UNIT 14

BALANCED INCOMPLETE
BLOCK DESIGN

Structure
14.1 Introduction 14.7 Complimentary Balanced
Incomplete Block Design
Expected Learning Outcomes
Construction of a Complimentary
14.2 Complete and Incomplete
Balanced Incomplete Block Design
Block Designs
14.8 Residual and Derived
Complete Block Designs
Balanced Incomplete Block
Incomplete Block Designs Design
14.3 Introduction to Balanced Residual Balanced Incomplete
Incomplete Block Design Block Design

Definition of a Balanced Incomplete Construction of Residual Balanced

Block Design Incomplete Block Design

Parameters of a Balanced Derived Balanced Incomplete Block

Incomplete Block Design Design
14.4 Incidence Matrix of a 14.9 Summary
Balanced Incomplete Block
14.10 Terminal Questions
Design
14.11 Solutions/Answers
14.5 Symmetric Balanced
Incomplete Block Design

14.6 Resolvable and Affine

Resolvable Balanced
Incomplete Block Design

14.1 INTRODUCTION
In Design of Experiments, generally different treatments which are to be
compared with each other in an experiment, need to be applied to a number of
blocks either once or may be replicated a number of times depending upon the
available resources allocated to the experiment, such as, amount of
experimental material, money, labour, time, etc. However, in the Unit 14 under

107
 Block 4 Incomplete Block Design

Block 4 of the course ‘Survey Sampling and Design of Experiments - I’ of

MSCAST programme; it has been mentioned that if the number of treatments
to be compared is large, there need to be a large number of blocks to
accommodate all the treatments. This requires more experimental material
and so the cost of experimentation becomes high. The Completely
Randomised Design and Randomised Block Design which are “Complete
Block Designs” may not be suitable in such situations. because they will
require a large number of experimental units to accommodate all the
treatments. The designs in which every block receives all the treatments are
called the “Complete Block Designs”.
In such cases, when the sufficient number of homogeneous experimental units
are not available to accommodate all the treatments in a block, then,
obviously, some alternative design is to be used. In fact, if in a block the
number of units or plots is smaller than the number of treatments, then the
block is said to be Incomplete Block and a design constituted of such blocks
is called an “Incomplete Block Design”.
This unit is aimed at to explain why and under what circumstances, incomplete
block designs seem to be preferable than complete block designs; what other
different types of incomplete block designs exist in the literature along with
their definitions, salient features and applications. In particular, the unit
highlights a special case of incomplete block design, namely, “Balanced
Incomplete Block Design” (B.I.B.D.). Some related theories and concepts
associated with Balanced Incomplete Block Design (B.I.B.D.) have also been
discussed.
In the Section 14.2 of this unit, we have tried to emphasise the concept of
“Complete Block Design” and ‘Incomplete Block Design” and to make the
basic difference between these two types of designs, which is necessary to
understand for Balanced Incomplete Block Designs (B.I.B.D.). Section 14.3
has exclusively been devoted to give a concise definition of the Balanced
Incomplete Block Design (B.I.B.D.), explanation of the parameters which
signifies a Balanced Incomplete Block Design (B.I.B.D.) and their inter-
relations and the actual meaning of balancing in the design.
Section 14.4 describes the concept of “Incidence Matrix” of a Balanced
Incomplete Block Design (B.I.B.D.) and its role in understanding the meaning
of a Balanced Incomplete Block Design (B.I.B.D.). The procedure of finding
the incidence matrix of a Balanced Incomplete Block Design (B.I.B.D.) has
also been explained. In Section 14.5, we have introduced the concept of
“Symmetric Balanced Incomplete Block Design (B.I.B.D.).” which plays an
important role, later on, for understanding the concepts of some other types of
Balanced Incomplete Block Design (B.I.B.D.). While the Section 14.6
introduces the idea and concepts of “Resolvable and Affine Resolvable
Balanced Incomplete Block Design (B.I.B.D.)”; the Section 14.7 defines the
meaning of “Complimentary Balanced Incomplete Block Design (B.I.B.D.).”
and their construction on the basis of a given symmetric Balanced Incomplete
Block Design (B.I.B.D.). Section 14.8 deals with the study of two other types of
Balanced Incomplete Block Design (B.I.B.D.), namely, “Residual Balanced
Incomplete Block Design (B.I.B.D.)” and “Derived Balanced Incomplete Block
Design (B.I.B.D.)”. The section also explains how to construct a residual
design given a symmetric Balanced Incomplete Block Design (B.I.B.D.).
108
Unit 14 Balanced Incomplete Block Design

