CHAPTER 5

SINGLE FACTOR
EXPERIMENTS:
BLOCKING
Instructor: Lena Ahmadi

During
the
Lecture E6-2004
Skype: L2ahm Ext.
Office L2ahm adi@u 37160
After Hour adi w...
the
Lecture

In Person Message Call

Course notes are copyrighted material. © Professor Alex Penlidis, 2019.
This copy is for individual use only in connection with this course.
It may not be resold or used to make additional copies.
Tel: (519) 888-4567 x36634, E-mail: [email protected]
1
Major Topics
• Chapter 1:
Statistical Background
• Chapter 2:
Regression Analysis
• Chapter 3:
Statistical Design of Experiments
• Chapter 4:
Design/Analysis of Single Factor Experiments
• Chapter 5:
Blocking
• Chapter 6:
Multifactor Experiments
• Chapter 7:
Multifactor Experiments
• Chapter 8
Response Surface Methods/CCD
• Chapter 9:
Response Surface Methods/BBD,
Face Centered Designs, Nonlinear
Regression
• Chapter 10:
Data Transformations
• Chapter 11:
The Analysis of Undesigned Data,
• Chapter 12:
Concluding Remarks
2
 LEARNING OBJECTIVES

• Understand blocking and

nuisance factors
• Learn about the randomized
block design or the RBD
• Understand the extension of
the ANOVA to the RBD

3
 THE BLOCKING PRINCIPLE

• Blocking is a technique for dealing

with nuisance factors
• A nuisance factor is a factor that
probably has some effect on the
response, but it’s of no interest to the
experimenter…however, the variability
it transmits to the response needs to
be minimized
• Typical nuisance factors include
batches of raw material, operators,
pieces of test equipment, time (shifts,
days, etc.), different experimental units
• Many industrial experiments involve
blocking (or should)
• Failure to block is a common flaw in
designing an experiment
(consequences?)

4
 EXAMPLE
BLOCKING:
PAIRED COMPARISONS
An experiment is carried out to determine the
performance (wear) of two different synthetic
materials, A and B for producing shoe soles.
Material B is considerably cheaper than A
and hence, if equally durable, would be the
preferred material in this application. At the
95% confidence level determine whether or
not there is any difference in wear between
A and B?
Material A Material B
13.2 14.0
8.2 8.8
10.9 11.2
14.3 14.2
10.7 11.8
6.6 6.4
9.5 9.8
10.8 11.3
8.8 9.3
13.3 13.6
5
 BLOCKING: PAIRED
COMPARISONS

• In this experiment, one factor (material type) is

being studied at two levels, A and B:

yA = 10.63 yB = 11.04
s2 = 6.009 s2 = 6.3426
A B

• We will assume that the variances of the two

samples are the same and pool:

(n − 1)s 2 + (n − 1)s2
s2p = A A B B
(n A + nB − 2)
9s2A + 9s2B
= = 6.18
18
• What is 18 above?

6
 BLOCKING: PAIRED
COMPARISONS

• We now have all the information we need to

compare the two averages using a hypothesis
test:

H 0 :  A − B = 0 H 1 :  A − B  0

Tobserved =
(yA − yB )− 0
 1 1 
s 2p  + 
 n A nB 

=
(10.63 −11.04) − 0 = −0.37
6.18 + 
1 1
10 10 
t18,0.025 = 2.101
 Tobserved  tcritical

• Therefore, fail to reject the null hypothesis:

there is no difference between the two
materials and we should use B (cheaper).
7
 BLOCKING: PAIRED
COMPARISONS

In reality, each person was given a

“special” pair of shoes (one shoe soled
with A and the other with B). The decision
about whether to use A for the left or right
shoe was made by flipping a coin.

Person Material A Material B D = B-A

1 13.2 14.0 0.8
2 8.2 8.8 0.6
3 10.9 11.2 0.3
4 14.3 14.2 -0.1
5 10.7 11.8 1.1
6 6.6 6.4 -0.2
7 9.5 9.8 0.3
8 10.8 11.3 0.5
9 8.8 9.3 0.5
10 13.3 13.6 0.3

8
 BLOCKING: PAIRED
COMPARISONS

• With this design (pairing), it is possible to

compare the materials by doing a hypothesis
test on the difference D = B - A:

H 0 : D = 0 H1 : D  0

D = 0.41 2 = 0.1498
sD nd = 10
0.41− 0
Tobserved = = 3.35
0.1498
10
t9,0.025 = 2.262
Tobserved  tcritical reject H0

• The effect of pairing is to eliminate the variability

present among persons. This is important here
since no two persons wear shoes out at the
same rate. (Thus, blocking eliminates potential
bias; makes test more robust in detecting
differences!)
9
 RANDOMIZED BLOCK
DESIGNS

• Extension of completely randomized

designs (material of Ch 4, notes) to the
situation where block effects are
eliminated from treatment effects and
residuals.

