INTRODUCTION

DEFINITION
 Experiment:
An experiment is a planned inquiry to discover
new facts, or to confirm or deny the results of previous
investigation
 Treatment:
A treatment is a procedure whose effect on the
experimental material is to be measured
 Factor:
A particular class of related treatments is often called
factor
 Levels of the factor:
The states of a factor, i.e., the treatments
within the class, are called levels of the factor
 Experimental unit:
Experimental unit is the piece of
experimental material to which one trial of a single experiment is
applied
 Sampling unit:
A sampling unit is that fraction of the
experimental unit on which the effect of the treatment is
measured
 Block:
A group of homogenous experimental units is called a
block
 Experimental design:
An experimental design is a set of rules
by which the treatments to be used in an experiment are assigned
to the experimental units
 Randomization:
If treatments are assigned to a set of unit in
such a way that every unit is equally likely to receive any
treatment, the assignment is set to be random and the phenomena
is called randomization
 Replication:
When a treatments appears more than once in an
experiment, the treatment said to be replicated and phenomena is
called replication
 Experimental error:
Experimental error is the variation among
experimental units which have been treated alike
Variation comes from two main sources:
1. Firstly there is the inherent variability that exist in the
experimental material to which treatments are applied.
2. Second, there is the variation which result from any lack in
uniformity in the physical conduct of the experiment
Experimental error can be reduced in the following ways:
• Blocking
• Increase in number of replication
• Proper use of efficient statistical design
• Size and shape of experimental units
 Design of an experiment:
It is the planning of an experiment
in such a way that the information that would be collected is
relevant to the problem under investigation.
• Therefore, design of an experiment is a complete sequence of
steps taken ahead of time to ensure that the appropriate data will
be obtained in a way that allows an objective analysis leading to
valid inferences with regard to the set hypotheses
 Statistical Design:
• A set of statistical rules for allocating the given number of
experimental units/plots to the given number of treatments.
• Functions of a Statistical Design:
• sort out treatment differences/variation from
variation/differences among experimental units/plots; say to
separate wheat from chaff.
• To provide unbiased estimates of treatment differences
• Measure uncertainty in the estimates of treatment differences
in term of standard deviation(SD), standard error of
estimate(SE) etc.
 Principles of a statistical design:
1) Replication of treatments
2) Randomization of treatments
3) Local control of variation among experimental units
1. Replication:
• It is the number of distinct non-overlapping experimental units
assigned to “a” treatments. No experiment should be conducted
without complying with the basic principle. This is the most
important ingredient of an experiment.
• Replication is must as it helps to:
1) Estimate experimental error
2) Give precise estimates of treatment effects as any effects due to
uncontrolled factors go into experimental error and only treatment
effects remain in the estimates of the treatment effects
 Cont.….
The number of replications in an experiment depends upon:
1. Resources available
2. magnitude of experimental error
3. Size of treatment differences which is biologically important
4. A small difference to be detected requires more replications
than a large difference
NOTE:
Use as many number of replication for which error degree of
freedom is not less than 8 and number of replication for which
error degree of freedom greater than 20 is wastage of resources
 Cont.….
2. Randomization:
• The allocation of treatment to the experimental units in such a
way that each unit has an equal chance of receiving any
treatment is called randomization
• Randomization is necessary in order to eliminate any personal
bias in the allocation of treatments to the experimental units
• Every treatment gets the same chance to go to any
experimental unit/plot
Why randomize?
• To validate statistical analysis
• To protect against bias in estimates of treatment differences
 Cont.….
3. Local Control:
• It refers to amount of balancing, blocking and grouping of
experimental units that is employed in the adopted statistical
design
• The purpose of local control is to make the experimental
design more efficient
• It makes any test of significance more sensitive or test
procedure more powerful
• This increase in efficiency or sensitivity or power occurs
because a proper use of local control will reduce the
magnitude of the estimate of experimental error
 Completely Randomized Design
• Completely Randomized Design (CRD) is a design in which the
treatments are assigned to the experimental units completely at
random. That is the randomization is done without any
restriction. The design is completely flexible, i.e., any number of
treatments and replication may be used. Moreover, the number
of replication for any treatment needs not to be equal. A
completely randomized design is considered to be most useful in
situations where:
• The experimental units/materials is homogeneous
• The experiments are small such as laboratory experiments
• Some experimental units are likely to be destroyed or fail to
respond.
• Advantages of CRD
• The design is completely flexible, i.e., any number of
treatments and of replications may be used. Moreover, the
number of replication for any treatment needs not be equal.
• It gives maximum degrees of freedom for error sum of
squares as compared with the other designs for the same
situation.
• The design is very simple and is easily laid out
• The statistical analysis is very simple both for equal and
unequal number of replication.
• If the data from some experimental units is missing, it does
not complicate the analysis. The missing observations can be
discarded without affecting the results of the experiment and
efficiency of the design is still not affected
 Disadvantages of CRD
• The design is applicable only to a small number of
treatments.
• It is only applicable if the experimental material is
homogeneous
• If the experimental units are not homogeneous, then
the design gives large experimental error.
• There is possibility of entering the whole of the
variation among the experimental units into the
experimental error, as the randomization is not
restricted in any direction.
•
 Example: Consider a data of sunflower head diameter
(cm) of 7 genotypes grown in 3 replications.

