 𝜎 = =
⁄ 𝑛 40⁄ 7.302967433
√ √30
= −1.643167673
𝒛𝒄 ≅ −𝟏. 𝟔𝟒
d) Decision Rule:

Reject Ho if |𝑧𝑐 | ≥ 𝑧𝛼 = |𝑧𝑐 | ≥ 𝑧0.05 = |𝑧𝑐 | ≥ 𝑧0.025

2 2
Reject Ho if |𝑧𝑐 | ≥ 1.96, (based on the Z-table, Table B.1)
 𝑧0.025 = 1.96

1 − 0.9750 = 0.025

e) Decision on Ho:

Since |𝑧𝑐 |= |−1.64| is not greater than or equal to 1.96, we do not reject the null hypothesis.

f) Conclusion:

There is not enough evidence to say that the average lifetime of the manufactured
lightbulbs is not equal to 800 hours.

Note: You always state your conclusion based on the alternative hypothesis. If your decision is
to reject the null hypothesis, Ho, you should always state at what level of significance you rejected
the null, as we know that the level of significance, denoted by 𝛼, is the maximum probability of
committing a type I error (incorrect decision). However, if your decision is not to reject the null
hypothesis, Ho, it is not necessary to mention the level of significance.
Now, using the 𝑝 − 𝑣𝑎𝑙𝑢𝑒 in performing this test:

Since the test is two-tailed, we will have

𝑝 − 𝑣𝑎𝑙𝑢𝑒 = 𝑃(|𝑍| ≥ 1.64|𝐻𝑜 is true)

= 𝑃(𝑍 ≤ −1.64|𝐻𝑜 is true) + 𝑃(𝑍 ≥ 1.64|𝐻𝑜 is true)
= 2𝑃(𝑍 ≤ −1.64|𝐻𝑜 is true), in terms of area below
= 2 (0.0505)
= 0.101

With 𝑝 − 𝑣𝑎𝑙𝑢𝑒 = 0.101 not lesser than 𝛼 = 0.05, we do not reject the null hypothesis
that the average lifetime of the manufactured light bulbs is 800 hours. This is the same
with our earlier decision on the null hypothesis using the critical value approach.
We can also construct a 95% confidence interval on the population mean as follows (from Module
7, Lesson 7.2) and then test the null hypothesis after:

𝜎 𝜎 𝜎
Recall case 1: (𝑋̅ − 𝑧𝛼/2 , 𝑋̅ + 𝑧𝛼/2 ) ↔ 𝑋̅ ± 𝑧𝛼/2
√𝑛 √𝑛 √𝑛

Applying this here, we will have the following 95% confidence interval on the population
mean:

788 ± (1.96)(7.302967433) ≅ 14.3138 ↔ (773.69, 802.31)

We observe that 𝜇𝑜 = 800 does fall in this interval. Thus, we also do not reject the null
hypothesis based on interval estimation. We can say that we are 95% confident that the
true mean lies in the interval constructed.

Note: As you can see in the results, using the three approaches in hypothesis testing (critical
value approach, p-value approach, and interval estimation approach) you surely arrived with the
same conclusion.

2. Test the hypothesis that the mean content of containers of a particular lubricant is 10 liters if
the contents of a random sample of 10 containers are 10.2, 9.7, 10.1, 10.3, 10.1, 9.8, 9.9, 10.4,
10.3, and 9.8 liters. Use a 0.01 level of significance and assume that the distribution of contents
is normal.
This statement means, 𝜇 = 10 𝑙𝑖𝑡𝑒𝑟𝑠, since it has equality sign,
then it would be our Ho. It was not specified the direction of
Solution:
the value of the unknown population mean that is why your Ha
a) The null and alternative hypotheses: would be 𝜇 ≠ 10 𝑙𝑖𝑡𝑒𝑟𝑠. You will use the not equal sign.

Ho: 𝜇 = 10 𝑙𝑖𝑡𝑒𝑟𝑠, The average content of containers of a particular lubricant is 10 liters.

