IIE 3017

Quality Engineering

Lecture 4-2: Variables Control Charts

(Recall Sampling Distributions)

Fall 2025
Professor Hyunsoo Yoon
Industrial Engineering at Yonsei University

Lecture 4-2
Contents

Lecture 4-2 2
Central Limit Theorem

Lecture 4-2 3
 8.1-8.2 Random Sampling and Statistic
Definition 8.1
A population consists of the totality of the observations with which we are concerned.

Definition 8.2
A sample is a subset of a population.

Definition 8.3
Let 𝑋1 , 𝑋2 , … , 𝑋𝑛 be 𝑛 independent random variables, each having the same probability distribution 𝑓(𝑥). Define
𝑋1 , 𝑋2 , … , 𝑋𝑛 to be a random sample of size n from the population 𝑓 𝑥 and write its joint probability distribution as
𝑓 𝑥1 , 𝑥2 , … , 𝑥𝑛 = 𝑓(𝑥1 )𝑓(𝑥2 ) ⋯ 𝑓(𝑥𝑛 ).

Definition 8.4
Any function of the random variables constituting a random sample is called a statistic.

Lecture 4-2 4
 Central Limit Theorem
Definition 8.5
The probability distribution of a statistic is called a sampling distribution.

Suppose that a random sample of n observations is taken from a normal population with mean 𝜇 and variance 𝜎 2 .
Then from Theorem 7.11,

1
𝑋ത = (𝑋 + 𝑋2 + ⋯ + 𝑋𝑛 )
𝑛 1
has a normal distribution with

1 1 𝜎2
𝜇𝑋ത = 𝜇 + 𝜇 + ⋯ + 𝜇 = 𝜇 and 𝜎𝑋2ത = 𝜎2 + 𝜎2 + ⋯+ 𝜎2 =
𝑛 𝑛2 𝑛

As 𝑛 approaches ∞, 𝜎𝑋2ത approaches 0.

ഥ −μ𝑥ഥ )
(X ഥ −μ)
(X
Also, Z = = has a normal distribution 𝑛(𝑧; 0, 1)
σX
ഥ σΤ 𝑛

Lecture 4-2 5
Central Limit Theorem
Central Limit Theorem

Although a random sample of n observations is NOT taken from a normal population with mean 𝜇 and variance 𝜎 2 , 𝑋ത
follows a normal distribution with mean 𝜇 and variance 𝜎 2 /𝑛 approximately if the sample size is big (e.g., n ≥ 30)

𝑋ത
𝑋ത for n = 1,
Figure 8.1 Illustration of the Central Limit Theorem (distribution of X
moderate n, and large n)

Lecture 4-2 6
Central Limit Theorem
Central Limit Theorem

Theorem 8.2
Central Limit Theorem: If 𝑋ത is the mean of a random sample of size n taken from a population with mean 𝜇 and finite
variance 𝜎 2 , then the limiting form of the distribution of
𝑋ത − 𝜇
𝑍= ,
𝜎/ 𝑛
as 𝑛 → ∞, is the standard normal distribution n(𝓏; 0, 1).

Lecture 4-2 7
Central Limit Theorem
Central Limit Theorem

Application: hypothesis testing on population mean (𝝈𝟐 is known)

Suppose that a manufacturing process produces automobile cylindrical component
parts. The engineer involved conjecture the mean diameter of the part population is
5.0 millimeters. The engineer samples 100 parts and 𝑥ҧ = 5.027.
Does this sample information appear to support or refute the engineer’s conjecture?
It is known that 𝜎 = 0.1

Since sample size 100 is big, we know that 𝑋ത follows 𝑛(𝜇, 0.12Τ100)
approximately using the central limit theorem. If the population mean 𝜇 is 5.0, then
𝑋ത follows 𝑛(5, 0.12 Τ100)

Figure 8.3 Area for Case Study 8.1

Lecture 4-2 8
Central Limit Theorem
Central Limit Theorem

Application: hypothesis testing on population mean (𝝈𝟐 is known)

Now, what is the chance that ഥ

X will deviate by as much as 0.027 millimeters?

𝑋ത − 5
𝑃 𝑋ത − 5 ≥ 0.027 = 𝑃 𝑋ത − 5 ≥ 0.027 + 𝑃 𝑋ത − 5 ≤ −0.027 = 2𝑃 ≥ 2.7 = 0.007
0.1Τ 100

Since the chance is 0.007 (very small value), we do not certainly support the conjecture that 𝜇 = 5.0.

