Chapter 4 Lesson 3: Estimating Population Proportion (p) for The Large Sample

Size
Sometimes, we are concerned, and so population that share common
characteristics. What proportion of the population are senior citizens? How many percent
of the population have IQ's greater than 120? These are some of the questions which
require population proportion (p). In this chapter we will be dealing with population
proportion using large sample size (𝑛 ≥ 30) . Estimating population proportions using
small sample size is not included. Estimating population proportion is similar to estimating
population mean. When the sample proportion 𝑝̂ is computed from the large sample n,
then the interval estimate of the population proportion p at certain 𝛼 can be computed
as:

𝑝̂ 𝑞̂ 𝑝̂ 𝑞̂
𝑝̂ − 𝑧𝛼⁄2 √ < 𝑝 < 𝑝̂ + 𝑧𝛼⁄2 √
𝑛 𝑛

Where :
𝑝̂ = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑝𝑟𝑜𝑝𝑜𝑟𝑡𝑖𝑜𝑛
𝑞̂ = 1 − 𝑝̂
𝑛 = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑠𝑖𝑧𝑒
𝑧𝛼⁄2 = 𝑧 𝑣𝑎𝑙𝑢𝑒 𝑡ℎ𝑎𝑡 𝑙𝑒𝑎𝑣𝑒𝑠 𝑎𝑛 𝑎𝑟𝑒𝑎 𝑜𝑓 𝛼⁄2

𝑝̂ 𝑞̂
𝑧𝛼⁄2 √ = 𝑚𝑎𝑟𝑔𝑖𝑛 𝑜𝑓 𝑒𝑟𝑟𝑜𝑟(𝑀𝐸)
𝑛

𝑝̂ 𝑞̂
√ = 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑒𝑟𝑟𝑜𝑟 (𝑆𝐸)
𝑛

The confidence interval of the difference of two proportion is given by:

𝑝̂1 𝑞̂1 𝑝̂2 𝑞̂2 𝑝̂1𝑞̂1 𝑝̂ 2 𝑞̂2

|𝑝̂1 − 𝑝̂ 2 | − 𝑧𝛼⁄ √ + < (𝑝1 − 𝑝2 ) < |𝑝̂1 − 𝑝̂ 2 | + 𝑧𝛼⁄2 √ +
2 𝑛1 𝑛2 𝑛1 𝑛2

Formula for estimating a sample size n of a population proportion:

𝑝̂ 𝑞̂
𝑛= (𝑧𝛼⁄2 )2
𝑀𝐸 2
Example 1:
200 randomly selected workers were asked whether they believe that the
country's economy will improve under the new president. One hundred twenty of them
said yes. Construct a 90% confidence interval for the proportion of workers who believe
that the economy will improve.
𝑥 120
𝑝̂ = =
𝑛 200
120
Solution: x = 120, 𝑝̂ = = 0.6, thus, 𝑞̂ = 1 − 0.6 = 0.4 ,
200

Confidence level = 90% , 𝛼 = 0.10, 𝛼⁄2 = 0.05, 𝑧𝛼⁄2 = 1.64

𝑧𝛼⁄2 : 0.5 − 0.05 = 0.4500

𝑝̂ 𝑞̂
𝑝 = 𝑝̂ ± 𝑧𝛼⁄2 √
𝑛

0.6 ∙ 0.4
𝑝 = 0.6 ± 1.64√
200

0.6 ∙ 0.4
𝑝 = 0.6 ± 1.64√
200

𝑝 = 0.6 ± 1.64 ∙ 0.034

𝑝 = 0.6 ± 0.056
𝑝 = (0.544, 0.656)
54.4% < 𝑝 < 65.6%
Interpretation: We are 90% confident that about 54.4% to 65.6 % of the workers
believe that the economy will improve under the new president.
Example 2:
Dr Doof discovered a new treatment in increasing the cure rate of skin diseases.
He claims that it is more effective than the existing standard treatment with a 65% cure
rate based on a study on 100 patients. He stressed that out of 30 patients with skin
diseases, 22 were totally cured using his treatment. Is the claim of Dr Doof true? Get the
95% confidence interval for the difference in the two proportions.

𝑝̂1 𝑞̂1 𝑝̂2 𝑞̂2 𝑝̂1𝑞̂1 𝑝̂ 2 𝑞̂2

|𝑝̂1 − 𝑝̂ 2 | − 𝑧𝛼⁄ √ + < (𝑝1 − 𝑝2 ) < |𝑝̂1 − 𝑝̂ 2 | + 𝑧𝛼⁄2 √ +
2 𝑛1 𝑛2 𝑛1 𝑛2

Solution: 𝑝̂1 = 0.65, 𝑞̂1 = 0.35; 𝑝̂2 = 0.73, 𝑞̂2 = 0.27; 𝑛1 = 100, 𝑛2 = 30
Confidence level: 95% 𝛼 = 0.05, 𝛼⁄2 = 0.025, 𝑧𝛼⁄2 = 1.96

𝑧𝛼⁄2 : 0.5 − 0.025 = 0.475

𝑝̂1 𝑞̂1 𝑝̂ 2 𝑞̂2

(𝑝1 − 𝑝2 ) = |𝑝̂1 − 𝑝̂ 2 | ± 𝑧𝛼⁄ √ +
2 𝑛1 𝑛2

0.65 ∙ 0.35 0.73 ∙ 0.27

(𝑝1 − 𝑝2 ) = |0.65 − 0.73| ± 1.96√ +
100 30

(𝑝1 − 𝑝2 ) = 0.08 ± 0.184

(𝑝1 − 𝑝2 ) = (− 0.104, 0.264)

−0.104 < (𝑝1 − 𝑝2 ) < 0.264

Therefore, the 95% confidence interval of the difference between the two
proportions is(− 0.104, 0.264).
Interpretation: since the confidence interval ranges from a negative to a positive value,
then zero lies within the interval. This means that there is a possibility of getting a zero
difference in the proportion of patients that the two treatments are able to cure, or that
there is no significant difference between the existing standard treatment and the new
treatment.
Example 3:
A political campaign manager wishes to survey a number of voters to estimate the
proportion of those who are in favor of his candidate. If a previous survey shows that 55%
of the registered voters plans to vote for his candidate, what is the minimum sample size
required to make his surveys accurate with a 95% confidence level and a margin of error
of 2.5%?
𝑝̂ 𝑞̂
𝑛= (𝑧𝛼⁄2 )2
𝑀𝐸 2
Solution: 𝑝̂ = 0.55, 𝑞̂ = 0.45, ME = 0.025
Confidence level: 95% 𝛼 = 0.05, 𝛼⁄2 = 0.025, 𝑧𝛼⁄2 = 1.96

𝑧𝛼⁄2 : 0.5 − 0.025 = 0.475

0.55 ∙ 0.45
𝑛= (1.96)2 =1521.2736 ≈ 1522
0.0252
Therefore, the campaign manager needs to survey at least 1522 registered voters.
 Chapter V: Hypothesis Testing
Lesson 1: Introduction to Hypothesis Testing

A hypothesis is a proposed explanation, assertion, or assumption about a

population parameter or about the distribution of a random variable. It is a supposition,
advanced as a basis for argument. It is tested using statistical methods, generically using
experimental samples.

