DESIGN OF EXPERIMENT:
Designing experiment means deciding how the observations or measurements should be taken to answer a particular problem under investigation in a valid efficient and economic way.
DIFFERENT EXPERIMENTAL DESIGNS:
There are different experimental designs suitable for different purpose.
1. Completely Randomized Design.
2. Randomized Completely Block Design
3. Factorial Experiment / Split Plot Design
4. Latin Square Design
PRINCIPLES OF EXPERIMENTAL DESIGN:
There are three principles of experimental design
1. Replication
2. Randomization
3. Error control / Local control
1. Replication:
In case of experiment, there is necessity for replicating the entire set of treatment in the experimental unit. Each of this set is called replication.
Purpose of replication:
1. Replication provides the greater precision of the experimental result by reducing the standard deviation of the treatment mean.
2. To provide an estimation of experimental error.
3. To increase the scope of the inference of the experiment.
Though replication improves the experiment but one cannot take too many replication because of limitation of space, labor, money, time and availability of an experimental material.
Thus a simple thumb rule for determination of replication of an experiment is
(Treatment d. f.) X (Replication d. f.) = 20
Replication d. f. = 20
Treatment d. f.
Replication – 1 = 20
Treatment – 1
Replication = 20 + 1
Treatment – 1
For example, let treatment 5. Then,
Replication = 20 + 1 = 6
(5-1)
Replication will be less when the treatment is more
Replication will be more when the treatment is less
2. Randomization:
Randomization is the arrangement of the treatment to experimental unit so that all units considered to have an equal chance to receive the treatment assign.
Purpose of randomization:
1. To avoid biasness for arrangement of treatments in the experimental units.
2. To ensure the unbiased examination of treatment mean an experimental error.
Randomization is done by using a table of numbers. For example,
Random Number Table
Number
|
Serial no.
|
Ranking order
|
684
|
1
|
7
|
793
|
2
|
8
|
416
|
3
|
4
|
250
|
4
|
1
|
287
|
5
|
2
|
896
|
6
|
9
|
375
|
7
|
3
|
504
|
8
|
5
|
560
|
9
|
6
|
3. Local Control:
Reduction of experimental error blocking / grouping of experimental unit which eliminates the extra genus sources of variation from treatment comparison. For this purpose experimental units are divided into a set of homogenous groups or blocks. Blocking is done in such a way that the variation among the block is as large as possible and units that within the block is as small as possible.
The distance between the blocks should be as far as possible and the distance between the units in a block should be as less as possible.
No comments:
Post a Comment