Ha: 𝜇 ≠ 10 𝑙𝑖𝑡𝑒𝑟𝑠, The average content of containers of a particular lubricant is not equal to
10 liters.
b) Desired level of significance: 𝛼 = 0.01
 c) Test statistic: t, since population variance, 𝜎 2 is unknown and the sample size is small (n≤30)

Given:
∑ 𝑋𝑖 100.6
Sample (i) 𝑋𝑖 𝑋𝑖2 𝑥̅ = = = 10.06 (sample mean)
n=10 𝑛 10
1 10.2 104.04
Since population
2 9.7 94.09 𝑛 ∑ 𝑋𝑖2 − (∑ 𝑋𝑖 )2
3 10.1 102.01 𝑆𝑥 = √ standard deviation, 𝜎,
4 10.3 106.09 𝑛(𝑛 − 1) is unknown, we are
5 10.1 102.01 going to estimate the
6 9.8 96.04 10(1,012.58) − (100.6)2
=√ population standard
7 9.9 98.01 10(9)
8 10.4 108.16 deviation using the
9 10.3 106.09 sample data. That is
10 9.8 96.04 5.44 the sample standard
Total (∑) 100.6 1,012.58 =√ deviation, S, using
90
either of these two
Or you can use this another formula to get the 𝑆𝑥 = 0.245854518 formulas.
sample standard deviation, S,
𝑺𝒙 ≅ 𝟎. 𝟐𝟓 (sample standard deviation)
∑𝑛𝑖=1(𝑥𝑖
−𝑋 ̅ )2
𝑆𝑥 =
𝑛−1

Now, we compute our test statistic, t:

𝑥̅ − 𝜇𝑜 10.06 − 10
𝑡𝑐 = 𝑠 =
⁄ 𝑛 0.245854518⁄
√ √10
0.06
= 0.245854518
⁄
√10
0.06
= 0.077746024

= 0.771743645
𝒕𝒄 ≅ 𝟎. 𝟕𝟕𝟐

Since our test is two tailed or Ha is 𝜇 ≠ 𝜇𝑜 , we will use

d) Decision rule: this region of rejection based on table 15, case 2.

Reject Ho if |𝑡𝑐 | ≥ 𝑡𝛼, (𝑣=𝑛−1) = |𝑡𝑐 | ≥ 𝑡0.01, 𝑣=9 = |𝑡𝑐 | ≥ 𝑡0.005, 𝑣=9
2 2
Reject Ho if |𝑡𝑐 | ≥ 3.250, (based on the t-table, Table B.2)
e) Decision on Ho:

Since |𝑡𝑐 | = 0.772 is not greater than or equal to 3.250, we do not reject the null hypothesis.
f) Conclusion:

There is insufficient evidence to say that the average content of containers of a particular
lubricant is not equal to 10 liters.

Now, using the 𝑝 − 𝑣𝑎𝑙𝑢𝑒 in performing this test:

Since the test is two-tailed, we will have

𝑝 − 𝑣𝑎𝑙𝑢𝑒 = 𝑃(|𝑇| ≥ 0.772|𝐻𝑜 is true), at 𝑣 = 9

= 𝑃(𝑇 ≤ −0.772|𝐻𝑜 is true) + 𝑃(𝑍 ≥ 0.772|𝐻𝑜 is true)
= 2𝑃(𝑇 ≥ 0.772|𝐻𝑜 is true), in terms of area above
> 2 (0.10)
> 0.2 (when using the T-table)
 With 𝑝 − 𝑣𝑎𝑙𝑢𝑒 > 0.2 not lesser than 𝛼 = 0.01, we do not reject the null hypothesis that
the average content of containers of a particular lubricant is 10 liters. This is the same with
our earlier decision on the null hypothesis using the critical value approach.

Note that, using excel, you can compute the exact value of the 𝑃(𝑇 ≥ 0.772) using the function
=T.DIST.RT(x,deg_freedom).
So, in one cell you type =T.DIST.RT(0.772,9) and then it will give you the value of 𝛼 or the
𝑷(𝑻 ≥ 𝟎. 𝟕𝟕𝟐) ≅ 𝟎. 𝟐𝟑 when round-off to 2 decimal places. So, the exact p-value of this problem
is:
𝑝 − 𝑣𝑎𝑙𝑢𝑒 = 𝑃(|𝑇| ≥ 0.772|𝐻𝑜 is true), at 𝑣 = 9
= 𝑃(𝑇 ≤ −0.772|𝐻𝑜 is true) + 𝑃(𝑍 ≥ 0.772|𝐻𝑜 is true)
= 2𝑃(𝑇 ≥ 0.772|𝐻𝑜 is true), in terms of area above
= 2 (0.23)
= 𝟎. 𝟒𝟔
We will also construct a 99% confidence interval on the population mean as follows (from Module
7, Lesson 7.2) and then test the null hypothesis after:

Recall case 2:
𝑆 𝑆 𝑆
(𝑋̅ − 𝑡𝛼,(𝑣=𝑛−1) , 𝑋̅ + 𝑡𝛼,(𝑣=𝑛−1) ) ↔ 𝑋̅ ± 𝑡𝛼,(𝑣=𝑛−1)
2 √𝑛 2 √𝑛 2 √𝑛

Applying this here, we will then have the following 99% confidence interval on the population
∑ 𝑋𝑖 100.6
mean with 𝑡0.01,(𝑣=10−1) = 𝑡0.005,(𝑣=9) = 3.250 (from Table B.2), sample mean, 𝑥̅ = = =
2
𝑛 10
10.06 and standard deviation,𝑆𝑥 = 0.245854518.

