o

Learning Module 7
Week 12

Agricultural
Statistics
(Stat 1e)

CAMIGUIN POLYTECHNIC STATE

COLLEGE
CATARMAN CAMPUS
Institute of Agriculture
Tangaro, Catarman, Camiguin

1
Table of Contents

Learning Module 7
RANDOMIZED COMPLETE
BLOCK DESIGN ..... ii

Description .................... 1

Blocking
……….……..3

Overview
.............. 3

Randomization and Lay-out

................... 5

Requirement, Features and

Assumptions of RCBD ................... 5

Example of Blocking
................... 5

Example Computation ................... 5

FINAL TERM
Submit your outputs on time.
Submission is on the schedule of
module retrieval. (See due date in
google classroom)

0
Learning Outcomes:
At the end of the unit, the students are expected to:
1. Define the terms concerning the basic principles of
experimental design.
2. Discuss the phases of an experimental procedure.
3. Discuss the components of an experimental design.
4. Discern, calculate and layout some experimental designs.

Description
Randomized Complete Block Design (RCBD)
 Is a widely used design
 Is the design which the researcher would use if he could not obtain or
locate completely homogenous experimental units
 eu’s are grouped into r blocks (r-number of replications = number of
blocks) in such a way that each block would contain more or less
homogeneous.
 Each block contains t experimental units (t = is number of treatments)
 The variability among the blocks is taken out of the experimental error
thereby improving the precision of the experiment.
 Blocking which is primarily aimed to reduce experimental error by
eliminating the contribution of known sources of variation among
experimental units, is the distinguishing feature of RCBD
 RCBD is especially appropriate for field experiments where the number
of treatment is not large and the experimental area has a predictable
productivity gradient.
 Distinguishing feature of RCBD:
 Blocks with equal size
 Each block contains all treatments

Blocking
How do we block?

 Gradient is unidirectional

1
 Randomization and Layout

 The experimental units are divided into r blocks considering

homogeneity of the eu’s in each block
 In each block, the t treatment are randomized, making sure that each eu
in a block will have equal opportunity of being assigned each treatment

Decisions:
 If the Fcomp value ≥ than the Ftab at the 1% level of significance, the
effect of the treatments is highly significant, designated by **
 If the Fcomp value ≥ than the Ftab at the 5% level of significance but
<Ftab at 1% level, the effect of the treatments is significant, designated
by *
 If the Fcomp value < than the Ftab at the 5% level of significance, the
effect of the treatments is non-significant, designated by ns

Overview
If the researcher could not find a completely homogenous experimental
units, then he may resort to grouping the experimental units into blocks
(r=replication = no. of blocks) in such a way that the experimental units are more
or less homogenous in a block. This experimental design is called RCBD. The main
purpose of blocking is to reduce the experimental error associated with the non-
homogeneity of the experimental units. Blocking is a form of local control or
experimental error control.

Randomization and Lay-out

The experimental units are grouped into r blocks in such a way that eu’s in each
block are homogenous. The randomization procedure is done by block. This make
sure that each treatment will have an equal chance to be assigned to each of the
homogenous experimental unit in a block. There will be r randomization
procedures, one for each block.

2
Example: Possible Lay-out: RCBD p= 4, r =3

I II III

Block I
T1 - 4
T2 - 2 1 1 1
T3 - 3
T4 - 1

Block II
2 2 2
T1 - 3
T2 - 1
T3 - 4
T4 - 2

3 3 3
Block III
T1 - 2
T2 - 3
T3 - 1
T4 - 4

4 4 4

Experimental units are blocked by size

3
Linear Model RCBD

𝑌𝑖𝑗 = 𝜇 + 𝜏𝑖 + 𝜌𝑗 + 𝜀𝑖𝑗
Where:
𝑌𝑖𝑗 = observation taken in the ith treatment and jth block
𝜇 = average of all possible observations
𝜏𝑖 = effect of the ith treatment
𝜌𝑗 = effect of the jth block
𝜀𝑖𝑗 = experimental error associated with the ith treatment and jth block
i= 1 to p j = 1 to r p = number of treatments r= number of blocks

Assumptions:
1. 𝜀𝑖𝑗 ~NID (0,𝜎𝜀 2 ) – normally distributed with mean zero and variance .
2. Model 1 – The treatments and blocks are fixed and have no interaction
with each other.
∑ 𝜏𝑖 = ∑ 𝜌𝑗 = 0
3. Model 2- The treatments and blocks are random and have no interaction
with each other
𝜏𝑖 ~NID (0,𝜎𝛼 2 ) 𝜏𝑖 ~NID (0,𝜎𝜌2 )

