OpenCourseWare | [Link].

my
Universiti Teknikal Malaysia Melaka

Understanding
Sampling
Distributions
BENG 2143
Engineering Statistics

DR FARAH SHAHNAZ FEROZ

shahnaz@[Link]
 OpenCourseWare | [Link]
Universiti Teknikal Malaysia Melaka

Calculate probabilities of sample

1 means for samples obtained from

populations that are normally
distributed.

2
Apply the concepts of central
tendency for large samples
obtained from populations that are
not normally distributed.

LESSON OUTCOMES 3 Calculate probabilities of sample

proportions for large samples.

4 Apply the concepts of sampling

distributions to solve mathematical
problems.
Review: Why sample?
 Review: Population Distribution
Probability distribution of the population data
e.g. If there are only 3 students in a Masters Degree class and the
number of siblings that they have is as below:
Student Amy Ben Harry
Number of siblings 2 2 4
random variable
The population distribution is as follows:
𝑿 Frequency (f) Relative 𝑷(𝑿 = 𝒙)
Frequency (rf)
2 2 0.67 0.67 values of
random variable
the random
4 1 0.33 0.33
variable
Total 3 1 1
Sampling Distribution of the Sample Mean
When an experiment is conducted repeatedly with ALL possibilities of 𝑥ҧ
e.g: Sampling distribution of 𝑥ҧ with a sample size of 2 is as follows:

Student Samples 𝑥ҧ 𝑥ҧ f p(𝑥)ҧ

A,B 2,2 2 2 1 1
= 0.33
B,C 2,4 3 3
3 2 2
C,A 4,2 3 = 0.67
3
3 1
Mean and Standard Deviation of the Sampling
Distribution of the Sample Mean
Population Sample
Mean 𝜇 𝑥ҧ
Standard 𝜎 𝑠
Deviation

Mean of the sample mean, 𝜇𝑥ҧ = 𝜇

𝜎
Standard deviation of the sample mean, 𝜎𝑥ҧ =
𝑛
 Example
A machine fills cola in bottles with a mean of 250 ml and a standard deviation of
0.5 ml. A quality assurance inspector takes a sample of 20 bottles once a week
and measures the amount of cola in these bottles.

Find the mean and standard deviation of the sample mean for these cola
bottles.

𝜇𝑥ҧ = 𝜇 = 250 𝑚𝑙

𝜎 0.5
𝜎𝑥ҧ = = = 0.1118 𝑚𝑙
𝑛 20
Shape of the Sampling Distribution of the
Sample Mean
Population is normally distributed Population is NOT normally distributed

𝜇𝑥ҧ = 𝜇 Central Limit Theorem =>

𝜎 for large samples (𝑛 ≥ 30), the sampling
𝜎𝑥ҧ =
𝑛 distribution of the sample mean is
approximately normal. Thus,
𝑥ҧ − 𝜇𝑥ҧ 𝑥ҧ − 𝜇 𝜇𝑥ҧ = 𝜇
𝑍= =𝜎 𝜎
𝜎𝑥ҧ ൗ 𝑛 𝜎𝑥ҧ =
𝑛

𝑥ҧ − 𝜇𝑥ҧ 𝑥ҧ − 𝜇
𝑍= =𝜎
𝜎𝑥ҧ ൗ 𝑛
Example
A random sample of 50 loaves of bread is taken from a bakery in
Hamburg this morning. It was found that the weight of breads are
normally distributed with a mean of 0.5 kg with a standard deviation
of 0.06 kg. What is the probability that the mean weight of breads is:
a) between 0.48 and 0.51 kg Step 0: 𝜇 = 0.5 𝑘𝑔, 𝜎 = 0.06 𝑘𝑔 & 𝑛 = 50
𝑃 0.48 < 𝑋ത < 0.51 Step 1: transform words to math symbols!
0.48−0.5 0.51−0.5 ҧ 𝑥ഥ
𝑥−𝜇 ҧ
𝑥−𝜇
𝑃 0.06 <𝑍< 0.06 Step 2: standardize with 𝑍 =
𝜎𝑥ഥ
=𝜎
ൗ 50 ൗ 50 ൗ 𝑛

𝑃(−2.3570 < 𝑍 < 1.1785) Step 3: Solve!

Using calculator: 𝑃 1.1785 − 𝑃 −2.3570 = 0.8807 − 0.0092 = 0.8715

A random sample of 50 loaves of bread is taken from a bakery in
Hamburg this morning. It was found that the weight of breads are
normally distributed with a mean of 0.5 kg with a standard deviation of
0.06 kg. What is the probability that the mean weight of breads is:
b) less than the population mean by 0.03 kg or more
Step 0: 𝜇 = 0.5 𝑘𝑔 & 𝜎 = 0.06 𝑘𝑔
𝑃 𝑋ത < 0.47
0.03 0.47 − 0.5
=𝑃 𝑍<
0.06ൗ
0.47 0.5 50
= 𝑃 𝑍 < −3.5356 = 0.0002
Example
The lives of 1.5 V batteries from VF Corporation is normally distributed
with a mean of 10 000 hours and standard deviation of 5000 hours. Find
the sample size, n selected by the QA Engineer, if he found that the
probability that the mean life of the batteries is less than the population
mean by 500 hours or more is 0.0655.