10.06 ± (3.250)(0.077746024) ≅ 0.2527 ↔ (𝟗. 𝟖𝟏, 𝟏𝟎. 𝟑𝟏)

We observe that 𝜇𝑜 = 10 does fall in this interval. Thus, we do not reject the null hypothesis
based on interval estimation. We can say that we are 99% confident that the true mean lies in the
interval constructed.

Important note:
Table 15. Summary of hypothesis tests for the population mean when sampling from a
normal population
Alternative
Null Hypothesis Region of
Hypothesis Test Statistic
𝑯𝒐 Rejection
𝑯𝒂
Case 1: 𝜎 is known 𝑧 ≤ −𝑧𝛼
𝜇 = 𝜇𝑜 𝜇 < 𝜇𝑜 𝑋̅ − 𝜇𝑜
𝜇 > 𝜇𝑜 𝑍= 𝑧 ≥ 𝑧𝛼
𝜇 ≠ 𝜇𝑜 𝜎/√𝑛 |𝑧| ≥ 𝑧𝛼
2
Case 2: 𝜎 is not 𝜇 < 𝜇𝑜 𝑡 ≤ −𝑡𝛼,(𝑣=𝑛−1)
known and 𝑛 is 𝜇 > 𝜇𝑜 𝑋̅ − 𝜇𝑜 𝑡 ≥ 𝑡𝛼,(𝑣=𝑛−1)
small 𝜇 ≠ 𝜇𝑜 𝑇=
𝑆/√𝑛 |𝑡| ≥ 𝑡𝛼,(𝑣=𝑛−1)
𝜇 = 𝜇𝑜 2
Case 3: 𝜎 is not 𝜇 < 𝜇𝑜 𝑧 ≤ −𝑧𝛼
known and 𝑛 is 𝜇 > 𝜇𝑜 𝑋̅ − 𝜇𝑜 𝑧 ≥ 𝑧𝛼
large 𝜇 ≠ 𝜇𝑜 𝑍=
𝑆/√𝑛 |𝑧| ≥ 𝑧𝛼
𝜇 = 𝜇𝑜 2

Note: Our region of rejection or decision rule is always based on the direction of our alternative
hypothesis (Ha). This means based on Table 15, for case 1, if your Ha is 𝝁 < 𝝁𝒐 , your decision
rule would be, you reject Ho if 𝒛 ≤ −𝒛𝜶. On the other hand, if your Ha is 𝝁 > 𝝁𝒐 , your decision
rule would be, you reject Ho if 𝒛 ≥ 𝒛𝜶 and if your test is two tailed or your Ha is 𝝁 ≠ 𝝁𝒐 , your
decision rule would be, you reject Ho if |𝒛| ≥ 𝒛𝜶 . It’s the same for the other two cases.
𝟐
Lesson 8.3: Tests for the Population Proportion
Learning Activity
Perform as indicated:
1. A television manufacturer claims in its warranty that in the past, less than 15% of its
television sets needed any repair during their first two years of operation. In order to test the
validity of this claim, a government testing agency selects a sample of 100 sets and finds that
12 sets required some repair within their first years of operation. Is the manufacturer’s claim
valid? Test at 0.01 level of significance. (Almeda et al, 2010) Based on this statement, since it gives
the direction of the unknown value of
Solution: the proportion from the hypothesized
value, we considered this to be our
The null and alternative hypotheses: alternative hypothesis.

𝐻𝑜: 𝑝 = 0.15 , The proportion of the television sets manufactured that needed any repair
during their first two years of operation is 15%.

𝐻𝑎: 𝑝 < 0.15, The proportion of the television sets manufactured that needed any repair
during their first two years of operation is less than 15%.

Desired level of significance, 𝛼 = 0.01

Test statistic: 𝑍, since 𝑝0 is not close to 0 or 1 and n is large

𝑦 12
Given: 𝑝̂ = 𝑛 = 100 = 0.12 (sample proportion)
𝑝0 𝑞𝑜⁄ (0.15)(0.85)
𝑆𝐸(𝑝) = √ 𝑛 = √ 100 = 0.035707142 ≅ 0.036

Note: Please make a correction in your learning guide. The standard error that we compute here
is not an estimated standard error (as written in the learning guide). It is the standard error of the
population proportion because we compute the standard error based on the null given the null is
true. That is why we wrote here 𝑆𝐸(𝑝) instead of 𝑠𝑒(𝑝̂ ).