Expected Learning Outcomes

After studying this unit, you should be able to:
❖ explain the meaning of a ‘Complete Block Design’ and an ‘Incomplete
Block Design’ and the difference between these too concepts;
❖ define the concept of “Balanced Incomplete Block Design” (B.I.B.D.) and
the meaning of its parameters and their inter-relations;
❖ define the ‘Incidence Matrix’ of a Balanced Incomplete Block Design
(B.I.B.D.) and can explain the steps of obtaining the incidence matrix of
a given B.I.B.D.;
❖ Give the concept of ‘Symmetric Balanced Incomplete Block Design
(B.I.B.D.).’ which is very important for understanding many of the
concepts of new B.I.B.D.s which could be derived for a given symmetric
Balanced Incomplete Block Design (B.I.B.D.).;
❖ mention briefly the idea of ‘Resolvable’ and ‘Affine resolvable’ Balanced
Incomplete Block Design (B.I.B.D.);
❖ explain what is meant by ‘Complimentary Balanced Incomplete Block
Design (B.I.B.D.)’ and its importance in the study of several designs.
You can also explain the steps of constructing a complimentary
Balanced Incomplete Block Design (B.I.B.D.) given a symmetric B.I.B.
design;
❖ explain the concepts of ‘Residual Balanced Incomplete Block Design
(B.I.B.D.)’ and Balanced Incomplete Block Design (B.I.B.D.)’ and the
construction of a residual Balanced Incomplete Block Design (B.I.B.D.).

14.2 COMPLETE AND INCOMPLETE BLOCK

DESIGNS
In order to understand the concept of Balanced Incomplete Block Designs
(B.I.B.D.); the subject matter to be discussed in this unit, it would be necessary
to know the meanings of ‘Complete Block Design’ and the ‘Incomplete Block
Design’, a mention of which has been made in the Introductory section above.
14.2.1 Complete Block Designs
Having decided on the block size and having grouped the experimental units
into blocks of similar units, the next step in any design of experiments, is to
assign the units to the levels of the treatment factors. One way of assigning of
experimental units to treatments might be to assign all the units within a block
to one treatment, all units within another block to a second treatment, and so
on; but this type of assignment would possibly be the worst because it would
not allow the analysis to distinguish block differences from treatment
differences. Perhaps, the best possible assignment would be one that
allocates to every treatment the same number of experimental units per block.
This can be achieved only when the block size is a multiple of the number of
treatments, that is, if the block size is denoted by k and the number of
treatments is denoted by , then k must be a multiple of . Such designs are
called “Complete Block Designs”. A special case of ‘Complete Block 109
 Block 4 Incomplete Block Design

Design’, where the block size is exactly equal to the number of treatments,
that is, k = , has been historically called “Randomised Complete Block
Design” (RCBD) or, more popularly as “Randomised Block Design” (RBD).
14.2.2 Incomplete Block Designs
If the block size is not a multiple of the number of treatments, that is, when k
is not a multiple of , then the block design is known as an “Incomplete
Block Design”. This term is sometimes reserved for the smaller designs
where the block size is smaller than the number of treatments, that is, when
k  ; but it is found convenient to classify all designs as either complete or
incomplete. Incomplete Block Designs are more complicated to design and
analyse than Complete Block Designs.
Thus, in the simplest way, it can be stated as:
The designs in which every block receives all the treatments are called the
Complete Block Designs, whereas, the designs in which every block does not
receive all the treatments but only some of the treatments are called
Incomplete Block Designs.
Examples of Incomplete Blocks Design:
Example 1: Consider an experiment related to the comparison of three brands
of Eyedrops for their ability to reduce eye irritation. As there is a strong
individual effect, patients are used in the experiment as blocks. If a patient can
only be tested in one treatment period, the researchers can apply one brand of
drop in the left eye and another brand of drop in the right eye. The natural
block size is, thus, limited to k = 2.
Then, the study may be looked upon as an incomplete block design, with block
size, k = 2 and number of treatments,  = 3, that is, k  .
Example 2: In this experiment the goal is to compare 5 commercial
advertisements, namely, A, B, C, D and E. For this, views of some experts will
be considered after they watch these advertisements. An expert may watch
multiple of advertisements. We can consider the experts as blocks in the
experiment. Although, the experiment can be considered to be a Randomised
Block Design if all the experts can watch all the five advertisements, but
generally, experts may lose patience after watching too many advertisements,
and they may also forget the first few advertisements they see. Therefore,
their response will be less accurate.
To ensure the quality of the response of subjects, we may restrict the number
of advertisements each expert watch to, say, 3. Thus, block size is limited to
k = 3. Let the following table shows the assignment of treatments to blocks:
Table 14.1: Design showing Assignment of Treatments to Blocks
Expert (Block)

1 2 3 4 5 6 7 8 9 10
A B E A C D B E D C
Ads. B D A C A E C B E D
C A B D E A D C B E

Obviously, it is an example of Incomplete Block Design.