• Extension of paired comparisons

(paired t-test) to more than two
treatments (level of treatments).

• In chemical engineering applications:

Blocks could be different blends of
chemicals, raw material suppliers,
batches, substrates, days of production,
operators, raw materials (with different
impurity levels), etc.
10
 One more example:
RANDOMIZED BLOCK DESIGNS
(more than two treatments)

Determine the wear of four different brands of

tires after 50,000 km of driving. The experiment
involved four different cars which are the blocks.

Tire Brand Block

totals
Car (block) A B C D
1 17 14 12 13 56
2 14 14 12 11 51
3 13 13 10 11 47
4 13 8 9 9 39
Treatment 57 49 43 44 193 =
totals grand total
The design involves k treatments in n blocks (here
k=n=4). The underlying model:

yti =  +  i +  t +  ti t = 1,2,..., k
i = 1,2,...,n
 i = block deviation for i th block

, t and  ti are as defined before.

11
 ANOVA
• Correction for the mean:
1  k n 2 1
  yti  = (193)2
= 2328.0625
kn  t=1i=1  4x4
• Treatment sum of squares:
k
1
STreatment = (Trmt Totals)2 − Correction for mean
n t =1

(
1
)
= 572 + 49 2 + 432 + 44 2 − 2328.0625 = 30.7
4
(df treatment = k − 1 = 3)
• Block sum of squares:
1 n
SBlock = (Block Totals)2 − Correction for mean
k i =1
(
1
)
1
= 562 + 512 + 47 2 + 39 2 − (193)2 = 38.7
4 16
(dfBlocks = n − 1 = 3)
• Total corrected sum of squares:
STotal =   y2ti − Correction for mean
t i

(
= 172 + 142 +......+112 + 92 − ) 1
16
(193)2 = 80.9

(dfTotal = kxn − 1 = 16 − 1 = 15)

12
 ANOVA TABLE

• As before, we can assemble the results into an

ANOVA table.
Source SS df MS Fobserved
Brands 30.7 3 10.23 7.8
Cars 38.7 3 12.9 9.9
Error 11.6 9 1.3
Total 80.9 15

F3,9 ,0.05 = 3.86

7.8  3.86
9.9  3.86
• Therefore, brands (treatments) do not show
the same amount of wear; the tread loss for
cars (blocks) is not the same. In this study, the
latter is not of particular interest. In fact, it was
expected!
• Using blocks made the experiment more
robust in the face of disturbances and
sensitive to detecting differences among
treatments. 17
 Concept Map

• Blocking is a technique for

dealing with nuisance factors
• Same rules as chapter 4
– Two treatments*→ t- test
– More than two treatments→ ANOVA
(f-test)
• ANOVA table is the same as
before, we just add block as
another row (source of
variability)

* # of treatments, and levels of treatment have

the same meaning in different stats books 18
More explanation!
 THE BLOCKING PRINCIPLE

• If the nuisance variable is known and

controllable, we use blocking
• If the nuisance factor is known and
uncontrollable, sometimes we can
use the analysis of covariance to
remove the effect of the nuisance
factor from the analysis
• If the nuisance factor is unknown and
uncontrollable (a “lurking” variable),
we hope that randomization balances
out its impact across the experiment

16
 IN GENERAL…

• A block is a specific level of the

nuisance factor
• A complete replicate of the basic
experiment is conducted in each
block
• A block represents a restriction
on randomization
• All runs within a block are
randomized

17
 ANOTHER EXAMPLE:
The Hardness
Testing Example
• A number of metal specimens could
be randomly selected. Suppose we
wish to determine whether or not
four tips produce different readings
on a hardness testing machine.
Experimenter has decided to obtain
four observations on Rockwell C-
scale hardness for each tip.
• Problem: If the metal coupons differ
slightly in their hardness, the
experimental units (the coupons) will
contribute to the variability observed
in the hardness data.

18
 The Hardness Testing
Example

• Suppose that we use b = 4 blocks:

• Notice the two-way structure of the

experiment
• Once again, we are interested in
testing the equality of treatment
means, but now we have to remove
the variability associated with the
nuisance factor (the blocks)

19