02-02-2026

Subir Chowdhury School of Quality and

DOE
Randomized Block Design
Reliability

Prof. V N Achutha Naikan

Indian Institute of Technology, Kharagpur

The Randomized Complete Block Design

Subir Chowdhury School of Quality and Reliability

In any experiment, variability arising from a nuisance factor can affect the
results.

Generally, we define a nuisance factor as a design factor that probably has

an effect on the response, but we are not interested in that effect.

Sometimes a nuisance factor is unknown and uncontrolled; that is, we don’t

know that the factor exists, and it may even be changing levels while we are
conducting the experiment.

Randomization is the design technique used to guard against such a

“lurking” nuisance factor.

In other cases, the nuisance factor is known but uncontrollable.

2
Indian Institute of Technology, Kharagpur

1
 02-02-2026

Subir Chowdhury School of Quality and Reliability

The Randomized Complete Block Design

If we can at least observe the value that the nuisance factor takes on at each
run of the experiment, we can compensate for it in the statistical analysis by
using the analysis of covariance.

When the nuisance source of variability is known and controllable, a design

technique called blocking can be used to systematically eliminate its effect
on the statistical comparisons among treatments.

Blocking is an extremely important design technique used extensively in

industrial experimentation .

3
Indian Institute of Technology, Kharagpur

Source of Variability and Blocking

• Source of variability
– Unknown and uncontrollable
– Known and Controllable
– Known and Uncontrollable
• If it is known and controllable, a design technique called blocking can
be used to systematically eliminate its effect.

2
 02-02-2026

Example 1: Hardness Testing

• Consider a hardness testing machine that
presses a rod with a pointed tip into a
metal specimen with a known force.
• By measuring the depth of the depression
caused by the tip, the hardness of the
specimen is determined. Two different
tips are available for this machine.
• An experiment could be performed as
follows. A number of metal specimens
(e.g., 20) could be randomly selected.
Half of these specimens could be tested
by tip 1 and the other half by tip 2.
• The exact assignment of specimens to
tips would be randomly determined.
• This is a completely randomized design.

Example 1: Hardness Testing

Subir Chowdhury School of Quality and Reliability

The order in which the tips are tested for a particular specimen would also
be randomly selected. The experiment, when performed according to this
design with 10 specimens, produced the (coded) data shown in Table 1.

Table 1

6
Indian Institute of Technology, Kharagpur

3
 02-02-2026

Subir Chowdhury School of Quality and Reliability

Example 1: Hardness Testing
Suppose now that we wish to determine whether or not four different tips
produce different readings on a hardness testing machine.

The experimenter has decided to obtain four observations on Rockwell C-

scale hardness for each tip.

There is only one factor—tip type (4) – 4 replications = 4 x 4= 16 runs with

metal coupons

7
Indian Institute of Technology, Kharagpur

Example 1: Hardness Testing Variabilities

Subir Chowdhury School of Quality and Reliability

• Thus, 16 different metal test coupons would be required in

this experiment, one for each run in the design
• There is a potentially serious problem with a completely
randomized experiment in this design situation.
• If the metal coupons differ slightly in their hardness, (as
might happen if they are taken from ingots that are produced
in different heats), the experimental units (the coupons) will
contribute to the variability observed in the hardness data.
• As a result, the experimental error will reflect both
- Random error and
-Variability between coupons.

8
Indian Institute of Technology, Kharagpur

4
 02-02-2026

Subir Chowdhury School of Quality and Reliability

The Randomized Complete Block Design

We would like to make the experimental error as small as possible; that is,
we would like to remove the variability between coupons from the
experimental error.

A design that would accomplish this requires the experimenter to test each
tip once on each of four coupons.

This design, shown in Table 2 , is called a randomized complete block

design (RCBD). The word “complete” indicates that each block (coupon)
contains all the treatments (tips).

By using this design, the blocks, or coupons, form a more homogeneous

experimental unit on which to compare the tips.