500
Solution:
Step 0: 𝜇 = 10000 hrs , 𝜎 = 5000 hrs & 𝑛 =?
9500 10 000
Example
Step 0: 𝜇 = 10000 hrs , 𝜎 = 5000 hrs & 𝑛 =?

Step 1: 𝑃 𝑋ത < 9500 = 0.0655

9500−10000
Step 2: 𝑃 𝑍< 5000 = 0.0655
ൗ 𝑛

From table, 𝑃 𝑍 < −1.51 = 0.0655. Thus:

500

−500
= −1.51
5000ൗ 9500 10 000
𝑛

𝑛 = 15.1000, 𝑛 = 228.01 ≈ 229

Normal distribution or sampling distribution
of the sample mean?
The average weight of meat that a person consumes a year is 180 kg. Assume
that the standard deviation is 18 kg and the distribution is normally
distributed.

a) Find the probability that a person selected at random consumes less than
180 kg meat a year.
𝑃(𝑋 < 180)

b) In a sample of 80 individuals, find the probability that the mean of the

sample will consume less than 188 kg of meat in a year.
𝑃(𝑋ത < 188)
Population and Sample Proportion
Proportion = relative frequency

𝑋 (number of elements of interest in a population)

Population proportion, 𝑝 =
𝑁 (Population size)

Sample proportion, 𝑝Ƹ =
𝑥 (number of elements of interest in a sample)
𝑛 (Sample size)
Example
Suppose a total of 789 654 disks are produced in a company and 111 of
them are defective. A sample of 240 disks is selected from this
company and 11 of them are defective.
What is the proportion of all disks that is defective?
111 𝑋 = 111, 𝑁 = 789 654, 𝑥 = 11, 𝑛 = 240
𝑝=
789 654

What is the sample proportion of defective disks?

11
𝑝Ƹ =
240
Population Distribution of the Proportion
Probability distribution of the population data
e.g. If there are only 4 machines in a factory and the current state of the
machines is as follows:
Machine A B C D
State Working In repair Working Working

The sampling distribution of the sample proportion of machines being

repaired for a sample size of 3 is as follows:
ෝ
𝒑 Frequency (f) Relative 𝐩(ෝ
𝒑)
Frequency (rf)
0 1 0.25 0.25
0.33 3 0.75 0.75
Total 4 1 1
Mean and Standard Deviation of the Sampling
Distribution of the Sample Proportion
Population Sample
Proportion 𝑝 𝑝Ƹ
Size 𝑁 𝑛
Mean 𝜇 𝑥ҧ
Standard 𝜎 𝑠
Deviation

Mean of the sample proportion, 𝜇𝑝ො = 𝑝

𝑝𝑞
Standard deviation of the sample proportion, 𝜎𝑝ො =
𝑛
Central Limit Theorem: Sampling Distribution
of the Sample Proportion
The sampling distribution of 𝑝Ƹ is approximately normal if the sample
size is sufficiently large => 𝑛𝑝 > 5 and 𝑛𝑞 > 5. Thus:

𝑝Ƹ − 𝜇𝑝ො 𝑝Ƹ − 𝑝
𝑍= =
𝜎𝑝ො 𝑝𝑞
𝑛
 Example
Based on previous reports, the proportion of wafers that passed an
inspection test in a semiconductor factory is 0.95. Suppose 400
wafers are randomly sampled.

a) What is the probability that the sample proportion of wafers that

passed the inspection test is more than 0.93?

Step 0: 𝑝 = 0.95, 𝑞 = 0.05 & 𝑛 = 400

 Example
a) What is the probability that the sample proportion of wafers that
passed the inspection test is more than 0.93?
Step 0: 𝑝 = 0.95, 𝑞 = 0.05 & 𝑛 = 400
𝑃 𝑝Ƹ > 0.93 Step 1: transform words to math symbols!
0.93−0.95 ො
𝑝−𝑝
𝑃 𝑍> Step 2: standardize with 𝑍 = 𝑝𝑞
(0.95)(0.05)ൗ
400 𝑛

𝑃(𝑍 > −1.83) = 𝑅 −1.83 = 0.9664 Step 3: Solve!