The value of our test statistic 𝑍 is

𝑝̂ − 𝑝𝑜 0.12 − 0.15 −0.03

𝑧𝑐 = = = = −0.840168053 ≅ −0.84
0.035707142 0.035707142
√𝑝0 𝑞𝑜⁄𝑛

Decision rule and critical region:

Reject 𝐻𝑜 if 𝑧𝑐 ≤ −𝑧𝛼 = −𝑧0.01 = −2.326

[from the information at the bottom part of Table B.1 (Continued)]

Critical region: All value of 𝑍 such that 𝑍 ≤ −2.326.

Note: Our region of rejection or decision rule is always based on the direction of our alternative
hypothesis (Ha). In this case, since our test is one-tailed to the left 𝑝 < 0.15 or (𝑝 < 𝑝𝑜 ) we use
this decision rule, we reject Ho if 𝑧𝑐 ≤ −𝑧𝛼 (refer to Table 16.)

This means that if you Ha is 𝑝 < 𝑝𝑜 , your decision rule would be, you reject Ho if 𝑧 ≤ −𝑧𝛼 . On the
other hand, if your Ha is 𝑝 > 𝑝𝑜 , your decision rule would be, you reject Ho if 𝑧 ≥ 𝑧𝛼 and if your
test is two tailed or your Ha is 𝑝 ≠ 𝑝𝑜 , your decision rule would be, you reject Ho if |𝑧| ≥ 𝑧𝛼
2
Decision on 𝐻𝑜:

Since 𝑧𝑐 = −0.84 is not lesser than -2.326, we do not reject the null hypothesis.

Conclusion:

There is insufficient evidence to say that the proportion of the television sets manufactured
that needed any repair during their first two years of operation is less than 15%.

Note: You always state your conclusion based on the alternative hypothesis. If your decision is
to reject the null hypothesis, Ho, you should always state at what level of significance you rejected
the null, as we know that the level of significance, denoted by 𝛼, is the maximum probability of
committing a type I error (incorrect decision). However, if your decision is not to reject the null
hypothesis, Ho, it is not necessary to mention the level of significance.
Based on the 𝑝 − 𝑣𝑎𝑙𝑢𝑒:

We compute for the 𝑝 − 𝑣𝑎𝑙𝑢𝑒 in this one-tailed test by

𝑝 − 𝑣𝑎𝑙𝑢𝑒 = 𝑃(𝑍 ≤ −0.84|𝐻𝑜 is true)

= 0.2005

Since 𝑝 − 𝑣𝑎𝑙𝑢𝑒 = 0.2005 is not lesser than 𝛼 = 0.01, we do not reject the null hypothesis,
just like with using the critical value approach. We cannot say that the proportion of the
television sets manufactured that needed any repair during their first two years of
operation is less than 15%.
 We will also construct a 99% confidence interval on the population proportion as follows (from
Module 7, Lesson 7.2) and then test the null hypothesis after:

Using the formula that approximate 100(1- 𝛼)% confidence interval for the population
proportion, 𝑝

Estimated standard 𝑃̂(1 − 𝑃̂) 𝑃̂(1 − 𝑃̂ ) 𝑃̂(1 − 𝑃̂)

error of the sample (𝑃̂ − 𝑧𝛼 √ , 𝑃̂ + 𝑧𝛼 √ ) ↔ 𝑃̂ ± 𝑧𝛼 √
2 𝑛 2 𝑛 2 𝑛
proportion, 𝑠𝑒(𝑝̂ )
𝑦 12
with 𝑛 = 100, 𝑧0.01 = 𝑧0.005 = 2.575, 𝑃̂ = 𝑛 = 100 ≅ 0.12 and (1 − 𝑃̂) = 0.88 , we will have
2
the following 99% confidence interval on the population proportion:

0.12 ± (2.575)(0.032496153) ≅ 0.0837 ↔ (0.0363, 0.2037)

We observe that 𝑝𝑜 = 0.15 fall in this interval. Thus, we also do not reject the null
hypothesis based on interval estimation. We can say that we are 99% confident that the
true proportion lies in the interval constructed.
2.575
 It means 𝑝 = 0.20, thus it
would be our Ho.