Varieties R1 R2 R3 Total

V1 15.81 16.12 15.65 47.58

V2 18.93 18.22 18.36 55.51

V3 15.58 15.65 16.20 47.43

V4 15.60 15.20 15.15 45.95

V5 16.70 16.10 17.00 49.80

V6 15.10 14.95 15.45 45.50

V7 18.45 18.10 18.25 54.80

Total 116.17 114.34 116.06 -

Grand Total 346.57

 Randomized complete block design
• The completely randomized design is used when the experimental
material is homogeneous, but in situations when there is relatively large
variability in the experimental material and it is possible to make blocks in
such a manner that the experimental units within a particular blocks are
relatively homogeneous. The design applied in such situation is
randomized complete block design.
• The treatment are assigned at random to the experimental units within
each block, which means the randomization is restricted within blocks.
The main objective of the blocking is to reduce the variability among
experimental units within a block as much as possible and to maximize the
variation among blocks. Randomized complete block designs are
commonly used in agriculture experiments when the fertility and other
soil factors show variability across field. The blocks are made
perpendicular to the variation present in the experimental material. If
there are no differences among blocks, then RCBD would not contribute to
improve precision in detecting differences among treatment means. The
word “Complete” here indicates that each block contains the complete set
of treatments.
 Advantages of randomized complete block design
• The design is flexible, i.e., any number of treatment and of replication (but
not less than 2) may be used.
• The statistical analysis is fairly simple when there are no missing
observations.
• It is easy to adjust for missing observations.
• Grouping the experimental material controls the source of extraneous
variation and hence the estimate of the experimental error is decreased.
• When the variability among the blocks is large then precision increases
because the sum of squares for block is extracted from the sum of squares
for experimental error leading to smaller mean square error (MSE).
• Placing blocks under different conditions increases the scope of the
experiment.
• It provides unbiased estimates of the means of the blocking factor.
• A Randomized Complete Block Design is generally more efficient than
Complete Randomized Design subject to proper block orientation
 Disadvantages of randomized complete block design
• It controls variability only in one direction.
• It is not a suitable design when the number of treatment is
very large or when the blocks are not homogeneous.
• With the increase in block size, within block variability
increases which is not the aim of this design because our
purpose is to reduce variability within the blocks and to
maximize the among block variability.
• Missing observations in the data cause problem in the
analysis. One or two missing observations can be handled
easily but more missing observation affect the efficiency of
the design.
• The degree of freedom for error is less for Randomized
Complete Block Design because of the formation of blocks as
compared with the completely randomized design. This loss
can be dangerous when blocks effect is non-significant