CHAPTER FOUR

 Factorial Designs
What is Factorial Design Mean?

Factorial Experiments

Two-Factor Factorial Design

The 2k Factorial Design

The 22 Factorial Design

1
 Factorial Design
 A design which uses factorial experiment strategy.
 Two or more factors are investigated simultaneously in
the same study.
 Involves testing two or more different factors on the
same group at the same time.
 The investigator is concerned with testing several levels
of one factor while keeping all other factors at a
constant level.

2
 Cont..
 Many experiments involve the study of the effects of
two or more factors. Factorial designs are most efficient
for this type of experiment.
 In a factorial design, all possible combinations of the
levels of the factors are investigated in each replication.
 If there are a levels of factor A, and b levels of factor B,
then each replicate contains all ab treatment
combinations.

3
 Advantages of Factorial Designs
 They are more efficient than one-factor-at-a-time (OFAT) experiments.
 Tests the equality of two or more (p) population means when several
independent variables are used.

 Saves Time & Effort: Economizing the experimental resources

e.g., Could use separate Completely Randomized Designs (CRD) for
each variable.

 A factorial design is necessary when Interaction may be present to

avoid misleading conclusions.

 Controls Confounding effects by putting other variables into Model.

4
 Factorial Experiment
 It is an experiment in which the response variable is measured at
all possible combinations of the levels of the factors (independent
variables).
 The goal of factorial experiment is to determine which factors
have the largest effects on the response, and whether there are
interactions between factors.
 In the one-way ANOVA; we look for the effect of one “factor”.
 In the two-way ANOVA; we look for the effects of “two-factors”.
 A further complication arises because of the possibility of
“interactions” between the factors.
5
 Main effect and Interaction effects
 Main effect is the effect of independent variable
(factor) on a dependent variable (response).
 Interaction effect may arise when considering the
relationship among three (one for response variable
and the other two for factors) or more variables.
 It describes a situation in which the simultaneous
influence of two variables on a third which is not
additive.

6
Example: Factorial Experiments

7
 Factorial Experiments
No Interaction Interaction

8
 Two-Factor Factorial Design
 This design only considers two-factors and it may consider
the interaction of two-factors.
 Two versions of two-factor factorial design models:
– Two factor design with interaction and
– Two factor design without interaction.
Assume the number of observations in each cell are equal.
Let yi
jkbe the obse r
v edre spons ew h
en
• Factor A is at ith level (i = 1, 2, …, a)
• Factor B is at jth level (j = 1, 2, …, b)
 A two-factor factorial design has abn observations which
are taken randomly so that it can also be called as
Completely Randomized two-factor factorial design.
9
The data layout of a two-factor factorial design

Level i
Factor A
Y i
j
k
Observation k
Level j
Factor B 10
a levels of factor A; b levels of factor B;
 Statistical Analysis of the Fixed-Effects Model
 We may describe the linear statistical model by:

11
11
 Two-factor ANOVA
 Consists of three significance tests:

• Each of the main effects (A and B).

• Interaction of the two factors (AB).

 There is an F-test for each of the hypotheses: the mean

square for each main effect and the interaction effect

divided by the within-variance (MSE).

12
Two-Way ANOVA: Hypothesis Testing
1. The first hypothesis (main effect of factor A ): Row effect
looks at the mean response for each level of A
• No Difference in Means Due to Factor A
H0: 1.= 2
.=.
..= a
.
2. The second (main effect of factor B) : Column effect
looks at the mean response for each level of B
• No Difference in Means Due to Factor B
H0: .1= .
2=...= .
b

3. The third hypothesis (interaction effect):

asks whether or not factor A has an effect on factor B.
• No Interaction of Factors A & B H0
:
A
B=
i
j0
13
Hypothesis Test for a Two -Factorial Designs

14
Two-Way ANOVA:Total Variation Partitioning

Total Variation
SS(Total)

Variation Due to Variation Due to

Treatment A Treatment B

SSA SSB

Variation Due to Variation Due to

Interaction Random Sampling
SS(AB) SSE
15
 Two-Way ANOVA
Total Variation Partitioning

16
ANOVA partitioning of total variability:

Corresponding degree of freedom partitioning is:

17
Two-Factor Factorial Experiments with Interaction
Two-Way ANOVA Table: Summary Table

Same as Other Designs

18
Two-Factor Factorial Experiments

19
20
Two-Way ANOVA: Assumptions
1. Normality
• The populations from which the samples were
obtained must be normally or approximately
normally distributed.
2. Homogeneity of Variance
• The variances of the populations must be equal.

3. Independence of Errors
• The samples must be independent.

21
22
23
SOLUTION

24
CONT.………

25
26
27
28
Example: Two-Factor Factorial Experiments

29
Example: Two-Factor Factorial Experiments

30
Solution:

31
Cont’d

32
ANOVA TABLE

33
Cont’d

34
 Cont’d

Figure 14-8 Graph of average adhesion force versus

primer types for both application methods.
35
Factorials with More Than Two Factors

 Basic procedure is similar to the two-factor case;

allabc…kn treatment combinations are run in
random order

 ANOVA identity is also similar:

SS T  SS A  SS B    SS A B  SS A C  
 SS A B C    SS A B K  SS E
 Complete three-factor example in text,

36
General Factorial Experiments
 Model for a three-factor factorial experiment linear
statistical model:

37
Three-Factor Factorial Experiments
Three-Way ANOVA: Summary Table

38
 The 2k Factorial Design

 Special case of the general factorial design; in which

there arek factors, each at two levels.
 The two levels are usually called low and high (they
could be either quantitative or qualitative)
 Very widely used in industrial experimentation.
 Form a basic “building block” for other very useful
experimental designs (DNA).
 Special (short-cut) methods for analysis.
39
 The 22 Design; The simplest design
o The 22 Design is one with only two factors, say A and B, each
run at two levels. The levels of the factors may be arbitrary
called “low” and “high”. Let low be denoted by - and high
denoted by +.

o The layout of the data will look like

40
 22 Factorial Designs;
 “-” and “+” denote the low and high
levels of a factor, respectively.
 Low and high are arbitrary terms.

 Geometrically, the four runs form

the corners of a square.
 Factors can be quantitative or
qualitative, although their
treatment in the final model will be
different.

Figure 4-3: The 22 factorial design.

41
2k Factorial Designs: 22 Design
The main effect of a factor A is estimated by;

The main effect of a factor B is estimated by;

The AB interaction effect is estimated by;

42
 Estimation of Factor Effects
For manual calculations: The effect estimates are:
A  y A  y A
A = 8.33, B = -5.00, AB = 1.67
ab a b  (1)
  Practical interpretation?
2n 2n Thesign and themagnitude of the factors determine
 1
2n
[ ab  a  b  (1)] which variables are likely to be important.
B  yB  yB  For instance, if the effect of A is positive, then this
suggests that increasing A from low level to high level
ab b a  (1)
  will increase the value of the response variable.
2n 2n
 If the effect of A is negative, then this suggests that
 1
2n
[ ab  b  a  (1)]
increasing A from low level to high level will decrease
ab  (1) a b
AB   the value of the response variable.
2n 2n
 1
2n
[ ab  (1)  a  b ]
43
44
45
 Chemical Process Example
 Consider an investigation into the effect of concentration of the
reactant and the amount of the catalyst on the conversion (yield) in a
chemical process. Let the reactant concentration be factor A, and let
the two levels of interest be 15 and 25 percent. The catalyst is factor B,
with the high level denoting the use of only 1 pound. The data is given
below:
Temperature Level

Reactant Concentration 1 Pound 2 Pounds

15 percent 28,25,27 18,19,23

25 percent 36,32,32 31,30,29

46
In the 22 design the low and high levels of A and B are denoted by – and + respectively.
Factor A: 15% (-) and 25% (+) Factor B: 1 pound (-) and (+)

A = reactant concentration,B = catalyst amount,y = recovery

47
The above data can be displayed as follows:

Factor Replicate

Treatment
A B combination I II III Total

- - (1) 28 25 27 80

+ - a 36 32 32 100

- + b 18 19 23 60

+ + ab 31 30 29 90
48
Figure 4.1: Amount of catalyst and reactant
concentration factor experiment.

1
A  ( 90  100  60  80 )  8 . 33
2 ( 3)
1
B  ( 90  60  100  80 )  5 . 00
2 ( 3)
1
C  ( 90  80  100  60 )  1 . 67
2 ( 3)
49
Consider the sum of Squares for A, B and AB.

Contrast A
 ab  a  b  (1)
Contrast B
 ab  b  a  (1)

Contrast AB
 ab  (1)  a  b
2
 Contrast 
SS 
4n

 a  b  (1)   b  a  (1)  1  a  b
2 2 2
ab ab ab
SS A  SS B  and SS AB 
4n 4n 4n

50
 Consider Example (*):
 a  b  (1)   b  a  (1)    30 
2 2 2 2
ab 50  ab
SS A    208 . 33 SS B    75
4n 4 ( 3) 4n 4 ( 3)
1  a  b
2 2
ab (10 )
and SS AB    8 . 33
4n 4 ( 3)

a b n 2 2
y ... ( 330 )
  y ijk   28 2  25 2  ...  29 2   323
2
SS T
i 1 j 1 k 1 4n 4 ( 3)
SS E  SS T  SS A  SS B  SS AB  31 . 34

51
 Table 4.6: Analysis of variance for the Experiment in figure 4.1
Source of Sum of Degrees of Mean Square F0 P-Value
Variation Square Freedom
A 208.33 1 208.33 53.15 0.0001

B 75.00 1 75.00 19.13 0.0024

AB 8.33 1 8.33 2.13 0.1826

Error 31.34 8 3.92

Total 323.00 11

F 0 . 05 , 1, 8  5 . 32  Based on this and the p  values , we conclude that the main effects are statistica ly

significan t and there is no int eraction

52
Table 4.7: Algebraic signs for calculating the Effects in 22 Design

Treatment Effects

combinations I A B AB

(1) + - - +

a + + - -

b + - + -

ab + + + +
 The column headings are the main effects (A and B), the AB
interaction, and I, which represents the total or average of the
entire experiment.
 Notice that the column corresponding to I has only plus signs. The
row designators are the treatment combinations.
53
The 23 Factorial Design

54
Effects in The 23 Factorial Design
A  y A  y A
B  yB  yB
C  yC   yC 
etc, etc, ...

Analysis
done via
computer

55
An Example of a 3
2 Factorial Design

A = gap,B = Flow,C = Power,y = Etch Rate

56
Table of – and + Signs for the 23 Factorial Design

57
 Properties of the Table

 Except for columnI , every column has an equal number of + and – signs

 The sum of the product of signs in any two columns is zero

 Multiplying any column byI leaves that column unchanged (identity element)

 The product of any two columns yields a column in the table:

AB  AB

 Orthogonal design.
AB BC  2
AB C  AC

 Orthogonality is an important property shared by all factorial designs.

58
Estimation of Factor Effects

59
ANOVA Summary – Full Model

60
The General 2k Factorial Design

There will be k main effects, and

k 
  tw o - f ac to r i n ter ac ti o n s
2 
k 
  th r ee- f ac to r i n ter ac ti o n s
3 

1 k  f ac to r i n ter ac ti o n

61
The Split-Plot Designs
What is a Split-Plot Design?
• A split-plot design is a special case of a factorial
treatment structure.
• It is used when a factorial treatment structure has
two levels of experimental units.
• Basically a split-plot design consists of two
experiments with different experimental units of
different “size”. i. whole-plot
e subplot
&
• E.g., in agronomic field trials certain factors require
“large” experimental units, whereas other factors
can be easily applied to “smaller” plots of land.
62
 End of Chapter

Thank you!

63