2. A five-year old census recorded that 20% of the families in a large community lived below the
poverty level. To determine if this percentage has changed, a random sample of 400 families
is studied and 70 are found to be living below the poverty level. Does this finding indicate that
the current percentage of families earning incomes below the poverty level has changed from
what it was five years ago? Perform a two-tailed test and set 𝛼 = 0.05. (Johnson and
Bhattacharya, 2011)
meaning, 𝑝 ≠ 0.20, the statement did not
specify the direction of the unknown
Solution: population proportion, that is why we just
The null and alternative hypotheses: used the not equal sign.

𝐻𝑜: 𝑝 = 0.20 , The current percentage of families earning incomes below the poverty
level did not change from what it was five years ago.

𝐻𝑎: 𝑝 ≠ 0.20, The current percentage of families earning incomes below the poverty
level has changed from what it was five years ago.

Step 2: Desired level of significance, 𝛼 = 0.05

Step 3: Test statistic: 𝑍, since 𝑝0 is not close to 0 or 1 and 𝑛 can be considered large.
𝑦 70
Given: 𝑛 = 400; 𝑝̂ = 𝑛 = 400 = 0.175 (sample proportion)

𝑝0 𝑞𝑜⁄ (0.20)(0.80)
𝑆𝐸(𝑝) = √ 𝑛 = √ 400 = 0.02

The value of our test statistic 𝑍 is

𝑝̂ − 𝑝𝑜 0.175 − 0.20 −0.025

𝑧𝑐 = = = = −1.25
0.02 0.02
√𝑝0 𝑞𝑜 ⁄𝑛

Step 4: Decision rule and critical region:

Reject 𝐻𝑜 |𝑧𝑐 | ≥ 𝑍𝛼 = 𝑍0.05 = 𝑍0.025 = 1.96

2 2
[from the information at the bottom part of Table B.1 (Continued)]

Thus, we reject 𝐻0 if |𝑧𝑐 | ≥ 1.96

Critical region: All values of 𝑍 such that 𝑍 ≤ −1.96 and 𝑍 ≥ 1.96

Step 5: Decision on 𝐻𝑜:

Since |𝑧𝑐 | = 1.25 is not greater than or equal 1.96, we cannot reject the null
hypothesis.

Step 6: Conclusion:
There is no sufficient evidence to say that the current percentage of families earning
incomes below the poverty level has changed from what it was five years ago.

Based on the 𝑝 − 𝑣𝑎𝑙𝑢𝑒:

We compute for the 𝑝 − 𝑣𝑎𝑙𝑢𝑒 in this two-tailed test by

𝑝 − 𝑣𝑎𝑙𝑢𝑒 = 𝑃(|𝑍| ≥ −1.25|𝐻𝑜 is true)

= 𝑃(𝑍 ≤ −1.25|𝐻𝑜 is true) + 𝑃(𝑍 ≥ 1.96|𝐻𝑜 is true)
= 2𝑃(𝑍 ≤ −1.25|𝐻𝑜 is true), in terms of area below
= 2 (0.1056)
= 0.2112
Since 𝑝 − 𝑣𝑎𝑙𝑢𝑒 = 0.2112 is not lesser than 𝛼 = 0.05, we do not reject the null hypothesis,
just like with using the critical value approach. We can say that the current percentage of
families earning incomes below the poverty level did not change from what it was five
years ago.
We will also construct a 95% confidence interval on the population proportion as follows (from
Module 7, Lesson 7.2) and then test the null hypothesis after:

Using the formula that approximate 100(1- 𝛼)% confidence interval for the population
proportion, 𝑝

𝑃̂(1 − 𝑃̂) 𝑃̂(1 − 𝑃̂ ) 𝑃̂(1 − 𝑃̂)

(𝑃̂ − 𝑧𝛼 √ , 𝑃̂ + 𝑧𝛼 √ ) ↔ 𝑃̂ ± 𝑧𝛼 √
2 𝑛 2 𝑛 2 𝑛

𝑦 70
with 𝑛 = 400, 𝑧0.05 = 𝑧0.025 = 1.96, 𝑃̂ = 𝑛 = 400 ≅ 0.175 and (1 − 𝑃̂) = 0.825 , we will have
2
the following 95% confidence interval on the population proportion:

0.175 ± (1.96)(0.018998355) ≅ 0.0372 ↔ (0.1378, 0.2122)

We observe that 𝑝𝑜 = 0.20 does fall in this interval. Thus, we also cannot reject the null
hypothesis based on interval estimation. We can say that we are 95% confident that the
true proportion lies in the interval constructed.

Thank you!

If you still have questions or clarifications regarding the topic, just send me a message.

-MRLebante-