Effectively, this design strategy improves the accuracy of the comparisons

among tips by eliminating the variability among the coupons.
9
Indian Institute of Technology, Kharagpur

The Randomized Complete Block Design

Subir Chowdhury School of Quality and Reliability

Table 2

1
0
Indian Institute of Technology, Kharagpur

5
 02-02-2026

Subir Chowdhury School of Quality and Reliability

The Randomized Complete Block Design

• Within a block, the order in which the four tips are tested is
randomly determined.
• The RCBD is one of the most widely used experimental
designs.
• Situations for which the RCBD is appropriate are numerous.
- Units of test equipment or machinery are often
different in their operating characteristics and would be a
typical blocking factor.

- Batches of raw material, people, and time are also

common nuisance sources of variability in an experiment
that can be systematically controlled through blocking.

1
1
Indian Institute of Technology, Kharagpur

Statistical Analysis of the RCBD

Subir Chowdhury School of Quality and Reliability

Suppose we have, in general, a treatments that are to be compared and b

blocks. The randomized complete block design is shown in Figure 1.

There is one observation per treatment in each block, and the order in which
the treatments are run within each block is determined randomly. Because the
only randomization of treatments is within the blocks, we often say that the
blocks represent a restriction on randomization.

Figure 1. The randomized

complete block design

1
2
Indian Institute of Technology, Kharagpur

6
 02-02-2026

Subir Chowdhury School of Quality and Reliability

Statistical Analysis of the RCBD

(1)

we usually think of the treatment and block effects as deviations from the
overall mean so that

1
3
Indian Institute of Technology, Kharagpur

Statistical Analysis of the RCBD

Subir Chowdhury School of Quality and Reliability

It is also possible to use a means model for the RCBD, say

1
4
Indian Institute of Technology, Kharagpur

7
 02-02-2026

Subir Chowdhury School of Quality and Reliability

Statistical Analysis of the RCBD

1
5
Indian Institute of Technology, Kharagpur

Statistical Analysis of the RCBD

Subir Chowdhury School of Quality and Reliability

1
6
Indian Institute of Technology, Kharagpur

8
 02-02-2026

Subir Chowdhury School of Quality and Reliability

Statistical Analysis of the RCBD

It can be shown that the cross products (other terms) are zero.

1
7
Indian Institute of Technology, Kharagpur

Statistical Analysis of the RCBD

Subir Chowdhury School of Quality and Reliability

The expected value of the mean squares, if treatments and blocks are fixed,
can be shown to be

1
8
Indian Institute of Technology, Kharagpur

9
 02-02-2026

Subir Chowdhury School of Quality and Reliability

Statistical Analysis of the RCBD

Computing formulas can be expressed in terms of treatment and block totals.

These formulas are

1
9
Indian Institute of Technology, Kharagpur

Statistical Analysis of the RCBD

Subir Chowdhury School of Quality and Reliability

2
0
Indian Institute of Technology, Kharagpur

10
 02-02-2026

Subir Chowdhury School of Quality and Reliability

Statistical Analysis of the RCBD
Table 3

𝑀𝑆
𝑀𝑆

2
1
Indian Institute of Technology, Kharagpur

Example 2: Vascular Grafts

Subir Chowdhury School of Quality and Reliability

A medical device manufacturer produces vascular grafts (artificial veins).

These grafts are produced by extruding billets of polytetrafluoroethylene
(PTFE) resin combined with a lubricant into tubes. Frequently, some of the
tubes in a production run contain small, hard protrusions on the external
surface. These defects are known as “flicks.” The defect is cause for rejection
of the unit.