Lecture 4-2 9
Sampling Distributions of
Means

Lecture 4-2 10
 8.3-8.4 Sampling Distributions of Means
Sampling Distribution of the Difference between Two Means

Suppose we have two normal populations, the first with mean μ1 and variance 𝜎12 and the second with mean 𝜇2 and variance
𝜎22 . Let 𝑋ത1 represent the sample mean with sample size 𝑛1 from the first population, and 𝑋ത2 represent the sample mean with
sample size 𝑛2 from the second population.

We are interested in the sampling distribution of the difference 𝑋ത1 − 𝑋ത2.

Lecture 4-2 11
 8.3-8.4 Sampling Distributions of Means
Sampling Distribution of the Difference between Two Means

Theorem 8.3

Lecture 4-2 12
 8.3-8.4 Sampling Distributions of Means
Sampling Distribution of the Difference between Two Means

Application: hypothesis testing on population mean (𝜎12 and 𝜎22 are known)
Two independent experiments are run in which two different types of paints are compared. 18 specimens are painted using
type A and the drying time is recorded for each. The same is done with type B. The population variances of both types are
known to be 1.0.

From the experiments, it is observed that 𝑋ത𝐴 − 𝑋ത𝐵 = 1.0.

The hypothesis we want to test is 𝜇𝐴 − 𝜇𝐵 = 0 (mean drying time is equal for the two types of paint).

If the hypothesis is true, then from the sampling distribution of 𝑋ത𝐴 − 𝑋ത𝐵 , we know that
2 2
𝜎𝐴 𝜎𝐵 1
𝜇𝑋ത 𝐴−𝑋ത 𝐵 = 𝜇𝐴 − 𝜇𝐵 = 0 and 𝜎𝑋2ത 𝐴−𝑋ത𝐵 = 𝑛𝐴
+ 𝑛𝐵
= 9

Lecture 4-2 13
 8.3-8.4 Sampling Distributions of Means
Sampling Distribution of the Difference between Two Means

Application: hypothesis testing on population mean (𝜎12 and 𝜎22 are known)

From theorem 8.3,

(𝑋ത𝐴 −𝑋ത𝐵 ) − ( 𝜇𝐴 −𝜇𝐵 ) 1−0

𝓏 = = = 3.0 and 𝑃 𝑍 > 3.0 = 0.0013
1Τ9 1Τ9

Lecture 4-2 14
 6.7 Chi-Square Distribution
Probability density function
1 𝑣 Τ2−1 −𝑥ൗ2
𝑥 𝑒 , 𝑥>0
𝑓 𝑥; 𝑣 = ቐ2𝑣Τ2 𝛤(𝑣 Τ2)
0 otherwise

The chi-square distribution plays a vital role in statistical inference (hypothesis test)
Theorem 6.5
The mean and variance of the chi-squared distribution are 𝜇 = 𝑣 and 𝜎 2 = 2𝑣.

Probability density function Cumulative distribution function

Lecture 4-2 15
7.3 Linear Combinations of Random Variables
Theorem 7.12

Corollary 7.1

Corollary 7.2

Lecture 4-2 16
Sampling Distribution of
Variance

Lecture 4-2 17
8.5 Sampling Distribution of Variance
Suppose that a random sample of 𝑛 observations is taken from a normal population with mean 𝜇 and variance 𝜎 2 .

Note that

𝑛 𝑛

෍ 𝑋𝑖 − 𝜇 2
= ෍ 𝑋𝑖 − 𝑋ത + 𝑋ത − 𝜇 2

𝑖=1 𝑖=1
𝑛 𝑛 𝑛

= ෍ 𝑋𝑖 − 𝑋ത 2
+ ෍ 𝑋ത − 𝜇 2
+ 2 𝑋ത − 𝜇 ෍ 𝑋𝑖 − 𝑋ത
𝑖=1 𝑖=1 𝑖=1
𝑛

= ෍ 𝑋𝑖 − 𝑋ത 2
+ 𝑛 𝑋ത − 𝜇 2

𝑖=1

𝑛
σ𝑛𝑖=1 𝑋𝑖 − 𝜇 2
𝑋𝑖 − 𝑋ത 2
𝑛 𝑋ത − 𝜇 2
(𝑛 − 1)𝑆 2 𝑋ത − 𝜇 2
= ෍ + = +
𝜎2 𝜎2 𝜎2 𝜎2 𝜎 2 Τ𝑛
𝑖=1

