STAT ASSIENMENT

STAT-511

ANSHUMAN PAUL [DEPT – AGRONOMY]

1ST Year 2nd semester

241221523

ASSIENMENT-1

Multiple Choice Questions (MCQs)

1. In a Completely Randomized Design (CRD), the experimental units are:

A) Grouped into blocks
B) Randomly assigned to treatments without blocking
C) Assigned based on row and column effects
D) Grouped by treatment levels

Answer: B) Randomly assigned to treatments without blocking

2. Which of the following designs helps control two sources of variability?

A) CRD
B) RBD
C) RCBD
D) Latin Square Design

Answer: D) Latin Square Design

3. The primary purpose of blocking in RCBD is to:

A) Increase randomness
B) Decrease treatment effects
C) Control variability among blocks
D) Increase the number of replications

Answer: C) Control variability among blocks

4. Which of the following is NOT an assumption of CRD?

A) Homogeneity of variance
B) Normal distribution of errors
C) Independent errors
D) Presence of row and column effects

Answer: D) Presence of row and column effects

5. In Latin Square Design, each treatment appears:

A) Only once in each row and each column
B) Randomly in all plots
C) Twice in each block
D) Only in diagonal cells

Answer: A) Only once in each row and each column

Short Answer Question-

Q: What is the main difference between CRD and RCBD?

Answer:
The main difference is that in CRD, treatments are assigned randomly to all experimental units
without any blocking, while in RCBD, experimental units are grouped into blocks (based on some
source of variability), and treatments are randomly assigned within each block to control the
variability.

Q: What is a statistical experiment?

Answer: A statistical experiment is a planned procedure for collecting data under controlled
conditions, where the outcomes are uncertain and can be analyzed using probability and statistical
methods.

Q Why is local control considered a refinement over replication in experimental design?

A: Replication reduces random error by averaging results over repeated plots, but it cannot remove
systematic variation caused by field heterogeneity. Local control improves precision further by
grouping similar experimental units into blocks or arranging them systematically, thus removing
variation due to known factors (e.g., slope, fertility gradient) before analysis. This makes treatment
comparisons more accurate, especially in non-uniform fields.

Q: What statistical test is used in CRD to compare treatment means?

A: The F-test is used in the Analysis Of Variance (ANOVA) to determine whether the differences
among treatment means are statistically Significant. If the F-value is greater than the critical value,
we reject the Null hypothesis of equal means.

Long Answer Question-

Q: Explain the key features, advantages, and limitations of Latin Square Design. Also, compare it
briefly with RCBD.
Answer:
Latin Square Design is an experimental design used to control two sources of nuisance variability
(usually rows and columns) in addition to the treatments. It is arranged in a square format where
each treatment appears only once in each row and each column. This helps in reducing experimental
error from two directions.

Key Features:

 Controls for two blocking factors.

  Each treatment appears once per row and column.

 Requires equal number of treatments, rows, and columns.

Advantages:

 Effective for controlling two sources of variability.

 Reduces experimental error, increasing the precision of treatment comparisons.

 Requires fewer experimental units compared to a factorial design that controls for two
blocking factors.

Limitations:

 The number of treatments must equal the number of rows and columns, which may limit
flexibility.

 Difficult to arrange when the number of treatments is large.

 Missing data can significantly impact the analysis.

Comparison with RCBD:

 RCBD controls variability in only one direction (blocking factor), while Latin Square Design
controls variability in two directions.

 RCBD is simpler to implement and more flexible in terms of number of treatments.

 Latin Square is more efficient when both row and column effects need to be controlled, but
less practical for large treatment numbers.

Q: Explain the difference between a statistical experiment and a statistical model, with an
example.
Answer:
A statistical experiment is the actual process of collecting data by observing or manipulating
variables under controlled or natural conditions. Its goal is to gather information about a
population or process.
A statistical model, on the other hand, is a mathematical representation of the relationships
between variables in the collected data. It uses equations, probability distributions, or
assumptions to explain and predict outcomes.

Example:
Suppose we want to study the effect of fertilizer type on crop yield.

 The statistical experiment could involve dividing a field into plots, applying different
fertilizers, and measuring yields.

 The statistical model might be a regression equation Yield=β0+β1(Fertilizer Type)+ϵ\

text{Yield} = \beta_0 + \beta_1(\text{Fertilizer Type}) + \epsilonYield=β0+β1(Fertilizer Type)
+ϵ that estimates how yield changes with fertilizer type, including an error term for random
variation.
ASSIENMENT-2

ANOVA for an agronomic (RCBD) experiment :

Objective: test whether four fertilizer treatments (A, B, C, D) give different mean crop yields (t/ha).
Design: RCBD with 3 blocks (locations/replicates); each block contains all 4 treatments (one plot per
treatment per block).

Observed yields (t/ha):

Block A B C D Total

B1 5.20 6.10 5.80 6.40 23.5

B2 5.50 6.30 6.00 6.70 24.5

B3 5.30 6.00 5.90 6.60 23.8

Total 16 18.4 17.7 19.7 71.8

Mean 5.33 6.133 5.9 6.56

Grand mean = 5.9833 t/ha Correction Factor=429.60

How the numbers were computed (formulas)

Total Sum of Squares (TSS)

Measures total variation in the data

TSS=∑Yij2−N(∑Yij)2

=2.53667

Treatment Sum of Squares (TrSS)

Measures variation due to differences among treatment means.

TrSS=ri=1∑t(Yˉi.−Yˉ..)2

=2.37667

Replication Sum of Squares (RSS)

Measures variation due to differences among blocks/replications.

RSS=tj=1∑r(Yˉj −Yˉ..)2

=0.131667

Error Sum of Squares (ESS)

Measures unexplained variation after accounting for treatments and blocks.

ESS = TSS – TrSS − RSS

=0.02833

ANOVA table (RCBD) — computed from the data above

Source DF Sum of Squares Mean Square (MS F [Link] F table value

(SS) = SS/DF) (MS / MSE)

Treatment 3 2.3767 0.7922 167.7647 4.75706266

Block 2 0.1317 0.0658 13.9412 5.14325285

Error 6 0.0283 0.00472 —

Total 11 2.5367 — —

Key results:

Fcal> Ftab

So we reject the null hypothesis

4) Post-hoc: Fisher’s (pairwise comparisons):-

CD value= t(0.05,def) x √ 2 EMS /r

=0.137259905

 dfE=6, MSE=0.004722MSE = 0.004722MSE=0.004722, n=3n=3n=3 (replicates per treatment),

t0.025,6≈2.4469t_{0.025,6}\approx 2.4469t0.025,6≈2.4469.

 CD≈ 0.1373 t/ha.

treatment Mean difference significance

A-B 0.833 significant
A-C 0.57 Significant
A-D 1.23 Significant
B-C 0.233 Significant
B-D 0.427 Significant
C-D 0.66 significant
Interpretation: any pairwise treatment mean difference > 0.1373 is statistically significant at the 5%
level. Example: mean(D) − mean(A) = 6.5667 − 5.3333 = 1.2334 > LSD, so D differs from A.