The product developer responsible for the vascular grafts suspects that the
extrusion pressure affects the occurrence of flicks and therefore intends to
conduct an experiment to investigate this hypothesis. However, the resin is
manufactured by an external supplier and is delivered to the medical device
manufacturer in batches. The engineer also suspects that there may be
significant batch-to-batch variation, because while the material should be
consistent with respect to parameters such as molecular weight, mean particle
size, retention, and peak height ratio, it probably isn’t due to manufacturing
variation at the resin supplier and natural variation in the material.
2
2
Indian Institute of Technology, Kharagpur

11
 02-02-2026

Example 2: Vascular Grafts

Therefore, the product developer decides to investigate the effect of four different
Subir Chowdhury School of Quality and Reliability

levels of extrusion pressure on flicks using a randomized complete block design

considering batches of resin as blocks. The RCBD is shown in Table 4.

Note that there are four levels of extrusion pressure (treatments) and six batches of
resin (blocks). Remember that the order in which the extrusion pressures are tested
within each block is random.

The response variable is yield, or the percentage of tubes in the production run that
did not contain any flicks.

2
3
Indian Institute of Technology, Kharagpur

Randomized Complete Block Design

for the Vascular Graft Experiment
Table: Percentage of Defect free Vascular Grafts
Batch of Resin (Block) : j = 1..b
Extrusion
Pressure (PSI):
i = 1..a 1 2 3 4 5 6
8500 90.3 89.2 98.2 93.9 87.4 97.9
8700 92.5 89.5 90.6 94.7 87 95.8
8900 85.5 90.8 89.6 86.2 88 93.4
9100 82.5 89.5 85.6 87.4 78.9 90.7

12
 02-02-2026

Subir Chowdhury School of Quality and Reliability

Example 2: Vascular Grafts

Table 4

2
5
Indian Institute of Technology, Kharagpur

Example
Subir Chowdhury School of Quality and Reliability

2
6
Indian Institute of Technology, Kharagpur

13
 02-02-2026

Subir Chowdhury School of Quality and Reliability

Example

The ANOVA is shown in Table 5 on the next slide.

2
7
Indian Institute of Technology, Kharagpur

Example 2: Vascular Grafts

Table 5
Subir Chowdhury School of Quality and Reliability

2
8
Indian Institute of Technology, Kharagpur

14
 02-02-2026

ANOVA for RCBD

SL,α=0.05

Analysis of Variance for the Vascular Experiment (RCBD)

Source of Sum of Mean Sq F0.05

DF Fo
variation Squares Error (Table)

Treatments
178.171 3 59.390 8.107 3.287 Reject Ho
(Extr. Pr.)
Blocks
192.252 5 38.450 5.248 2.901 Reject Ho
(batches)
Error 109.886 15 7.325
Total 480.309 23

Analysis of Variance for the Vascular Experiment

(RCD) : No Block design, SL,α=0.05

Analysis of Variance for the Vascular Experiment (RCD) : No Block design

Source of Sum of Mean Sq F0.05
Fo
variation Squares DF Error (Table)
Treatments
(Extr. Pr.) 178.171 3 59.390 3.931 3.098 Reject Ho
Blocks (batches)
Error 302.138 20 15.107
Total 480.309 23

15
 02-02-2026

Analysis of Results
• We would still reject the null hypothesis and conclude that
extrusion pressure significantly affects the mean yield.
• However, note that the mean square for error has more than
doubled, increasing from 7.33 in the RCBD to 15.11.
• All of the variability due to blocks is now in the error term.
• This makes it easy to see why we sometimes call the RCBD a noise-
reducing design technique; it effectively increases the signal-to-
noise ratio in the data, or it improves the precision with which
treatment means are compared.
• This example also illustrates an important point. If an experimenter
fails to block when he or she should have, the effect may be to
inflate the experimental error, and it would be possible to inflate
the error so much that important differences among the treatment
means could not be identified.