A statistical hypothesis is a conjecture about a population parameter. This

conjecture may or may not be true.
The question of interest or under test is simplified into two competing
contradictory statements: The null hypothesis, denoted by 𝑯𝟎 , against the alternative
hypothesis, denoted by 𝑯𝟏 /𝑯𝒂 . These two hypotheses, however, are not treated on an
equal basis. Special consideration is given to the null hypothesis because it is the
hypothesis to be tested as to whether it should be rejected or not. The alternative
hypothesis, from the word itself, will be the choice if the null hypothesis were to be
rejected.
The null hypothesis (𝐻0 )is a statement of “zero” difference. Here, assumes that
there is no difference between two means (the population mean and sample mean) or
variables being compared. The null hypothesis (𝐻0 ) is a claim that denotes “absence”
such as absence of difference, absence of relationship, or equality to a certain value,
and the like. It usually comes with " =, ≥, 𝑜𝑟 ≤ " when written in symbol.
On the other hand, the alternative hypothesis (𝐻1 /𝐻𝑎 )is a statement that
assumes there is a significant difference between the two means or variables under test
of investigation. The alternative hypothesis (𝐻1 /𝐻𝑎 ) is a claim that denotes “presence”,
such as presence of difference, presence of relationship or inequality to a certain value
and the like. It usually comes with " ≠, >, 𝑜𝑟 < " when written in symbol.
Two-tailed test Right-tailed test Left-tailed test
𝐻0 : 𝜇 = 𝑘 𝐻0 : 𝜇 ≤ 𝑘 𝐻0 : 𝜇 ≥ 𝑘
𝐻1 : 𝜇 ≠ 𝑘 𝐻1 : 𝜇 > 𝑘 𝐻1 : 𝜇 < 𝑘
 Decision 𝐻0 𝑖𝑠 𝑡𝑟𝑢𝑒 Actual condition
𝐻1 𝑖𝑠 𝑡𝑟𝑢𝑒
Reject 𝐻0 Type I error Correct decision
Accept 𝐻0 Correct decision Type II error

After testing hypothesis, of course a decision shall be made as basis for a

conclusion that is to reject or not to reject the null hypothesis. In making a decision, 4
results can be made, 2 right decisions and two wrong decisions.
1. The null hypothesis is rejected when, in fact, it is true.
2. The null hypothesis is rejected when, in fact, it is false.
3. The null hypothesis is accepted when, in fact, it is true.
4. The null hypothesis is accepted when, in fact, it is false.

Steps in Testing Hypothesis (Critical Value Method)

1. Determining the null hypothesis and alternative hypothesis.
2. Identifying the statistical test to be used, the value of 𝛼, and the critical value of the
test statistic.
3. Computing the test statistic.
4. Making decision (reject or not to reject the 𝐻0 ).
5. Conclusion (in non-technical terms)

Example 1: In each of the following situations, answer the following:

a. State the null and alternative hypothesis in words.
b. State the null and alternative hypothesis in symbols.
c. What is the probability of committing type I error?
d. State the conclusion when H0 is rejected.
e. State the conclusion when H0 is not rejected.
1) A social worker wants to test (𝛼 = 0.05) whether the average body mass index (BMI)
of the pupils under feeding program is different from 28.2 kg.
a. H0 : The average body mass index (BMI) of the pupils under feeding program is not
different from 28.2 kg.
H1 : The average body mass index (BMI) of the pupils under feeding program is
different from 28.2 kg.
b. H0 𝜇 = 28.2
H1 𝜇 ≠ 28.2
c. The probability of committing type I error is 5%.
d. Therefore, the average body mass index (BMI) of the pupils under feeding program is
different from 28.2 kg.
e. Therefore, the average body mass index (BMI) of the pupils under feeding program is
not different from 28.2 kg. / Therefore, the average body mass index (BMI) of the pupils
under feeding program is 28.2 kg.

2) A DTI representative wants to test at 99% confidence level whether the average
content of soda X is less than 330mL as indicated in the label.
a. H0 : The average content of soda X is greater than or equal to 330mL.
H1 : The average content of soda X is less than 330mL.
b. H0 𝜇 ≥ 330
H1 𝜇 < 330
c. The probability of committing type I error is 0.01 / 1%.
d. Therefore, the average content of soda X is less than 330mL.
e. Therefore, the average content of soda X is greater than or equal to 330mL
3) An FDA officer claims that Pharma XYZ's new caplet drug contains greater than 250
mg of paracetamol. Test the claim using 99% confidence level.
a. H0 : Pharma XYZ's new caplet drug contains less than or equal 250 mg of paracetamol
H1 : Pharma XYZ's new caplet drug contains greater than 250 mg of paracetamol.
b. H0 𝜇 ≤ 250
H1 𝜇 > 250
c. The probability of committing type I error is 0.01 / 1%.
d. Therefore, Pharma XYZ's new caplet drug contains greater than 250 mg of
paracetamol.
e. Therefore, Pharma XYZ's new caplet drug contains less than or equal 250 mg of
paracetamol.

Testing Hypothesis about Population Mean

One of the most common tests for population mean is called the z-test, which
uses the properties of z-distribution or normal distribution when the population
standard deviation (𝜎) is known. This is also used when the sample size n is greater than
or equal to 30 (𝑛 ≥ 30) by virtue of central limit theorem.
𝑥̅ − 𝜇0
𝑧=
𝜎 ⁄√𝑛
Where:
𝜇0 = 𝑐𝑙𝑎𝑖𝑚𝑒𝑑 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 mean
𝜎 = 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 (𝑐𝑎𝑛 𝑏𝑒 𝑟𝑒𝑝𝑙𝑎𝑐𝑒𝑑 𝑏𝑦 𝑠 𝑖𝑓 𝑛 < 30)
𝑥̅ = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑚𝑒𝑎𝑛
𝑛 = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑠𝑖𝑧𝑒
Decision Rule:
✓ If zcomputed ≥ zcritical Reject H0
✓ If zcomputed< zcritical Do Not Reject H0

If the population standard deviation or variance is unknown, the z-test cannot be

used. For a special case where the population from which the samples are taken is
known to be normally distributed, the t-test can be used to test a claim or hypothesis
about a population mean. The formula for t-test is:
𝑥̅ − 𝜇0
𝑡=
𝑠⁄√𝑛
𝜇0 = 𝑐𝑙𝑎𝑖𝑚𝑒𝑑 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛
𝑠 = 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛
𝑥̅ = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑚𝑒𝑎𝑛
𝑛 = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑠𝑖𝑧𝑒

Decision Rule:
✓ If tcomputed ≥ tcritical Reject H0
✓ If tcomputed< tcritical Do Not Reject H0

Example 1: Determine the decision for each of the following given the computed and
critical value for z or t:
zcomputed zcritical decision tcomputed tcritical decision
Do Not Reject H0 Reject H0
1.82 1.96 2.02 1.771
Reject H0 Do Not Reject H0
2.54 2.33 2.24 2.552
Example 2: Determine the decision for each of the following given the computed z or t.
a. zcomputed = 1.29 confidence level = 90%, two-tailed
b. zcomputed = 1.87 𝛼 = 0.05, one-tailed
c. tcomputed = 2.84 confidence level = 99%, one-tailed, n =21
d. tcomputed = 2.58 𝛼 = 0.05, two-tailed, n=17

Solution:
a. zcomputed = 1.29 confidence level = 90%, two-tailed

𝛼
𝛼 = 0.10, = 0.05 zcritical = ±1.64
2

0.5000 – 0.0500 = 0.4500

Decision: Do not reject H0 zcomputed = 1.29 < zcritical = ±1.64
b. zcomputed = 1.87 𝛼 = 0.05, one-tailed