Lecture 4-2 18
8.5 Sampling Distribution of Variance
σ𝑛
𝑖=1 𝑋𝑖 −𝜇
2 ത
𝑋−𝜇 2
From Corollary 7.1, we know that 𝜎2
is a chi-squared random variable with 𝑛 degrees of freedom. 𝜎 2 Τ𝑛
is a chi-
(𝑛−1)𝑆 2
squared random variable with 1 degree of freedom. Therefore, from Theorem 7.12, is a chi-squared random variable
𝜎2
with 𝑛-1 degrees of freedom

Theorem 8.4

Lecture 4-2 19
8.5 Sampling Distribution of Variance

Figure 8.7 The chi-squared distribution

𝜒𝛼2 represent the 𝜒 2 value above which we find an area (probability) of 𝛼

Table A.5 gives the value of 𝜒𝛼2 for various values of 𝛼 and 𝜈.

2 2
Exactly 95% of a chi-squared distribution lies between 𝜒0.975 and 𝜒0.025

Lecture 4-2 20
8.5 Sampling Distribution of Variance

Application: hypothesis testing on population variance

A manufacturer of car batteries guarantees that the batteries will last, on average, 3 years with a variance of 1 year. If
lifetimes of five of these batteries is measured and 𝑠 2 = 0.815.

Should the manufacturer still be convinced that the batteries have a variance of 1 year? Assume that the batteries
lifetime follows a normal distribution.

If the variance of 1 year is true, then 𝜒 2 = 3.26 is a value from a chi-squared distribution with 4 degrees of freedom.
Since 95% of the 𝜒 2 values with 4 degrees of freedom fall between 0.484 and 11.143, the conjecture with 𝜎 2 = 1 is
reasonable.

Lecture 4-2 21
8.6 (Student’s) t Distribution
Suppose that a random sample of 𝑛 observations is taken from a normal population in which mean 𝜇 is known and variance
𝝈𝟐 is NOT known.

ത
Sampling Distribution of Mean (𝑋)

1 𝜎2
Note that 𝜇𝑋ത = 𝜇, 𝜎𝑋2ത = 𝜎2 + 𝜎2 + ⋯+ 𝜎2 = .
𝑛2 𝑛

𝑆2
Since 𝜎 2 is unknown, we can estimate it using 𝑆 2 with which the estimate of 𝜎𝑋2ത can be set to 𝑛
. Then the following statistic

𝑋ത − 𝜇
𝑇 = 𝑋ത − 𝜇𝑋ത Τ𝜎ෞ𝑋ത =
𝑆Τ 𝑛
has a t distribution.

Lecture 4-2 22
 8.6 (Student’s) t Distribution
Theorem 8.5

Corollary 8.1

Lecture 4-2 23
8.6 (Student’s) t Distribution

Figure 8.8 The t-distribution curves for v = 2, 5, and ∞ Figure 8.9 Symmetry property (about 0) of the t-distribution

𝑡𝛼 represents the t-value above which we find an area equal to 𝛼

Exactly 95% of the values of a t-distribution with 𝜈 = 𝑛 − 1 degrees of freedom lie between
− 𝑡0.025 and 𝑡0.025.

Lecture 4-2 24
8.6 (Student’s) t Distribution
Application: hypothesis testing on population mean (𝝈𝟐 is unknown)
A chemical engineer claims that the population mean yield of a certain batch process is 500 grams. To check this claim, he
samples 25 batches each month and 𝑥ҧ = 518 grams with 𝑠 = 40 grams. Assume the distribution of yields to be approximately
normal. If the computed t-value falls between −𝑡0.05 and 𝑡0.05 (90% area), he is satisfied with this claim.

𝑡0.05 is 1.711 with 𝜈 = 24 and so the interval is [-1.711, 1.711]. The computed t is 2.25. Thus the t-value is not in the interval.

If sample size n is large, then the value of s is reliable (does not fluctuate from sample to sample). In practice, if n≥30, the
statistic T follows the standard normal distribution. However, if n<30, it is better to use the t distribution.

Lecture 4-2 25
 8.7 F Distribution
Suppose we have two normal populations in which the variances 𝝈𝟐𝟏 and 𝝈𝟐𝟐 are NOT known.

We want know compare the variances using samples (𝑛1 and 𝑛2 ) chosen from the two populations (sample variances).

The statistic F is defined to be the ratio of two independent chi-squared random variables U and V, each divided by its number
of degrees of freedom (𝜐1 and 𝜐2 ).