ANOVA for RCBD

SL,α=0.01

Analysis of Variance for the Vascular Experiment (RCBD)

Source of Sum of Mean Sq F0.01

DF Fo
variation Squares Error (Table)

Treatments
178.171 3 59.390 8.107 5.417Reject Ho
(Extr. Pr.)
Blocks
192.252 5 38.450 5.248 4.556Reject Ho
(batches)
Error 109.886 15 7.325
Total 480.309 23

16
 02-02-2026

Analysis of Variance for the Vascular Experiment

(RCD) : No Block design, SL,α=0.01

Analysis of Variance for the Vascular Experiment (RCD) : No Block design

Source of Sum of Mean Sq F0.01
Fo
variation Squares DF Error (Table)
Treatments Do NOT
(Extr. Pr.) 178.171 3 59.390 3.931 4.94 Reject Ho
Blocks (batches)
Error 302.138 20 15.107
Total 480.309 23

Example 2: Vascular Grafts

Subir Chowdhury School of Quality and Reliability

Model Adequacy Checking

• As in the completely randomized design, residual analysis is the
major tool used in this diagnostic checking. The residuals for the
randomized block design in Example are listed in Figure 2:

• A normal probability plot of these residuals is shown in Figure 3.

• There is no severe indication of non-normality, nor is there any

evidence pointing to possible outliers.
• Following Formula can be used for calculation of Residuals.
Residuals 𝑒 𝑦 𝑦
𝑦 𝜇 𝜏 𝛽 𝜖
𝑦 𝐸 𝑦 𝑦.. 𝑦 . 𝑦.. 𝑦. 𝑦.. 0 𝑦 . 𝑦. 𝑦..
𝑒 𝑦 𝑦 . 𝑦. 𝑦..
3
4
Indian Institute of Technology, Kharagpur

17
 02-02-2026

Example 2: Vascular Grafts Model Adequacy Checking

Subir Chowdhury School of Quality and Reliability

• As in the completely randomized design, residual analysis is

the major tool used in this diagnostic checking.
• The residuals for the randomized block design in Example
𝑒 𝑦
𝑦 . 𝑦. 𝑦..
OR
𝑒 𝑦 𝑦 . 𝑦. 𝑦..
• A normal probability plot of these residuals is shown in Figure
3.
• There is no severe indication of non-normality, nor is there any
evidence pointing to possible outliers.

3
5
Indian Institute of Technology, Kharagpur

Predicted yij and Residuals

Predicted yij Residuals
𝑒 90.3 90.72 0.42
Predicted values of yij Residuals
1 2 3 4 5 6 1 2 3 4 5 6
90.72 92.77 94.02 93.57 88.35 97.47 -0.42 -3.57 4.18 0.33 -0.95 0.43
89.59 91.64 92.89 92.44 87.21 96.34 2.91 -2.14 -2.29 2.26 -0.21 -0.54
86.82 88.87 90.12 89.67 84.45 93.57 -1.32 1.93 -0.52 -3.47 3.55 -0.17
83.67 85.72 86.97 86.52 81.30 90.42 -1.17 3.78 -1.37 0.88 -2.40 0.28

𝑦 . 𝑦. 𝑦..
556.9 350.8 2155.1
6 4 24
90.72

18
 02-02-2026

Subir Chowdhury School of Quality and Reliability

Example

Figure 2: Residual
Estimation

Note: Predicted
values include block
corrections

3
7
Indian Institute of Technology, Kharagpur

Normality Check
Residua
x ls F(x)
1 -3.57 0.02
2 -3.47 0.06
3 -2.40 0.10 Normality Plot
4 -2.29 0.15 1.20
5 -2.14 0.19
6 -1.37 0.23 1.00
7 -1.32 0.27
8 -1.17 0.31 0.80

9 -0.95 0.35
F(x)

0.60
10 -0.54 0.40
11 -0.52 0.44
0.40
12 -0.42 0.48
13 -0.21 0.52 0.20
14 -0.17 0.56
15 0.28 0.60 0.00
16 0.33 0.65 -4.00 -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00
17 0.43 0.69 Residuals
18 0.88 0.73
19 1.93 0.77
20 2.26 0.81
21 2.91 0.85
22 3.55 0.90
23 3.78 0.94
24 4.18 0.98
Sum= 0.00

19