110
Unit 14 Balanced Incomplete Block Design
Now, you may try to answer the following self-assessment question:
SAQ 1
Differentiate between the Complete and Incomplete Block Designs. Give an
example of Incomplete Block Design.

14.3 INTRODUCTION TO BALANCED

INCOMPLETE BLOCK DESIGN
A Balanced Incomplete Block Design is, in fact, a special kind of Incomplete
Block Designs.
Generally, in Incomplete Block Designs, in order to ensure equal or nearly
equal precision of comparisons of different pairs of treatments, the treatments
are so allotted to the different blocks that each pair of treatments has the same
or nearly the same number of replications and each treatment has an equal
number of replications, say, r. Different patterns of values of the number of
replications of different pairs of treatments in a design provides different types
of incomplete block designs. However, when the number of replications of all
pairs of treatments in a design is the same, then an important series of
designs known as “Balanced Incomplete Block Design” (B.I.B.D.) is
obtained.
14.3.1 Definition of a Balanced Incomplete Block Design
A Balanced Incomplete Block Design may be defined as follows:
“A Balanced Incomplete Block Design (B.I.B.D.) is a design with  treatment
labels, each occurring r times, and with bk experimental units grouped into b
blocks of size k <  in such a way that the units within a block are alike and
units in different blocks are substantially different.
The plan of the design satisfies the following conditions:
(i) The design is binary (that is, each treatment label appears either once or
not at all in a block).
(ii) Each pair of labels appears together in  blocks, where  is a fixed
integer.”
Alternatively, we can define a Balanced Incomplete Block Design (B.I.B.D.) as
“An Incomplete Block Design with  treatments distributed over b blocks,
each of size k , where k is less than  such that each treatment occurs in r
blocks, no treatment occurs more than once in a block and each pair of
treatments occurs together in  blocks, is called a Balanced Incomplete Block
Design.”
14.3.2 Parameters of a Balanced Incomplete Block
Design
It can be seen that a Balanced Incomplete block design can completely be
characterised by five of its parameters, which are as follows:
(i)  : Number (label) of treatments;
(ii) b : Number of blocks;
(iii) k : Block size; 111
 Block 4 Incomplete Block Design

(iv) r : Number of replications;

(v) λ : a fixed integer.
Evidently, there are certain relations which the parameters of a balanced
incomplete block design must satisfy. These are
(i) r = bk = Total number of plots in the design and
(ii) λ (  − 1) = r ( k − 1) = Total number of replications of all possible pairs of
treatments in each of which any particular treatment is always present.
However, one of main drawbacks of a balanced incomplete block design is
that it may not exist for many choices of these parameters. In fact, Balanced
Incomplete Block Design (B.I.B.D.) is not balanced in the general sense that
all treatment-block combinations occur equally often. Rather they are balanced
in the looser sense by the criteria described below:
A “Balanced Incomplete Block Design” with
 treatments, b blocks, k as the size of each block, r replications of each
treatment is a design satisfying the followings:
(i) ‘Incompleteness’ implying that k  

(ii) ‘Balancing’ implying that each treatment appears at most once per
block and has the same number of replicates r.
(iii) Each pair of treatments appear in a block the same number of times λ.
We shall explain this fact below with the help of some of the examples of
incomplete block designs:
Example 3: Consider again the Example 2, presented under the Sub-section
14.2.2. The design of this experiment (shown below) is a Balanced Incomplete
Block Design with contents of the blocks as follows:
Expert (Block)

1 2 3 4 5 6 7 8 9 10

A B E A C D B E D C

Ads. B D A C A E C B E D

C A B D E A D C B E

We observe from the above design that

 = Number of Treatments = 5
b = Number of Blocks = 10
k = Size of each Block = 3

r = Number of Replications per Treatment = 6.