𝑈Τ𝜐1
𝐹=
𝑉Τ𝜐2

Lecture 4-2 26
8.7 F Distribution
Theorem 8.6

𝑓𝛼 represents the f-value above which we find an area equal to 𝛼

Figure 8.11 Typical F-distributions

Lecture 4-2 27
8.7 F Distribution

Table A.6 gives the values of 𝑓𝛼 only for 𝛼 = 0.05 and

𝛼 = 0.01 for various combinations of 𝜐1 and 𝜐2 .

Figure 8.12 Illustration of the 𝑓𝛼 for the F-distribution

Theorem 8.7

𝑈Τ𝜐1 (𝑛1 −1)𝑆12 (𝑛2 −1)𝑆22

In the definition of 𝐹 = , let 𝑈 = and 𝑉 = with 𝜐1 = 𝑛1 − 1 and 𝜐2 = 𝑛2 − 1. Then the following
𝑉 Τ𝜐2 𝜎12 𝜎22
theorem holds.

Lecture 4-2 28
8.7 F Distribution

Theorem 8.8

F-distribution is called variance ratio distribution.

Lecture 4-2 29
8.7 F Distribution
Application 1: hypothesis testing on two population variances
Two independent experiments are run in which two different types of paints are compared. 18 specimens are painted using
type A and the drying time is recorded for each. The same is done with type B (𝑛𝐴 = 𝑛𝐵 = 18). It is known that mean drying
time is different for the two types of paint.

We want to test the claim that population variances of both types are equal.
If the claim is true, then

𝑆𝐴2
𝐹= 2
𝑆𝐵

follows a F distribution with 17 (𝑛𝐴 − 1) and 17 (𝑛𝐵 − 1) degrees of freedom.

Lecture 4-2 30
Practice

Lecture 4-2 31
Q1.
Suppose that 𝑥 is a random variable with probability distribution of

𝑘 + 2𝑥 −2≤𝑥 ≤0
𝑓 𝑥 =ቊ
𝑘 − 2𝑥 0≤𝑥≤2

What is the mean of 𝑥?

(a) 0 (b) -1

(c) 1 (d) We cannot get the mean of 𝑥 because 𝑘 is unknown.

Lecture 4-2 32
Q2.
A mechatronic assembly is subjected to a final functional test. Suppose that defects occur at random in these
assemblies, and that defects occur according to a Poisson distribution with variance equal to 0.05. What is the
probability that an assembly will have one or more defect?

(a) 0.0196 (b) 0.9802

(c) 0.0488

(d) We cannot compute this probability because the Poisson distribution parameter 𝜆 is not given.

Hint: 𝑒 −0.05 = 0.9512

Lecture 4-2 33
Q3.
The inside diameters of bearings used in an aircraft landing gear assembly are known to have a standard deviation
of 𝜎 = 0.002 𝑐𝑚. A random sample of 15 bearings has an average inside diameter of 8.2535 cm. What of the
following answers is closest to the p-value for the hypothesis testing that the mean inside bearing diameter is 8.25
cm versus mean inside bearing diameter is not equal to 8.25 cm.

(a) Approximately zero (b) 0.05

(c) 0.1 (d) 1

Lecture 4-2 34
Q4.
In a statistical quality control setting, an x-bar control chart is used to monitor the process mean over time. Suppose
that the control limits for the x-bar chart are established using 3-sigma limits, assuming the underlying process
distribution is normal and in-control. Based on this setup, what is the Type I error rate (α), i.e., the probability that
a point will fall outside of the control limits even though the process is actually in control?
Choose the closest value to the true Type I error rate under these assumptions

(a) 0.0455 (b) 0.0327

(c) 0.02275 (d) 0.0654

Lecture 4-2 35
Q5.
The mean contents of coffee cans filled on a production line are
being studied. Standards specify that the mean contents must be
16.0 oz, and from experience it is known that the standard
deviation of the can contents is 0.4 oz. The hypotheses are

𝐻0 : 𝜇 = 16.0
𝐻1 : 𝜇 ≠ 16.0

A random sample of 16 cans is to be used, and the type I error

probability is specified as 𝛼 = 0.05. If the true mean contents
are 𝜇1 = 16.2 oz, what of the following answers is closest to the
type II error?

(a) 0.48 (b) 0.05

(c) 0.16 (d) 0.95

Lecture 4-2 36
Q & A

Lecture 4-2