Example
Based on previous reports, the proportion of wafers that passed an
inspection test in a semiconductor factory is 0.95. Suppose 400
wafers are randomly sampled.
Step 0: 𝑝 = 0.05, 𝑞 = 0.95 & 𝑛 = 400
b) What is the probability that the sample proportion of wafers that
failed the inspection test is less than 7%?
𝑃 𝑝Ƹ < 0.07 Step 1: transform words to math symbols!
0.07−0.05 ො
𝑝−𝑝
𝑃 𝑍< Step 2: standardize with 𝑍 = 𝑝𝑞
(0.95)(0.05)ൗ
400 𝑛

𝑃 𝑍 < 1.83 = P 1.83 = 0.9664 Step 3: Solve!

 Example
Based on previous reports, the proportion of wafers that passed an
inspection test in a semiconductor factory is 0.95. Suppose 400 wafers
are randomly sampled.
Step 0: 𝑝 = 0.95, 𝑞 = 0.05 & 𝑛 = 400
c) What probability of the sample proportion of wafers that passed
inspection lies within 0.01 of the true population proportion?
Step 1: 𝑃 0.94 < 𝑝Ƹ < 0.96
0.94−0.95 0.96−0.95
Step 2: 𝑃 <𝑍< 0.94 0.96
(0.95)(0.05)ൗ (0.95)(0.05)ൗ 0.95
400 400

Step 3: 𝑃 −0.9177 < 𝑍 < 0.9177 = 2 𝑄 −0.9177 = 0.6412

 Example
Based on previous reports, the proportion of wafers that passed an
inspection test in a semiconductor factory is 0.95. Suppose 400 wafers
are randomly sampled.
Step 0: 𝑝 = 0.95, 𝑞 = 0.05 & 𝑛 = 400
c) What probability of the sample proportion of wafers that passed
inspection lies within 0.01 of the true population proportion?
Step 1: 𝑃 0.94 < 𝑝Ƹ < 0.96 -0.01 +0.01

0.94−0.95 0.96−0.95
Step 2: 𝑃 <𝑍< 0.94 0.96
(0.95)(0.05)ൗ (0.95)(0.05)ൗ 0.95
400 400

Step 3: 𝑃 −0.9177 < 𝑍 < 0.9177 = 2 𝑄 −0.9177 = 0.6412

Exercise
The time taken to calibrate all lathe machines is normally distributed
with a mean of 40 minutes and a standard deviation of σ minutes.
i) Find σ if the probability that the time taken to calibrate a lathe
machine for less than 39.15 minutes is 0.32.
ii) A random sample of 30 lathe machines is then taken for a study. Find
the probability that the mean time taken to calibrate a lathe machine is
less than the population mean by at least 2 standard deviations.
iii) Next, find a new sample size for the lathe machines so that there is
a 90% chance for the sample mean to be within 0.5 standard deviation
of the population mean.
Answers: i) 1.8085 ii) 0 iii) 11
Exercise Table Q2 Weight of 30 cartons filled with a
liquid product
Suppose a filling machine is used to fill cartons with a
liquid product. Table Q2 below shows data for the 9.939 9.155 9.798
weight of 30 cartons. Assume that the weight of the 9.732 9.846 9.122
cartons filled with a liquid product is normally
distributed. 9.435 9.135 9.652
(a) What is the probability that on average, the weight 9.263 8.758 9.073
of a carton is more than 9.135 oz? 9.080 8.643 9.534
9.986 9.385 9.518
(b) What is the probability that on average, the weight 9.318 9.024 9.252
of a carton is between 9.125 oz and 9.525 oz?
9.422 9.858 9.626

(c) The specification strictly enforced for the filling 8.124 9.464 9.318
machine is that all cartons need to have a mean weight 9.275 9.025 9.737
of 9.0 ± 0.125 oz. If any carton is produced with weight
outside these bounds, it is considered by the supplier to
be defective. Calculate the probability that a carton
selected is considered defective. Answers:a) 0.9979 b) 0.9889 c) 0.9987
 ARE YOU ABLE TO :
OpenCourseWare | [Link]
Universiti Teknikal Malaysia Melaka

1 2 3 4
Calculate Apply the Calculate Apply the
probabilities of concepts of probabilities of concepts of
sample means central sample sampling
for samples tendency for proportions for distributions to
obtained from large samples large samples? solve
populations that obtained from mathematical
are normally populations that problems?
distributed? are not normally
distributed?

SELF REVIEW
 OpenCourseWare | [Link]
Universiti Teknikal Malaysia Melaka

REFERENCES
Farah Shahnaz Feroz, Nortazi
01 Sanusi, Hanissah Mohamad, A
Student’s Guide to Engineering
Statistics, Penerbit UTeM, 2019

Sharifah Sara, Hanissah,

02 Fauziah, Nortazi, Farah Shahnaz,
Introduction To Statistics &
Probability: A Study Guide,
Pearson-Prentice Hall, 2008
OpenCourseWare | [Link]
Universiti Teknikal Malaysia Melaka

THANK
YOU
DR FARAH SHAHNAZ FEROZ
shahnaz@[Link]