3. Definition of experimental design
Experimental design is
the branch of statistics
that deals with design
and analysis of
experiment.
The experiment design is
the laying out of a
detailed experimental
plan in advance of doing
the experiment that will
maximize results
Experimental design is
a way to carefully plan
experiment in advance
so that your result are
both objective and
valid.
4. Applications of experimental design
Agriculture
Medicine
Biology
Industrial production
The methods of experimental design are widely used in following fields
5. History of experimental design
James Lind performed wee organize experiment in 1747
and developed a cure for scurvy.
His experiment miss only randomization.
Ronald fisher present methodology for designing
experiment in his books
His work fall under the first era(1918-1940) of design of
experiment.
The Arrangement of field
experiments(1926)
The design of experiment
(1935).
6. Why we need experimental design
From a statisticians perspective experiment is performed to decide:
Weather there is actual difference among treatments and it is not due to
difference in experimental units.
Secondly wot is the size of difference among treatments
But there is not always such ideal conditions that there is no variation among
experimental units.
So due to difference among experimental units, variation between treatments is
caused other than actual difference need to be tested.
7. In simple words We need experimental design to control variability caused by
other factors.
So that the treatment effect can be identified prominently and accurately.
Technically experimental design are for reduction of experimental error
(The difference among experimental plots treated alike is called experimental
error
If experimental error is more , it mean the basic variation among treatment is
due to chance.
If experimental error is less it mean there is differ
8. Example
Suppose there are two experimental plots x&y.
A rice breeder has to do experiment to check the yield of two new varieties(A & B) he
made.
He sow variety A in X plot while variety B in Y plot.
The results he got from this simple experiment are sure not be accurate.
This is because along with the variety difference(treatment difference) , there are many
other differences in plots like soil condition, moisture, sunlight etc.
Even if he plant same variety A in plot x & Y he will get different results(experimental
error)
Breeder will follow principles of experimental design and design a suitable experiment
for proper evaluation.
10. Replication
Reptation of basic experiment is called replication.
Helps in getting more accurate estimate of experiment error.
More replication means more precision.
Replication reduce standard error.
Example:
Rice breeder, planted his variety A and B in more than
one experimental units.
The more number of time he plant any variety in
experimental unit the more replication it would be.
Variety A Variety B
Replication
1
Plot 1 Plot 5
Replication
2
Plot 2 Plot 6
Replication
3
Plot 3 Plot 7
Replication
4
Plot 4 Plot 8
n=r= No of replications
One Plots= one experimental unit
11. Randomization
It is process of assigning the treatment to experimental units
Each treatment has same probability to get a experimental unit.
If treatments 3 (A,B,C) and replications 4 than no of experimental units = 12 .
A C B C
C B A B
A C B A
12. Example
Treatment Treatment
Replication
1
Plot 1(A) Plot 5(B)
Replication
2
Plot 2(A) Plot 6(B)
Replication
3
Plot 3(A) Plot 7(B)
Replication
4
Plot 4(A) Plot 8(B)
Rice breeder having two treatments(variety A & B).
Number of replications =4
Number of experimental units(plots)=8
The non-random layout is:
There is chance for fertility gradient or any other
biasness from one direction to other.
Breeder follows any randomization method and
have following layout.
B A
A B
A B
B A
13. Methods to do randomization
BY Random number table
By Drawing lots
By Software(GEN STAT, Excel)
By Cards
14. By Random number table
Random
number
Sequence Rank
204 1 3
279 2 4
711 3 6
100 4 1
197 5 2
619 6 5
889 7 8
Assign the plot number from 1 -8.
Open random number table and move finger any where from
left to right with eyes close.
Stop any where and select a three-digit number and move
vertically to took 8 numbers.
Write sequence of random numbers and rank from smallest to
largest.
Plot no Variety Sequence
1 A 1
2 A 2
3 A 3
4 A 4
5 B 5
6 B 6
7 B 7
8 B 8
15. Now assign the 8 treatments to 8 plots by seeing sequence corresponding to
rank.
E.g: The treatment A with sequence 1 will assigned to plot 3 because its rank
is 3
Plot 1(A) Plot 5(B)
Plot 2(B) Plot 6(A)
Plot 3 (A) Plot 7(B)
Plot 4(A) Plot 8(B)
Remember: Rank correspond to new plot number
16. Local control
Choose a design in such a manner that all extraneous sources of variation are
brought under control
Balancing
Blocking
Proper plot techniques
17. Blocking
Arrangement of experimental units into blocks of similar entities is called as blocking.
Blocking means that the like experimental units should be collected together to far
relatively homogeneous groups.
Observations collected under similar experimental conditions are grouped in the same
block.
Blocking reduces sources of deviations and thereby bringing more precision to the
experimental design.
18. Balancing
Balancing means that the treatment should be assigned to the experimental
units in such a way that the result is a balanced arrangement of treatment.
19. Proper plot technique
Plot with homogenous soil
Equal slope
Equal block size and shape
Equal number of replications
Removal of border plants
Thinning
Furrowing for row spacing
Fertilizer application
Labeling
Pesticide application