The following table shows the blocks to which a particular treatment is
allocated. For example, it shows that A has been allocated to blocks 1, 2, 3, 4,
5 and 6; B has been allocated to blocks 1, 2, 3, 7, 8 and 9; C has been
allocated to blocks 1, 4, 5, 7, 8 and 10 and so on. Thus, every treatment has
been replicated to same number of replications, which is 6 here.
112
Unit 14 Balanced Incomplete Block Design

BLOCK

Treatment 1 2 3 4 5 6 7 8 9 10

A      
B      
C      
D      
E      

Further, Balanced Incomplete Block Design (B.I.B.D.) requires each pair of

treatments appears in a block the same number (λ) of times. The following
Two-way table is prepared between treatment-pairs and blocks to verifies this
fact. In the table, we have all the treatment-pairs in the first row and it has
been observed which treatment-pair to which blocks and how many times.
BLOCK

Treatment-
1 2 3 4 5 6 7 8 9 10
Pair ↓
AB ✓ ✓ ✓
AC ✓ ✓ ✓
AD ✓ ✓ ✓
AE ✓ ✓ ✓
BC ✓ ✓ ✓
BD ✓ ✓ ✓
BE ✓ ✓ ✓
CD ✓ ✓ ✓
CE ✓ ✓ ✓
DE ✓ ✓ ✓

From this table, it is seen that every treatment-pair occurs in a block same
number of times which is 3. Thus, λ = 3 in the design.
It is now easy to verify that the three conditions to be satisfied by a design to
be a Balanced Incomplete Block Design (B.I.B.D.) as mentioned under the
Sub-section 14.3.2 are satisfied, implying that the concerned design is a
Balanced Incomplete Block Design (B.I.B.D.).
The two conditions of Balanced Incomplete Block Design (B.I.B.D.), mentioned
as
(i) r = bk
(ii) λ (  − 1) = r ( k − 1)

are respectively known as “First Balancing Condition” and “Second

Balancing Condition”. Further, from condition (ii), we have
r ( k − 1) 12
λ= = = 3, which is an integer.
(  − 1) 4

Example 4: Suppose an experiment is to be run to compare  = 7

compositions of a metal alloy in terms of its tensile strength. Further, suppose
that only three observations can be taken per day and that the experiment
113
 Block 4 Incomplete Block Design

must be completed within seven days. It may be thought advisable to divide

the experiment into blocks, with each day representing a block, since different
technicians may work on the experiment on different days and the laboratory
temperature may vary from day to day. Thus, an incomplete block design with
b = 7 blocks of size k = 3 and with  = 7 treatment labels is needed.
The plan is shown below in the first column of Table 14.2 is of the correct size.
Table 14.2: Randomisation of a Complete Block Design

Unrandomised Design after Design after

Block Label Day
Design step (i) step (ii)
I 1 2 4 1 6 7 2 2 7 6
II 2 3 5 2 2 3 5 5 3 2
III 3 4 6 3 3 4 6 4 3 6
IV 4 5 7 4 1 2 4 2 1 4
V 5 6 1 5 5 6 1 1 5 6
VI 6 7 2 6 4 5 7 4 5 7
VII 7 1 3 7 7 1 3 7 3 1

It is binary, with every treatment appearing 0 or 1 times per block and r = 3

times in total. Also, all pairs of treatments occur together in a block exactly
once, so the treatment labels are evenly distributed over the blocks.
Randomisation now proceeds in two steps. We explain here only the step(i) of
randomisation.
Step (i): The block labels need to be randomly assigned to days. Suppose we
obtain the following pairs of random digits from a random number generator
and associate them with the blocks as follows:
Random digits: 71 36 65 93 92 02 97
Block labels: I II III IV V VI VII
Then, sorting the random numbers into ascending order, the blocks are
reordered and assigned to the seven days as follows:
Block labels: VI II III I V IV VII
Days: 1 2 3 4 5 6 7
The randomly ordered blocks after step (i) are shown in the fourth column of
Table 14.2. Similarly, the randomly ordered blocks after step (ii) are shown in
the fifth column of Table 14.2.
Since, in the above design,  = 7,b = 7, k = 3 and r = 3 ; the two conditions of a
Balanced Incomplete Block Design (B.I.B.D.) are satisfied and we get  = 1, an
integer. So, the design is a Balanced Incomplete Block Design (B.I.B.D.).
Example 5: Let us now consider Table 14.3 given below showing a design
with  = b = 8;r = k = 3.