z-critical = 1.64
0.5-0.05= 0.4500, z = 1.64
Decision: Reject H0

c. tcomputed = 2.84 confidence level = 99%, one-tailed, n =21

df= n-1 = 21-1=20, df = 20 @ 𝛼 = 0.01
t-critical = 2.528

Decision: Reject the H0

d. tcomputed = 2.58 𝛼 = 0.05, two-tailed, n=17
df = 16, 𝛼 = 0.05, two-tailed t-critical = 2.120 Decision: Reject the H0
Example 3: A printer manufacturing company claims that its new ink-efficient printer
can print an average of 1500 pages of word documents with standard deviation of 60.
Thirty-five (35) of these printers showed a mean of 1475 pages. Does this support the
company's claim. Use 95% confidence level.
𝜇 = 1500
𝑥̅ = 1475
𝑛 = 35
𝜎 = 60
Confidence level = 95%
Solution:
1. Determining the null hypothesis and alternative hypothesis.
H0: The new ink-efficient printer can print an average of 1500 pages. (CLAIM)
H1: The new ink-efficient printer cannot print an average of 1500 pages.
H0: 𝜇 =1500
H1: : 𝜇 ≠1500
2. Identifying the statistical test to be used, the value of 𝛼, and the critical value of the
test statistic.
✓ n = 35; statistical test: z-test, two-tailed
✓ 𝛼 = 0.05, confidence level = 95%
✓ 𝑧 − 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 = ±1.96

0.5000 – 0.025 = 0.4750

3. Computing the test statistic.
𝑥̅ − 𝜇0 1475 − 1500 −25
𝑧= = = = −2.47
𝜎 ⁄√𝑛 60⁄√35 10.14

4. Making decision (reject or not to reject the 𝐻0 ).

Since the computed z is greater than the z-critical, therefore, reject the null hypothesis.
𝑧 − 𝑐𝑜𝑚𝑝𝑢𝑡𝑒𝑑 = −2.47 > 𝑧 − 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 = ±1.96
Decision: Reject H0
5. Conclusion (in non-technical terms)
Therefore, the new ink-efficient printer cannot print an average of 1500 pages.
There is no enough evidence to support the claim of the company that the new printer
can print 1500 pages.
There is a sufficient evidence to deny the claim of the company that the new printer can
print 1500 pages.

Testing Hypothesis about Proportion

To test a claim about population proportion, we use the Z test for population
proportion. The former below is used:
𝑝̂ − 𝑝
𝑧=
√𝑝𝑞/𝑛
Where:
p = claimed/ hypothesized proportion
𝑝̂ = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑝𝑟𝑜𝑝𝑜𝑟𝑡𝑖𝑜𝑛
𝑞 = 1−𝑝
n = sample size
 Example 4: The researcher wants to test his belief that 50% of the population of
rats is female. From his collected samples, 23 out of 50 are females. Will this support his
claim? The confidence level is 95%.

1. Determining the null hypothesis and alternative hypothesis.

H0: The population of female rats is 50%. (claim)
H1: The population of female rats is not 50%.
H0: 𝑝 = 0.50
H1: 𝑝 ≠ 0.50

2. Identifying the statistical test to be used, the value of 𝛼, and the critical value of the
test statistic.
z-test, two-tailed test (non-directional)
𝛼 = 0.05, 𝑐𝑜𝑛𝑓𝑖𝑑𝑒𝑛𝑐𝑒 𝑙𝑒𝑣𝑒𝑙 𝑜𝑓 95%; 𝑧 − 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 = ±1.96

3. Computing the test statistic.

23
𝑝̂ = = 0.46
50
𝑝̂ − 𝑝 0.46 − 0.5 −0.04
𝑧= = = = −0.56
√𝑝𝑞/𝑛 √(0.50)(0.50)/50 0.071
4. Making decision (reject or not to reject the 𝐻0 ).
𝑧 − 𝑐𝑜𝑚𝑝𝑢𝑡𝑒𝑑 = −0.56 < 𝑧 − 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 = ±1.96
Decision: Do not reject H0. Since the z-computed is less than the z-critical, therefore, do
not reject the null hypothesis.

5. Conclusion (in non-technical terms)

The population of the female rats is 50%.

Therefore, there is enough evidence to support the claim of the researcher that
the population of the female rats is 50%.
Therefore, there is no sufficient evidence deny the claim of the researcher. Thus,
50% of the population of the rats is female.