MAHARAJA AGRASEN COLLEGE
PRACTICAL FILE
2372573501
Introduction
To
Design of experiments
Chaitanya Chauhan
23/252
[Link]. Mathema5cal Sciences
Index
1. Analysis of CRD, RBD & LSD.
2. Analysis of 22 and 23 factorial experiments in CRD and RBD.
3. Analysis of a completely confounded two level factorial designs in
2 and 4 blocks.
4. Analysis of a partially confounded two level factorial design.
5. Analysis of a single replicate of a 2n design.
6. Analysis of one-half and one-quarter fractions of 2n factorial
design.
Practical – 1
Analysis of CRD
Completely Randomized Design (CRD)
Q. A researcher wants to compare six different treatments (T1 to T6) to
study their effect on a particular response variable.
Each treatment is replicated 3 times and the responses obtained are given
below:
Solution :-
Conclusion :
ANOVA Decision
The ANOVA results show that:
• F calculated = 62.628
• F critical = 3.105
• P-value = 3.5 × 10⁻⁸
Decision Rule
• If F calculated > F critical, we reject H₀.
• If P-value < 0.05, we also reject H₀.
Conclusion
Since
• 62.628 is much greater than 3.105, and
• the P-value is extremely small (less than 0.05),
We reject the null hypothesis (H₀).
This means that all treatment means are not equal.
Practical – 2
Analysis of RBD
Randomized Block Design (RBD)
Q . A researcher wants to compare six treatments (T1 to T6) while
controlling variability due to three blocks (R1, R2 and R3). The observed
responses are recorded in the following table where columns represent
blocks and rows represent treatments.
Solution :
Conclusion :
For Treatments (Rows):
• F calculated = 60.2523
• F critical = 3.325835
• P-value = 3.85E-07
Decision
• Since F calculated > F critical and P-value < 0.05 → Reject H₀ (for
treatments)
Conclusion
There is a significant difference among treatment means. At
least one treatment mean is different.
Practical – 3
LSD TEST QUESTION
Q. A researcher compares 3 fertilizer treatments (T1, T2, T3) on crop yield.
Each treatment is replicated 4 times and the yields (in kg) are recorded.
Tasks:
1. Perform ANOVA (CRD).
2. Compute LSD at α = 0.05.
3. Identify which treatment pairs differ significantly.
Data are provided in the 'Data' sheet.
Solution :
Practical – 4
2² Factorial Experiment in Completely Randomized Design (CRD)
Question :- A researcher conducts a 2² factorial experiment under a Completely
Randomized Design (CRD) to study the effects of two factors:
• Factor A at two levels: –1 (Low) and +1 (High)
• Factor B at two levels: –1 (Low) and +1 (High)
The experiment also includes the interaction AB.
Each treatment combination is replicated once (i.e., total 4 treatment combinations, 8
observations).
The coded values of A, B, the interaction AB, and the observed response YYY are
shown below:
SoluEon :
Practical – 5
2² Factorial Experiment Under Randomized Block Design (RBD)
Q. A researcher wants to study the effect of two factors:
• Factor A at two levels: –1 (Low) and +1 (High)
• Factor B at two levels: –1 (Low) and +1 (High)
A 2² factorial experiment is conducted using a Randomized Block Design (RBD)
with 2 blocks.
To incorporate blocks into regression, a dummy variable is used:
• Block 1 → Dummy = 0
• Block 2 → Dummy = 1
Each treatment combination appears once in each block, resulting in 8 total
observations.
The coded factors, block dummy variable, and observed response values YYY are
shown below:
Solution :
Practical – 6
2³ Factorial Experiment in Completely Randomized Design (CRD)
Question . A researcher conducts a 2³ factorial experiment under a Completely
Randomized Design (CRD) to study the effects of:
• Factor A (–1, +1)
• Factor B (–1, +1)
• Factor C (–1, +1)
The factorial design includes all 8 treatment combinations, each
replicated 1 time, giving 12 total observations.
The coded factor levels and responses are shown below:
Solution :
Practical – 7
2³ Factorial Experiment in Randomized Block Design (RBD)
Question . A researcher conducts a 2³ factorial experiment arranged in a Randomized Block Design
(RBD) with 3 blocks.
The factors studied are:
• Factor A (2 levels: –1, +1) • Factor B (2 levels: –1, +1)
• Factor C (2 levels: –1, +1)
There are 8 treatment combinations and 12 total observations because 4 extra observations were
taken to complete the blocks.
The data (coded levels and responses) are shown below:
SOLUTION :
Practical – 8
Completely Confounded 2³ Factorial in 2 & 4 Blocks
A researcher conducts an experiment using a 2³ factorial design, but due to
restrictions on experimental size, the experiment is arranged in 2 blocks of size 4
using a complete confounding scheme.
The three factors studied are:
• A (–1, +1)
• B (–1, +1)
• C (–1, +1)
The interaction ABC is confounded with blocks, and a block indicator variable D1 is
used:
• Block 1 → D1 = 0
• Block 2 → D1 = 1
The coded design matrix and observed responses are given below:
Solution :
Practical – 9
Completely-Confounded 2³ Factorial Design (with Blocks)
Q. A researcher runs a 2^3 experiment but, due to prac9cal constraints, uses a
confounding/blocking scheme so that some effects are confounded with blocks. The
experiment has 8 runs (observa9ons). The coded factor levels and responses are
shown in the table below (–1 = low, +1 = high).
Solution: