2/2/2026

Basic Definitions and Principles

Subir Chowdhury School of Quality and Reliability

Many experiments involve the study of the effects of two or more factors.
Subir Chowdhury School of Quality and

DOE In general, factorial designs are most efficient for this type of
experiment. By a factorial design, we mean that in each complete trial or
Factorial Designs replicate of the experiment all possible combinations of the levels of the
factors are investigated.
Reliability

For example, if there are a levels of factor A and b levels of factor B, each
Prof. V N A Naikan replicate contains all ab treatment combinations.
IIT Kharagpur
The effect of a factor is defined to be the change in response produced by a
change in the level of the factor.

2
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

1
 2/2/2026

Basic Definitions and Principles Basic Definitions and Principles

Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

This is frequently called a main effect because it refers to the primary The main effect of factor A in this two-level design can be thought of as the
factors of interest in the experiment. difference between the average response at the low level of A and the average
response at the high level of A. Numerically, this is
For example, consider the simple experiment in Figure 1. This is a two-factor
factorial experiment with both design factors at two levels.

We have called these levels “low” and “high” and denoted them “-” and “+,” That is, increasing factor A from the low level to the high level causes an
respectively. average response increase of 21 units. Similarly, the main effect of B is

Figure 1. A two-factor factorial

The difference in response between the levels of one factor is not the same at
experiment, with the response (y)
all levels of the other factors. When this occurs, there is an interaction
shown at the corners
between the factors.
Interaction Effect, AB 1
Or, AB 1
3 4
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

2
 2/2/2026

Basic Definitions and Principles- 2 FF Experiment

Interaction Effect Another Method

Subir Chowdhury School of Quality and Reliability

For example, consider the two-factor factorial experiment shown in Figure 2.
• The Interaction effect of A and B can also be Figure 2. A two-factor factorial experiment
with interaction
found by the average of difference of
Reponses between the diagonal values of the Main Effect of A, 𝐴 1
Main Effect of B, 𝐵 9
rectangle.

• 𝐴𝐵 1
Because the effect of A depends on the level chosen for factor B, we see that
there is interaction between A and B.

6
Indian Institute of Technology, Kharagpur

3
 2/2/2026

Basic Definitions and Principles – No Interaction Basic Definitions and Principles - Interaction
Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

These ideas may be illustrated graphically. Figure 3 plots the response data in Similarly, Figure 4 plots the response data in Figure 2.
Figure 1 against factor A for both levels of factor B.

Figure 4. A factorial experiment

with interaction
Figure 3. A factorial experiment
without interaction

Note that the 𝐵 𝑎𝑛𝑑 𝐵 lines are approximately parallel, indicating a lack of
interaction between factors A and B.

7 8
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

4
 2/2/2026

Basic Definitions and Principles – Basic Definitions and Principles – Regression coefficients
Regression Model
Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

There is another way to illustrate the concept of interaction. Suppose that both
of our design factors are quantitative (such as temperature, pressure, time, etc.).
Then a regression model representation of the two-factor factorial experiment
could be written as:
𝑦 𝛽 𝛽𝑥 𝛽 𝑥 𝛽 𝑥 𝑥 𝜖

The parameter estimates obtained in the manner for the factorial design with all
factors at two levels (- and +) turn out to be least squares estimates
The parameter estimates in this regression model turn out to be related to the
effect estimates. For the experiment shown in Figure 1 we found the main
effects of A and B to be:
1
9
0
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

5
 2/2/2026

Response Surface and Contour Plots – No Interaction Response Surface and Contour Plots
Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

In Figure 5a we have a plot of the plane of y-values generated by the various
combinations of 𝑥 and 𝑥 . This three dimensional graph is called a response
Figure 5 presents graphical representations of this model. surface plot.

Figure 5 b shows the contour lines of constant response y in the 𝑥 , 𝑥 plane.

Notice that because the response surface is a plane, the contour plot contains
parallel straight lines.

Now suppose that the interaction contribution to this experiment was not
negligible; that is, the coefficient 𝛽 was not small. Figure 6 presents the
response surface and contour plot for the model

(We have let the interaction effect be the average of the two main effects = 8)
Figure 5.
1 1
1 2
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

6
 2/2/2026

Response Surface and Contour Plots – Interaction Main and Interaction Effects - Comparing
Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

Generally, when an interaction is large, the corresponding main effects have
little practical meaning. For the experiment in Figure 2, we would estimate
the main effect of A to be

which is very small, and we are tempted to conclude that there is no effect due
to A. However, when we examine the effects of A at different levels of factor B,
we see that this is not the case. Factor A has an effect, but it depends on the
level of factor B. That is, knowledge of the AB interaction is more useful than
• Figure 6: Response surface and contour plot for the model, knowledge of the main effect.
𝑦 35.5 10.5𝑥 5.5𝑥 8𝑥 𝑥
A significant interaction will often mask the significance of main effects.
• Notice that the significant interaction effect “twists” the plane in Figure 6a. These points are clearly indicated by the interaction plot in Figure 4. In the
• This twisting of the response surface results in curved contour lines of presence of significant interaction, the experimenter must usually examine the
constant response in the 𝑥1, 𝑥2 plane, as shown in Figure 6b. levels of one factor, say A, with levels of the other factors fixed to draw
• Thus, interaction is a form of curvature in the underlying response surface conclusions about the main effect of A.
1 model for the experiment. 1
3 4
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

7
 2/2/2026

The Advantage of Factorials The Two-Factor Factorial Design

The simplest types of factorial designs involve only two factors or sets of
Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

Factorial designs have several advantages.
treatments.
 They are more efficient than one-factor-at-a-time experiments.
There are a levels of factor A and b levels of factor B, and these are arranged
 Furthermore, a factorial design is necessary when interactions may be in a factorial design; that is, each replicate of the experiment contains all ab
present to avoid misleading conclusions. treatment combinations. In general, there are n replicates.

An Example
 Finally, factorial designs allow the effects of a factor to be estimated at
several levels of the other factors, yielding conclusions that are valid over
a range of experimental conditions. As an example of a factorial design involving two factors, an engineer is
designing a battery for use in a device that will be subjected to some extreme
variations in temperature. The only design parameter that he can select at this
point is the plate material for the battery, and he has three possible choices.
When the device is manufactured and is shipped to the field, the engineer has
no control over the temperature extremes that the device will encounter, and he
knows from experience that temperature will probably affect the effective
battery life.
1 1
5 6
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

8
 2/2/2026

The Two-Factor Factorial Design The Two-Factor Factorial Design

However, temperature can be controlled in the product development laboratory
Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

Factors: Material type ( a = 3 types), Temperature (b = 3 Levels),
for the purposes of a test. Replications: (n = 4)

Table . 1

temperature, and all 36 tests are run in random order. The experiment and the
resulting observed battery life data are given in Table 1.

1 1
7 8
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

9
 2/2/2026

Life (in hours) data for the battery The Two-Factor Factorial Design
Design Example

Subir Chowdhury School of Quality and Reliability

Temperature (oF)
Material
Type 15 70 125
130 155 34 40 20 70
1
74 180 80 75 82 58
150 188 136 122 25 70
2 In general, a two-factor factorial experiment will appear as in Table 2. The order
159 126 106 115 58 45 in which the abn observations are taken is selected at random so that this design
is a completely randomized design.
138 110 174 120 96 104
3
168 160 150 139 82 60
2
0
Indian Institute of Technology, Kharagpur

10
 2/2/2026

The Two-Factor Factorial Design The Two-Factor Factorial Design

The observations in a factorial experiment can be described by a model. There
Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

Table 2 are several ways to write the model for a factorial experiment. The effects
model is

2 2
1 2
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

11
 2/2/2026

The Two-Factor Factorial Design The Two-Factor Factorial Design

Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

Statistical Analysis of the Fixed Effects Model

2 2
3 4
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

12
 2/2/2026

The Two-Factor Factorial Design The Two-Factor Factorial Design

Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

Notice that the total sum of

equation for the two-factor factorial.

