Design and Analysis

of Experiments

Chapter 4

BLOCKING

Dr. Tran Thanh Hung

Department of Automation Technology,
College of Engineering, Can Tho University
Email: [email protected]
 Chapter objectives

At the end of this chapter, students can:

• explain the concept of Blocking in Design of
Experiment
• use ANOVA to analyze block designs
• apply of Latin Square Designs in presence of
two nuisance factors
 Nuisance factors

• What are nuisance factors?

• Effects of nuisance factors?

• How to deal with nuisance factors?

Example: hardness testing

• One factor: tip type

(Tip 1, Tip 2, Tip 3, Tip 4)
• Four tests per tip type
tip  4 x 4 runs

• These 4 tips produce

different hardness
reading on a hardness
tester?
 Example: hardness testing

• Assignment of the tips to an experimental

unit; that is, a test coupon (mẫu thử)
 4 x 4 runs = 16 coupons
• Structure of a completely randomized design
(CRD)
• Problems?
 The test coupons are a source of nuisance
variability

• The experimenter may want to test the tips

across coupons of various hardness levels 5
The need for blocking
 Example: hardness testing
• Conduct this experiment as a randomized complete
block design (RCBD= thiết kế đóng khối hoàn chỉnh
ngẫu nhiên), assign all 4 tips to each coupon
• Each coupon is called a “block”; that is, it’s a
homogenous (đồng nhất) experimental unit on which
to compare the tips
• Variability between blocks can be large, but
variability within a block should be relatively small
• 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 6
 Example: hardness testing
• Suppose that we use b = 4 blocks:

• 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)
7
 Randomized complete block
design (RCBD)
• The RCBD is one of the most widely used experimental
designs.
• Situations for which the RCBD is appropriate:
Units of test equipment or machinery, batches of raw
material, people, time,…

Suppose that there are a

treatments (factor levels) and b
blocks
 Extension of the ANOVA to the
RCBD
• A statistical model (effects model) for the RCBD is
 i  1, 2,..., a
yij     i   j   ij 
 j  1, 2,..., b
  an overall mean,  ij  N  0,  2  : random error
 i  i th treatment effect, a b

 i  0,  j 0
 j  jth block effect,
i 1 j 1

• The relevant (fixed effects) hypotheses are

H 0 : 1   2     a where i  (1/ b) j 1 (    i   j )    i

b

 H 0 : 1   2     a  0 9
 Extension of the ANOVA to the
RCBD
Let:
yi. , yi. :total & average of the observations under treatment i,
y. j , y. j :total & average of the observations in block j,
y.. , y.. :total & average of all observations,
b
yi.
yi.   yij , yi .  , i  1, 2,..., a
j 1 b
a y. j
y. j   yij , y. j  , j  1, 2,..., b
i 1 a
a b
y..
y..   yij , y..  ,
i 1 j 1 N
10
N  a  b :total number of observations
 Extension of the ANOVA to the
RCBD
ANOVA partitioning of total variability:
a b a b

 ij ..  i. .. . j .. ij i. . j ..
( y  y
i 1 j 1
)  [( y2
 y )  ( y 
i 1 j 1
y )  ( y  y  y  y )] 2

a b a b
 b ( yi.  y.. )  a  ( y. j  y.. )   ( yij  yi.  y. j  y.. ) 2
2 2

i 1 j 1 i 1 j 1
a b a b
2  ( yi.  y.. )( y. j  y.. )  2  ( yi.  y.. )( yij  yi.  y. j  y.. )
i 1  j 1       i 1  j 1          
0 0
a b
2  ( y. j  y.. )( yij  yi.  y. j  y.. )
i 1  j 1          
0
11
 SST  SSTreatments  SS Blocks  SS E
 Extension of the ANOVA to the
RCBD
The degrees of freedom for the sums of squares in
SST  SSTreatments  SS Blocks  SS E
are as follows:

ab  1  a  1  b  1  (a  1)(b  1)
SSTreatments SS Blocks
MSTreatments  , MS Blocks  ,
a 1 b 1
SS E MSTreatments
MS E  , F0 
(a  1)(b  1) MS E
F0 can be used to test the hypothesis of equal treatment
means 12
 ANOVA Display for the RCBD

• The reference distribution for F0 is the Fa-1, (a-1)(b-1)

distribution (randomization distribution)
• Reject the null hypothesis H0 if
F0  F ,a 1,( a 1)(b 1) 13
 Example: Vascular Graft (p.144)
• To conduct this experiment as a RCBD, assign all
4 pressures to each of the 6 batches of resin
• Each batch of resin is called a “block”; that is, it’s a
more homogenous experimental unit on which to
test the extrusion pressures

14
 Example: Vascular Graft
ANOVA Output

F0  F0.05,3,15  3.29  Reject H 0

15
Model Adequacy Checking:
Residual Analysis

16
Model Adequacy Checking:
Residual Analysis

17
 Model Adequacy Checking:
Residual Analysis
• Basic residual plots indicate that normality,
constant variance assumptions are satisfied
• No obvious problems with randomization
• No patterns in the residuals vs. block
• Can also plot residuals versus the pressure
(residuals by factor)
• These plots provide more information about the
constant variance assumption, possible outliers

18
Multiple Comparisons for the Vascular
Graft Example – Which Pressure is
Different?

19
 Other Aspects of the RCBD
See Text, Section 4-1.3, pg. 150
• The RCBD utilizes an additive model (mô
hình cộng) – no interaction between
treatments and blocks
• Random Treatments and/or Blocks
• Missing values: Missing some observations in
one of the blocks
• What are the consequences of not blocking
if we should have?
• Sample sizing in the RCBD? The OC curve
approach can be used to determine the
number of blocks to run..see page 153 20
 The Latin Square Design

• Text reference, Section 4-2, pg. 158

• These designs are used to simultaneously
control (or eliminate) two sources of
nuisance variability
• A significant assumption is that the three
factors (treatments, 2 nuisance factors) do
not interact
• If this assumption is violated, the Latin square
design will not produce valid results
• Latin squares are not used as much as the
RCBD in industrial experimentation
21
 The Rocket Propellant Problem –
A Latin Square Design

•• This is a
 
• Page 162 shows some other Latin squares
• Table 4-13 (page 162) contains properties of Latin
squares
• Statistical analysis?
22
 Statistical Analysis of the
Latin Square Design
• The statistical (effects) model is
 i  1, 2,..., p

yijk     i   j   k   ijk  j  1, 2,..., p
k  1, 2,..., p


• The statistical analysis (ANOVA) is much like

the analysis for the RCBD.
• See the ANOVA table, page 160 (Table 4-10)
• The analysis for the rocket propellant
example is presented on text page 161
23
 Thực hành chương 4

• Bài 1: Dùng Minitab phân tích kết quả thí

nghiệm trong ví dụ kiểm tra độ cứng.
Số liệu kết quả thí nghiệm:

Tip Coupon Hardness Tip Coupon Hardness

1 1 9.3 3 1 9.2
1 2 9.4 3 2 9.4
3 3 9.5
1 3 9.6 3 4 9.7
1 4 10.0 4 1 9.7
2 1 9.4 4 2 9.6
2 2 9.3 4 3 10.0
2 3 9.8 4 4 10.2
2 4 9.9
 Thực hành chương 4

• Bài 2: Thiết kế thí nghiệm xếp máy bay

với 3 độ rộng cánh khác nhau trên 3 loại
giấy khác nhau. Mỗi loại phóng 10 lần.
Dùng Minitab phân tích kết quả.