Table 14.3: An Incomplete Block Design

Block Block
I 1 3 8 V 5 7 4
II 2 4 1 VI 6 8 5
III 3 5 2 VII 7 1 6
IV 4 6 3 VIII 8 2 7
114
Unit 14 Balanced Incomplete Block Design

The first condition of a Balanced Incomplete Block Design (B.I.B.D.) given as

r = bk is satisfied, both being 24. However, we can find the parameter λ
using the second condition given as λ(  − 1) = r (k − 1) .

We have
r ( k − 1) 6
λ= = , which is not an integer.
(  − 1) 7

Since λ is not an integer, it indicates that a Balanced Incomplete Block Design

cannot exist with this information.
Now, you may try to answer the following self-assessment question:

SAQ 2
Define a Balanced Incomplete Block Design (B.I.B.D.) and mention its
parameters with their meanings. Give at least example of a design which is a
Balanced Incomplete Block Design (B.I.B.D.).

14.4 INCIDENCE MATRIX OF A BALANCED

INCOMPLETE BLOCK DESIGN
We will now look at one way to describe the information in a Balanced
Incomplete Block Design (B.I.B.D.) by putting it in matrix form.
A Balanced Incomplete Block Design (B.I.B.D.) with parameters (  , b,r,k,  ) is
a pair (X, ) that satisfies the following properties:
1. X is a set of  elements (called points).
2. is a family of b subsets of X, each of cardinality k (called blocks).

3. Every point occurs in exactly r blocks.

4. Every pair of distinct points occurs in exactly  blocks.
It is generally required that k  ; that is why B.I.B.D.s are called Incomplete
Block Designs. It is to see that the five parameters are not independent but are
related as
r = b k and λ (  − 1) = r ( k − 1) .

Hence, since the other two parameters can then be deduced, it is not
uncommon to write a Balanced Incomplete Block Design (B.I.B.D.) using the
three parameters ,k,  , that is, as a ( ,k,  ) − B.I.B.D.

Let (X, ) be a Balanced Incomplete Block Design (B.I.B.D.) with parameters

(  , b,r,k,  ), where X = { x1, x2 ,,x } and = { A 1 , A 2 ,, Ab }. Then
the Incidence Matrix of this B.I.B.D. is the  × b matrix M whose
0 if x i  A j
entries mij are defined as mij =  .
1 if x i  A j
In other words, the Incidence Matrix of a (  , b,r,k,  ) – B.I.B.D. is a matrix
containing  rows (each row representing a point), and b columns (each
column representing a block), where the entry in row i and column j is 0 if the
point x i not contained in the block A j , and where the entry in row i and 115
 Block 4 Incomplete Block Design

column j is 1 if the point x i is contained in the block A j . Thus, M tells us which

points of X are contained in which blocks of .
Let us take an example of a (7, 7, 3, 3, 1) – B.I.B.D. with
X = { 1, 2, 3, 4, 5, 6, 7} and = [{1,2,3}, {3,4,5}, {5,6,1}, {1,7,4}, {2,7,5},
{3,7,6}, {2,4,6}].
Then, obviously we observe that m11 = 1, since, x1 A1 = 1. 2. 3; m12 = 0,
since, x1 A 2 = 3, 4, 5; m13 = 1, since, x1 A 3 = 5, 6,1; m14 = 1, since,
x1 A 4 = 1, 7, 4; m15 = 0 , since, x1 A 5 = 2, 7, 5; etc. Similarly, obtaining rest
of the mij values, we get the incidence matrix of the given Balanced
Incomplete Block Design (B.I.B.D.) as
1 0 1 1 0 0 0
1 0 0 0 1 0 1

1 1 0 0 0 1 0
 
M = 0 1 0 1 0 0 1
0 1 1 0 1 0 0
 
0 0 1 0 0 1 1
0 0 
 0 0 1 1 1

The incidence matrix of a Balanced Incomplete Block Design (B.I.B.D.)

satisfies the equation
MMT = (r − λ ) I + λJ;

where I is a   identity matrix and J is a   matrix of 1’s.

Now, you may try to answer the following self-assessment question:

SAQ 3
What is meant by the Incidence Matrix of a B.I.B.D.? Explain how would you
obtain the incidence matrix of a B.I.B.D.

14.5 SYMMETRIC BALANCED INCOMPLETE

BLOCK DESIGN
A Balanced Incomplete Block Design (B.I.B.D.) in which the number of blocks =
number of treatments, that is, b =  is called a “Symmetric Balanced
Incomplete Block Design (B.I.B.D.).”
Since, b =  for a symmetric B.I.B.D., so from the condition b = rk , we have
r = k for a symmetric B.I.B.D., Also, since  (  − 1) = r (k − 1) and r = k , for a
(
symmetric B.I.B.D. it follows that  (  − 1) = k 2 − k . )
It can be shown that in a symmetric B.I.B.D., the number of pairs of treatments
common between any two blocks = .

