The document discusses key concepts in experimental design including: 1. Randomization, replication, and local control are essential principles to generate valid estimates of error variance and make valid statistical tests. 2. Experimental design involves understanding the research question to determine appropriate treatment levels and the relationship between treatments and responses. 3. It is important to identify sources of variation beyond the treatment being studied, such as genetic and environmental factors, and employ local control techniques. 4. Examples are provided for components of experimental design like hypotheses, independent and dependent variables, experimental units, blocks, and sources of variation.

1. DESIGN OF
EXPERIMENTS

2. Have you ever got
confused when
trying to design an
experiment?

3. Do you struggle
to analyze your
experimental
results?

4. “... you can observe a lot
just by watching.”
- Yogi Berra
Great New York Yankees Catcher
Manipulate the input variables and
observe changes in the system output
EXPERIMENTS

5. Experiment
test in which purposeful changes are made to the
input variables of a process or system so that we
may observe and identify the reasons for changes
Experimental Design
Planning an experiment to obtain appropriate data
and drawing inference out of the data with respect to
any problem under investigation

6. COMPONENTS OF EXPERIMENTAL DESIGN
1. Any problem under investigation
• Research question, hypothesis
2. Plan an experiment
• Sample size, Treatment, Response
3. To obtain appropriate data
• Results and Observation
4. To draw inference out of the data
• Statistical Analysis
• Conclusion

7. COMPONENT OF EXPERIMENTAL DESIGN
Example
Problem under Investigation:
Find out the effects of fertilizer dose on crop yield
Ho: There is a significante effect of fertilizer dose on crop yield
Plan an Experiment:
Define doses to be checked (2, 3, or more)
Grow crop plants receiving different (defined) doses of fertilizer
Define statistical analysis to be done

8. COMPONENT OF EXPERIMENTAL DESIGN
Example
To obtain appropriate data:
Crop yield under different doses of fertilizer
To draw inference out of the data:
If different doses of fertilizer affect crop yield
Statistical analysis to interpret data
Reject or accept null hypothesis

9. OBSERVATIONAL STUDY VS. EXPERIMENT
• Treatment – deliberately apply on
a group of objects or subjects
• Purpose – to observe the
response
Observational Study
• Collection and analysis of data
altering existing
without
conditions
Experiment

10. EXPERIMENTAL DESIGN
Testable Hypothesis
• A hypothesis that can be tested
 Crop yield is affected by fertilizer dose
Independent Variable
• One or more variables – precisely manipulated
 Amount of fertilizer/fertilizer dose – 3 doses can
be given
Dependent Variable
• One dependent variable – precisely measured
 Crop yield

11. EXPERIMENTAL DESIGN
Experimental Unit
• Which receives a treatment
• A leaf, a tree, animals, or an individual
• Sometimes referred to as a plot
 single crop plant receiving specific fertilizer dose
 If 100 crops plants in experiment, it is 100
experimental units or plots

12. EXPERIMENTAL DESIGN
Block
• Collection of plots/experimental units
• Also known as samples or groups
 A group of crop plants receiving similar dose of
fertilizer
 Multiple samples (2,3, or more) in an experiment

13. EXPERIMENTAL DESIGN
Sources of Variation
• Observations vary considerably – among
experimental units - variation
 Different plants will grow at different rates and
crop yield of the samples / blocks will vary
• Sources of Variation:
1. Treatment
2. Extraneous factors / Experimental Error

14. EXPERIMENTAL DESIGN
Treatments
• One source of variation
• Done by manipulation of a certain
interest known as treatment /
variable of
predictor /
independent / factor variable.
• Variation introduced by the experimenter to see the
influence / effect.
 Variable of interest – fertilizer dose
 Treatments / categories of variable – different
doses defined by experimenter

15. EXPERIMENTAL DESIGN
Extraneous Factors
• Also known as confounding variable
• Largely uncontrolled:
environmental factors,
genetic variation,
observer’s fatigue, and
instrument defects.
• Unavoidable – called as “experimental error”
• “Variation produced by set of known factors beyond
the control of experimenter”
 Genetic make-up of plants
 Environmental factors affecting plant growth

16. PRINCIPLE OF EXPERIMENTATION
• Almost all experiments involve three factors
1. Randomization
2. Replication
3. Local Control
• These three are in a way complementary to each
other
 To increase the accuracy of experiment
 To provide a valid test of significance

17. PRINCIPLE OF EXPERIMENTATION
Randomization
• Every experimental unit will have the same chance
of receiving any treatment
• Limit the occurence of bias
• Minimize effect of extraneous
factor
 100 units
Every unit has equal chance
of receiving any of the treatments

18. PRINCIPLE OF EXPERIMENTATION
Randomization:
Simple Random Sampling
• assign number of
participants
• choose randomly and assign
to treatment groups.
• simple to implement

19. PRINCIPLE OF EXPERIMENTATION
Randomization:
Permuted Block Randomization
• balance participants into
groups or blocks
• equally distribute treatment
to each blocks

20. PRINCIPLE OF EXPERIMENTATION
Randomization: Stratified Random Sampling
• target popultion to split up into strata (subsets)
• random sampling to choose elements with
each stratum

21. PRINCIPLE OF EXPERIMENTATION
Replication
• Repetition of experiment
under identical conditions
• Number of experimental
units under the same
treatment
• Greater number of
replicates – greater is the
precision of the
experiment
 100 plants or units
 5 treatments or doses of
fertilizers
 20 replicates of each
treatment

22. PRINCIPLE OF EXPERIMENTATION
Local Control
• Control of all factors (unwanted) except the ones
about which we are investigating
• Reduces variations due to extraneous factors
• Increases the precision of experiment
 Control genetic variation – get plant seeds from
same source
 Environmental factors - homogenous
environmental conditions
• Only effective source of variation - treatments

23. PRINCIPLE OF EXPERIMENTATION
Local Control
• Important to include a control or placebo group
 A sample or group of plants receiving no fertilizer
NO
TREA
TMENT
TREATMENT A TREATMENT B TREATMENT B
CONTROLLED GENETIC VARIATION
CONTROLLED ENVIRONMENTAL CONDITION
TREATMENT D

24. SUMMARY
• Randomization, Replication and Local
Control
• All of these – essential for making:
• Valid estimate of error (within) variance
• Valid test of significance
• Increase the accuracy and precision

25. SUMMARY
• Understanding research question and experimental
conditions
• Extremely important
• Clear understanding of research question
 Helps in deciding levels of treatment variable
 Relationship between treatment and response variable
• Through knowledge of the confounding factors and
field conditions
 Identifying sources of variation other than treatment
 Precise local control
 Enough replicates for statistical analysis

26. Practice (10 minutes)
Think of an experimental design title based on your field of
specialization i.e. The Effect of Speed of Aerator to the Level of
Dissolved Oxygen in Aquarium.
Determine the following:
Hypothesis: Dissolve oxygen is affected by speed of aerator
Independent Variable: Speed of aerator
Dependent Variable: Amount of dissolved oxygen
Confounding Variable: Temperature
Principle of Experimentation: Local Control (Room Temperature)

27. Thank you for listening