We may write the ANOVA Equation symbolically as

2 2
5 6
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

13
 2/2/2026

The Two-Factor Factorial Design The Two-Factor Factorial Design

Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

n replicates; thus there are ab(n - 1) degrees of freedom for error.

2 2
7 8
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

14
 2/2/2026

The Two-Factor Factorial Design The Two-Factor Factorial Design

Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

The test procedure is usually summarized in an analysis of variance table, as The total sum of squares is computed as usual by :
shown in Table 3.

Table 3

2 3
9 0
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

15
 2/2/2026

The Two-Factor Factorial Design The Two-Factor Factorial Design

Example 1. The Battery Design Experiment
Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

Table 4 presents the effective life (in hours) observed in the battery design
example (Table 1) .

3 3
1 2
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

16
 2/2/2026

The Two-Factor Factorial Design The Two-Factor Factorial Design

Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

The row and column totals are shown in the margins of the table, and the
circled numbers are the cell totals.
The sums of squares are computed as follows:

3 3
3 4
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

17
 2/2/2026

The Two-Factor Factorial Design The Two-Factor Factorial Design

Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

The ANOVA is shown in Table 5.

temperature are also significant. Temperature is more significant than

material type.
3 3
5 6
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

18
 2/2/2026

ANOVA Table (for SL 𝛼 0.05 ANOVA Table (for SL 𝛼 0.01

Source of
Source of Variation SS DF MS Fo
Variation SS DF MS Fo Ftable (SL Decision
Ftable (SL Decision Total 77646.97 35 =0.01)
Total 77646.97 35 =0.05) Material Reject
Material Reject Types 10683.72 2 5341.861 7.911 5.4881 Ho
Types 10683.72 2 5341.861 7.911 3.3541 Ho Temperatu Reject
Temperatu Reject re 39118.72 2 19559.36 28.968 5.4881 Ho
re 39118.72 2 19559.36 28.968 3.3541 Ho Accept
Reject Interaction 9613.778 4 2403.444 3.560 4.1056 Ho
Interaction 9613.778 4 2403.444 3.560 2.7277 Ho
Error 18230.75 27 675.213
Error 18230.75 27 675.213

19
 2/2/2026

The Two-Factor Factorial Design The Two-Factor Factorial Design

Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

To assist in interpreting the results of this experiment, it is helpful to construct The significant interaction is indicated by the lack of parallelism of the lines.
a graph of the average responses at each treatment combination. This graph is In general, longer life is attained at low temperature, regardless of material
shown in Figure 7. type.

Changing from low to intermediate temperature, battery life with material

type 3 may actually increase, whereas it decreases for types 1 and 2.

From intermediate to high temperature, battery life decreases for material

types 2 and 3 and is essentially unchanged for type 1.

Material type 3 seems to give the best results if we want less loss of effective
life as the temperature changes.

Figure 7. Material type–temperature plot for Example 1

3 4
9 0
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

20
 2/2/2026

The Two-Factor Factorial Design The Two-Factor Factorial Design

Model Adequacy Checking Model Adequacy Checking
Subir Chowdhury School of Quality and Reliability

Subir Chowdhury School of Quality and Reliability

.
Residuals = 𝑒 𝑦 𝑦 𝑦 𝑦 . 𝑦
TABLE 6 . Residuals for Example 1

FIGURE 8. Normal probability plot of residuals for Example 1

4 4
1 2
Indian Institute of Technology, Kharagpur Indian Institute of Technology, Kharagpur

21
 2/2/2026

Model Adequacy Checking

Residual Plot
1.200

1.000

0.800
F(x)

0.600

0.400

0.200

0.000
-80 -60 -40 -20 0 20 40 60
Residuals, Eijk

22