14.6 RESOLVABLE AND AFFINE RESOLVABLE

BALANCED INCOMPLETE BLOCK DESIGN
An Incomplete Block Design in which each treatment appears r times is
116 “Resolvable” if the blocks can be divided into r groups such that each group
Unit 14 Balanced Incomplete Block Design

is a complete replication of the treatments (that is, each group of blocks

contains each treatment exactly once).
It has been shown that a necessary condition for the existence of a Resolvable
Balanced Incomplete Block Design (B.I.B.D.) is given by b  + r − 1.
If equality holds, that is, if b = + r − 1, the B.I.B.D. is said to be “Affine
Resolvable”.

14.7 COMPLIMENTARY BALANCED

INCOMPLETE BLOCK DESIGN
Given a B.I.B.D. with certain parametric values, it is possible to construct a
new B.I.B.D. which is said to be the “Complimentary Balanced Incomplete
Block Design (B.I.B.D.)” of the original Balanced Incomplete Block Design
(B.I.B.D.) in fact, there are several methods by which new B.I.B.D.s can be
obtained from the old one. However, all such B.I.B.D.s cannot be the
complimentary of the original one. Being the complimentary one of the given
B.I.B.D., there must exist some kind of relations between the parametric sets
of the given B.I.B.D. and its complimentary one.
We shall show how the parameters of the given B.I.B.D. and that of its
compliment are related to each other.
The compliment of a B.I.B.D. is obtained by replacing every block by its
compliment with respect to the point set. If the given B.I.B.D. is a ( ,b,r,k,  ) −
B.I.B. D., then the compliment of it will be a ( ,b,b − r,  − k,b − 2r + ) −
B.I.B.D.

14.7.1 Construction of a Complimentary Balanced

Incomplete Block Design
From a given B.I.B.D., it is always possible to obtain a Complimentary B.I.B.D.
by doing certain steps with the parameters of the given B.I.B.D. For this, let us
consider the B.I.B.D. presented in the first two columns of the Table 14.2
above, which is reproduced below:
Table 14.4: Partial Reproduction of Table 14.2

Block Label Unrandomised Design

I 1 2 4
II 2 3 5
III 3 4 6
IV 4 5 7
V 5 6 1
VI 6 7 2
VII 7 1 3

It is seen that the parameters of this design are (  , b,r,k, ) = ( 7, 7, 3, 3,1) .

In order to construct the Complimentary B.I.B.D. of this ( 7, 7, 3, 3,1) − B.I.B.D.,

we follow the following step:
In the jth block of the complimentary design take all those treatments which do
not occur in the jth block of the original design. For example, in the original
117
 Block 4 Incomplete Block Design

design, the block I consists of treatments 1, 2 and 4; but do not consist of the
treatments 3, 5, 6 and 7. Therefore, in the complimentary design, block I will
consist of treatments 3, 5, 6 and 7. Similarly, repeating this process of
replacement in all the blocks, we get the Complimentary B.I.B.D. of the original
B.I.B.D. You can see that the complimentary design is given in following table:
Block Label Unrandomised Design
I 3 5 6 7
II 1 4 6 7
III 1 2 5 7
IV 1 2 3 6
V 2 3 4 7
VI 1 3 4 5
VII 2 4 5 6

Obviously, the parameters of the Complimentary B.I.B. Design will be (  * ,

b* ,r * ,k * , * ) = ( 7, 7, 4, 4, 2) .

Now, you may try to answer the following self-assessment question:

SAQ 4
What is a Complimentary B.I.B.D. of a given Symmetric B.I.B.D.? Explain the
method of obtaining a complimentary B.I.B.D. given the following B.I.B.D.:
Block I II III IV V VI VII
Design 2 7 6 5 3 2 4 3 6 2 1 4 1 5 6 4 5 7 7 3 1

14.8 RESIDUAL AND DERIVED BALANCED

INCOMPLETE BLOCK DESIGN
We mentioned under the Section that any two blocks of a Symmetric B.I.B.D.
have λ objects in common. Using this fact some new B.I.B.D.’s can be
constructed which we shall define here.
14.8.1 Residual Balanced Incomplete Block Design
Under the Section 14.4, we represented a B.I.B.D. in the matrix form as
(X, ) with parameters (  , b,r,k,  ), where X = { x1, x2 ,,x } and = { A1 ,
A 2 ,, Ab } with the properties:

1. X is a set of  elements (called points).

2. is a family of b subsets of X, each of cardinality k (called blocks).
3. Every point occurs in exactly r blocks.
4. Every pair of distinct points occurs in exactly  blocks.
Let us assume that (X, ) be a Symmetric (  , k, ) – B.I.B.D. and let A0 ∈
. Then, define
Res(X, , A0 ) = (X \ A0 , { A \ A0 : A  , A  A0 })

as the Residual Balanced Incomplete Block Design (B.I.B.D.).