ANOVA table for RCBD

SV df SS MS Fc Ftab5% Ftab1%
Block r-1 BLSS MSBL MSBL/ F[dfblock,ddferror] F[dfblock,ddferror]
MSE
Treat p-1 TrSS MSTr MSTr/ F[dftreat,ddferror] F[dftreat,ddferror]
MSE
Error P(r- ESS MSE
1)
Total Pr-1 TSS

Formulas for CF and SS

CF = (y….)2/pr
TSS = 𝛴𝛴𝑦𝑖𝑗 2 − 𝐶𝐹 using all observations
BLSS = 𝛴𝑦𝑗 2 − 𝐶𝐹 using the block totals
TrSS = 𝛴𝑦𝑖 2 − 𝐶𝐹 using the treatment totals

4
ESS = TSS – BLSS - TrSS by subtraction
MSBL = BLSS/dfblock MS Blocking
MSTr = TrSS/dftreat MS Treatment
MSE = ESS/dferror MS Error
Fcb = MSBL/MSE F – computed for Blocking
Fcb = MSTr/MSE F – computed for Treatment

Coefficient of Variation – measure of precision in the conduct of the experiment.

√𝑀𝑆𝐸
𝐶𝑉 = 𝑥 100%
𝑦̅. .
Standard error of a treatment mean
√𝑀𝑆𝐸
𝑆𝑥 =
𝑟
Standard error of the difference between treatment means
√2𝑀𝑆𝐸
𝑆𝑑 =
𝑟

Notes:

Example. RCBD Linear Model

The gain in weight (kg) of 15 dairy calves, grouped by cake takers, fed with 5
different brands of feeds (T1, T2, T3,T4, T5) are measured. The experiment was
conducted in RCBD. The calves are grouped into three blocks with one care taker
for each block. Define the linear model.

5
Linear Model RCBD

𝑌𝑖𝑗 = 𝜇 + 𝜏𝑖 + 𝜌𝑗 + 𝜀𝑖𝑗
Where:
𝑌𝑖𝑗 = weight gain in kgs of calf given ith brand of feeds under the jth care
taker.
𝜇 = average weight gain
𝜏𝑖 = effect of the ith brand of feeds
𝜌𝑗 = effect of the jth care taker
𝜀𝑖𝑗 = experimental error associated with the ith brands of feeds and jth care
taker
P = 5 r=3 i=1 to 5 j=1 to 3

Randomization procedure in RCBD

Example: Number of treatments = p=4,

Number of blocks = r=3

Steps:
1. Number the experimental unis (eu) 1 to 4 in each block
Block I Block II Block III
eu1 eu1 eu1
eu2 eu2 eu2
eu3 eu3 eu3
eu4 eu4 eu4

2. Prepare four (4) pcs. of paper numbered 1 to 4 corresponding to the

number of experimental units in each block.
3. Begin randomization in Block I,
a. draw one paper and assign eu number to T1 – example 4
b. draw another paper and assign eu number to T2 – example 2
c. draw another paper and assign eu number to T3 – example 3
d. assign the last eu number to T4 – example 1

6
Example result:
Block I Block II Block III
eu1-T4 eu1 eu1
eu2-T2 eu2 eu2
eu3-T3 eu3 eu3
eu4-T1 eu4 eu4

4. Continue randomization of eu’s for blocks 2 and 3

Possible Layout: RCBD p=4, r=3

Block I Block II Block III

eu1-T4 eu1-T2 eu1-T3
eu2-T2 eu2-T4 eu2-
eu3-T3 eu3-T1 eu3
eu4-T1 eu4-T3 eu4

Requirement, Features and Assumptions of

RCBD
Requirement, Features and Assumptions of RCBD
1. Experimental units are grouped into r blocks (r= replications).
2. Experimental units in each block must be homogenous.
3. Each block must have a complete set of treatments.
4. Randomization procedure is done of each block.
5. Useful in field and big experiments.
6. Smaller experimental error due to blocking.
Example of Blocking
Example of Blocking
1. Soil Fertility. In field experiments, like rice or corn experiments, plots are
blocked perpendicular to the slope of the soil or soil fertility gradient.
2. Sunlight. Plots in open field experiments are in North- South orientation,
that means that rows of plants are also North-South orientation so that
experimental units in each block will receive the same amount of sunlight.

7
 3. Care Taker. In animal experiments where it is necessary to employ more
than one care taker then care takers can be considered blocking.
4. Age. In animal experiments, experimental units can be grouped
according to age.
5. Sex. In animal experiments, animals can be grouped by sex, that is
separating the male from female.
6. Location. Different locations can be used as block.

Example Computation
Example Computation Procedure for RCBD
1. The data below are the weight gained (kg) of 15 dairy calves, grouped by
care taker, fed with 5 different brands of feeds (T1, T2, T3, T4, T5). The
experiment was conducted in RCBD with the care takers as blocks.

Block I Block II Block III

T1R1 = 2.9 T1R2 = 5.7 T1R3 = 5.2
T2R1 = 2.5 T2R2 = 3.2 T2R3 = 3.4
T3R1 = 2.8 T3R2 = 2.5 T3R3 = 2.0
T4R1 = 3.2 T4R2 = 3.8 T4R3 = 3.5
T5R1 = 3.2 T5R2 = 5.2 T5R3 = 5.0

Number
Treatment: brands of feeds 5
Blocks: care taker (5 calves per care taker) 3
Experimental unit: calves 15
Response variable: weight gained (kg) 1

Illustrate a possible layout of the experiment

Treatments: T1, T2, T3, T4, T5

8
 Care Taker I Care Taker II Care Taker III
c11 = T5 c12 = T2 c13 = T2
c21 = T2 c22 = T5 c23 = T4
c31 = T1 c32 = T1 c33 = T5
c41 = T4 c42 = T3 c43 = T3
c51 = T3 c52 = T4 c53 = T1

State the null (Ho) and Alternative (Ha) hypothesis

I. Blocking
Ho: The effects of different Care Takers to the gain in
weight of calves are the same.
Ha: The effects of different Care Takers to the gain in
weight of calves are not the same.

II. Treatment
Ho: The effects of different brands of feeds to the gain in
weight of calves are the same.
Ha: The effects of different brands of feeds to gain in
weight of calves are not the same.

Data Presentation in Tabular Form

Treatments Block Totals Means
1 2 3
T1 2.9 5.7 5.2 13.8 4.60
T2 2.5 3.2 3.4 9.1 3.03
T3 2.8 2.5 2.0 7.3 2.43
T4 3.2 3.8 3.5 10.5 3.50
T4 3.2 5.2 5.0 13.4 4.47
Blocks Totals 14.6 20.4 19.1 y.. = 54.1

Blocks Means 2.92 4.08 3.82 = 3.6

9
Degrees of Freedom

Sv df
Block r-1 = 3-1 =2
Treat p-1 = 5-1 =4
Error (p-1) (r-1) = (5-1) (3-1) = 8
Total pr-1 = (5)(3) -1 = 14

Using F-test statistic from F-table

Sv df Ftab 5% FTab1%
Treat 4 Ftab (4,8) = 3.84 Ftab(4,8) = 7.01
Block 2 Ftab (2,8) = 4.46 Ftab(2,8) = 8.65
Error 8

Decision Rule:

Blocking
At , Reject Ho if Fc= MSBL/MSE ≥ 4.46
At , Reject Ho if Fc = MSBL/MSE ≥ 8.65

Treatment
At , Reject Ho if Fc= MSTr/MSE ≥ 3.84
At , Reject Ho if Fc = MSTr/MSE ≥ 7.01

Formulas for CF
CF = (y….)2/pr
CF = (54.1)2 /(5*3)
CF = 2926.81/15
CF = 195.1207

Total Sum of Squares

TSS = Sum of the squares of the observations
TSS =
TSS = (2.92 + 5.72 + 5.22 … + 52) – 195.1207
TSS = 213.29 – 195.1207
TSS = 18.16933

10
Treatment Sum of Squares
TrSS = (Sum of the squares of the treatment totals) divided by r - CF
TrSS =
TrSS = (13.82 +9.12 +7.32 +10.52 +13.42)/3 – 195.1207
TrSS = (616.35)/3 -195.1207
TrSS = 205.45 – 195.1207
TrSS = 10.33

Block Sum of Squares

BLSS = (Sum of the squares of the block totals) divided by p - CF
BLSS =
BLSS = (14.62 + 20.42 + 19.12)/5 – 195.1207
BLSS = (994.13)/5 -195.1207
BLSS = 198.826 – 195.1207
BLSS = 3.71

Error SS
ESS = TSS – TrSS - BLSS
ESS = 18.16933 – 10.32933 – 3.705333
ESS = 4.13

Mean Square Error

MSE = ESS/dferror
MSE = (4.134667)/(8)
MSE = 0.52

Means Square Treatment

MSTr = TrSS/dftreat
MSTr = (10.32933)/ (4)
MSTr = 2.58

Means Square Block

MSBL = BLSS/dfblock
MSBL = (3.705333)/ (2)
MSBL = 1.85
F-computed
11
Blocking
Fcb = MSBL/MSE
Fcb = (1.852667)/(0.516833)
Fcb = 3.58