118
Unit 14 Balanced Incomplete Block Design

It is clear that the residual design is constructed by deleting all points in a

block A0 .

It can be proved that the Residual B.I.B.D., Res(X, , A0 ) is a (  −k ,  −1

, k , k −  ,  ) - B.I.B.D. provided that k ≥  +2.

14.8.2 Construction of Residual Balanced Incomplete

Block Design
It is easy to construct a Residual B.I.B.D. from the given Symmetric B.I.B.D.
with specific parameters. In order to explain the construction process, let us
consider the Symmetric B.I.B.D. presented in Table 14.4, which has been
given below:
Block Label Unrandomized Design
I 1 2 4
II 2 3 5
III 3 4 6
IV 4 5 7
V 5 6 1
VI 6 7 2
VII 7 1 3

We have seen that the parameters of this B.I.B.D. are (, b, r, k,  )

= ( 7, 7, 3, 3,1) .

A Residual B.I.B.D. can be constructed by the process of block section. In

block section, any one of the blocks appearing in the given B.I.B.D. is deleted
along with all of the treatments appearing in that particular block from
remaining portion of the original B.I.B.D. The resultant B.I.B.D. will be the
residual B.I.B.D. with parameters (  −k ,  −1 , k , k −  ,  ).
Let us select the Block I consisting of treatments 1, 2 and 4. Then in the
residual design, Block I will be deleted and its treatments 1, 2 and 4 will also
be deleted from the treatments of all the remaining blocks of the given B.I.B.D.
It is seen that due to this process we get the Residual B.I.B.D. as

Block Label Unrandomised Design

I 3 5

II 3 6

III 5 7

IV 5 6

V 6 7

VI 7 3

This is a (4, 6, 3, 2, 1) – B.I.B.D.

14.8.3 Derived Balanced Incomplete Block Design

Let (X, ) be a Symmetric (  , k, ) – B.I.B.D. and let A0 ∈ . Then,
Define
Der(X, , A0 ) = ( A0 , { A ∩ A0 : A ∈ , A  A0 }) 119
 Block 4 Incomplete Block Design

as the Derived Balanced Incomplete Block Design (B.I.B.D.).

Derived design is formed by taking a block A0 and then forming new blocks
such that they consist those points common to the old block and A0 and then
deleting A0 .

It can be proved that the Derived B.I.B.D., Der(X, , A0 ) is a (  − k , 

−1, k , k −  ,  ) – B.I.B.D. provided that k ≥  +2.
Now, you may try to answer the following self-assessment question:

SAQ 5
Define the Residual B.I.B.D. and Derived B.I.B.D. of a given Symmetric
B.I.B.D. Mention the rule of constructing a Residual B.I.B.D. corresponding to
a specific B.I.B.D.

14.9 SUMMARY
On the basis of definitions, theories and concepts related to the study of
“Balanced Incomplete Block Design”, explained and discussed in this unit; a
summary has been presented under this Section which is as follows:
• The terms ‘Complete Block Design’ and “Incomplete Block Design’ have
been explained and their basic difference has been mentioned.
• The idea of ‘Balanced Incomplete Block Design’ (B.I.B.D.) is explained in
detail along with the parameters of it which categorize a B.I.B.D. The
inter-relations of the parameters have been stated and explained.
• The concept of ‘Incidence Matrix’ of a Balanced Incomplete Block
Design (B.I.B.D.) has been explained. Also, the process of finding the
Incidence Matrix of a specific Balanced Incomplete Block Design
(B.I.B.D.) is also explained.
• The concepts of ‘Symmetric Balanced Incomplete Block Design
(B.I.B.D.)’ of a given Balanced Incomplete Block Design (B.I.B.D.),
‘Resolvable Balanced Incomplete Block Design (B.I.B.D.)’ and ‘Affine
Resolvable B.I.B.D.’ have been mentioned in brief.
• The ‘Complimentary Balanced Incomplete Block Design (B.I.B.D.)’ for a
specific symmetric design has been explained. Also, the construction
process of a complimentary design has been explained.
• The ideas of ‘Residual Balanced Incomplete Block Design (B.I.B.D.)’ and
‘Derived Balanced Incomplete Block Design (B.I.B.D.)’ have been
mentioned.

14.10 TERMINAL QUESTIONS

1. Let X = {1, 2, 3, 4, 5} and = { {1, 2, 3,}, {1, 2, 4}, {1, 2, 5}, {1, 3, 4}, {1,
3, 5}, {1, 4, 5}, {2, 3, 4}, {2, 3, 5}, {2, 4, 5}, {3, 4, 5}}; then (X, ) is a (5,
3, 3)- B.I.B.D. Find the Incidence Matrix of this B.I.B.D.
2. The following table depicts an Incomplete Block Design with
b = 8,k = 3,  = 8,r = 3 :
120
Unit 14 Balanced Incomplete Block Design

Block Treatment Labels Block Treatment Labels

I 1 3 8 V 5 7 4

II 2 4 1 VI 6 8 5

III 3 5 2 VII 7 1 6

IV 4 6 3 VIII 8 2 7

State whether it is a Balanced Incomplete Block Design or not. Give the

reason(s) of your answer.

14.11 SOLUTIONS / ANSWERS

Self-Assessment Questions (SAQs)
1. Hint: The answer of the question can be sorted out from the contents
of the Sub-sections 14.2.1 and 14.2.2.
2. Hint: You are advised to go through the materials of the Section 14.3
along with the Sub-sections 14.3.1 and 14.3.2.
3. Hint: You are advised to consult the content of the Section 14.4 for
answering the question.
4. Hint: For the first part of the question, go through the material of the
Section 14.7. For the second part of the problem, consult the
sub-section 14.7.1. However, following the rule mentioned in the
sub-section 14.7.1, you may get the complimentary Balanced
Incomplete Block Design (B.I.B.D.) as follows:
Block Label Design
I 1 3 4 5
II 1 4 6 7
III 1 2 5 7
IV 3 5 6 7
V 2 3 4 7
VI 1 2 3 6
VII 2 4 5 6

5. Hint: For answering the complete question, you are advised to go

through the materials of the Section 14.8, Sub-sections 14.8.1,
14.8.2 and 14.8.3.

Terminal Questions (TQs)

1. Given the (5, 3, 3)-B.I.B.D.
with X = {1, 2, 3, 4, 5} and = { {1, 2, 3,}, {1, 2, 4}, {1, 2, 5}, {1, 3, 4}, {1,
3, 5}, {1, 4, 5}, {2, 3, 4}, {2, 3, 5}, {2, 4, 5}, {3, 4, 5}}; we shall find the
incidence matrix below:
We know that the elements of the Incidence Matrix M are given by
0 if x i  A j
mij = 
 1if x i  A j

121
 Block 4 Incomplete Block Design

We observe that for x1 = 1 there are six sets in containing 1, which

are for j = 1, 2. 3, 4, 5 and 6, therefore, elements m11 ,
m12 ,m13 ,m14 ,m15 ,m16 all will be 1 in the matrix M. Similarly, since for x1 =
1, there are four sets in which do not contain 1, which are for j = 7, 8,
9 and 10; therefore, elements m17 ,m18 ,m19 ,m1,10 will be zero in M. Thus,
we decided the 10 elements in the first row of the matrix M. Following
the same rule for other rows, it is easy to find the remaining elements of
the matrix M. The final Incidence Matrix M is then found as
1 1 1 1 1 1 0 0 0 0
1 1 1 0 0 0 1 1 1 0 

M = 1 0 0 1 1 0 1 1 0 1
 
0 1 0 1 0 1 1 0 1 1
0 0 1 0 1 1 0 1 1 1

2. The given design is not a Balanced Incomplete Block Design, because

of two reasons. The first reason is that treatment label 2, for example,
appears in one block with each of labels 1, 3, 4, 5, 7, and 8 but never
with label 6. Similarly, label 1 never appears with label 5; label 3 never
appears with label 7 and so on. This simple argument shows that no
Balanced Incomplete Block Design can possibly exist for this size of
experiment. Another reason being that for the design given, we have
λ(  − 1) = r (k − 1)

r (k − 1) 6
or λ= =
 −1 7
which is not an integer, whereas for a Balanced Incomplete Block
Design (B.I.B.D.), λ must be an integer.
*******

122