Treatment
Fct = MSTr/MSE
Fct = (2.582333)/(0.516833)
Fct = 4.99

Summary of ANOVA
SV df SS MS Fc Ftab5% Ftab1%
Block 2 3.71 1.85 3.58ns 4.46 8.65
Treatment 4 10.33 2.58 4.99* 3.84 7.01
Error 8 4.13 0.52
Total 14 18.17

Treatment
Decision: Since (Fc = 4.99) > (Ftab 5% = 3.84) therefore we reject Ho

Conclusion:
The effects of the different brands of feeds to gain in weight of calves are
significantly different.
Or
The gain in weight of calves is significantly different among different brands of
feeds.
Or
There are significant differences on the effects of different brands of feeds on the
gain in weight of calves.
Or
We can say that, with 95% correctness, the gain in weight of calves are differs
with different brands of feeds.

12
Blocking

Decision: Since (Fc = 3.58) is not > (Ftab 5% = 4.46) therefore we fail to reject
Ho.

Conclusion:
The effects of different Care Takers to the gain in weight of calves are
significantly the same.

Coefficient of Variation – measure of precision in the conduct of the experiment.

√𝑀𝑆𝐸
𝐶𝑉 = 𝑥 100%
𝑦̅. .
√0.52
𝐶𝑉 = 𝑥 100%
3.61
𝐶𝑉 = 19.93%

Standard error of a Treatment Means

√𝑀𝑆𝐸
𝑆𝑥 =
𝑟
√0.52
𝑆𝑥 =
3
𝑆𝑥 = 0.42

Standard error of the difference between treatment means

√2𝑀𝑆𝐸
𝑆𝑑 =
𝑟
√2(0.52)
𝑆𝑑 =
3
𝑆𝑑 = 0.59

13
 Assessment:

Problems for RCBD

Instruction: Perform the problem assigned by your professor.

Problem 1.

A researcher investigated the effect of different organic fertilizers on

the yield a certain variety of corn. Sixteen plots were equally used for
4 kinds of organic fertilizer. The following are the yield in tons/ha of an
experiment conducted in RCBD. The plots are blocked based on the
slope of the field. Perform the analysis of variance to find out if there
are significant differences on the different organic fertilizers applied to
corn on its yield.
Blocks
Treatment I II III IV
T1 4.2 3.2 3.7 2.5
T2 4.5 4.2 3.9 5.1
T3 3.9 3.5 3.7 3.5
T4 5.5 4.0 3.8 5.0

Problem 2.
The following data are the weight gained in kgs. of 15 dairy calves,
grouped by cage, fed with 5 different brands of feeds (T1, T2, T3,
T4,T5) mixed with their feeds (kg). The experiment was conducted in
RCBD. The calves are grouped into three cages as blocks.

Blocks
Treatment I II III
T1 2.8 5.8 5.3
T2 2.7 3.6 3.5
T3 2.8 2.5 2.1
T4 3.2 3.8 3.5
T5 3.2 5.2 5

14
 WORKSHEET

Problem No. ____

Number
Treatment: ______________________ ________
Experimental units: _______________ ________
Blocks: ___________________ ________
Response Variable: _______________ ________

Possible Layout of the experiment

State the null (Ho) and alternative (Ha) hypothesis

Blocking:
Ho:
Ha:

Treatment:
Ho:
Ha:

Data Presentation in Tabular Form

Treatments Blocks Totals Means
1 2 3

Block Total
Block
Means

Degrees of Freedom (CRD)

Sv df
Block
Treat

15
 Error
Total

Using F-test statistic from F-table

Sv df Ftab 5% FTab1%
Block
Treat
Error

Decision Rule:

Blocking
At 𝛼 = 5%, Reject Ho if Fc=
At 𝛼 = 1%, Reject Ho if Fc =

Treatment
At 𝛼 = 5%, Reject Ho if Fc=
At 𝛼 = 1%, Reject Ho if Fc

Grand Total
y.. =
y.. =
y.. =

Formulas for CF
CF =
CF =
CF =
CF =

Total Sum of Squares

TSS =
TSS =
TSS =
TSS =
16
TSS =

Treatment Sum of Squares

TrSS =
TrSS =
TrSS =
TrSS =
TrSS =
TrSS =

Block Sum of Squares

BLSS =
BLSS =
BLSS =

Error SS
ESS =
ESS =
ESS =

Treatment Means Square

MSTr =
MSTr =
MSTr =

Error Mean Square

MSE =
MSE =
MSE =

F-computed
Fc =
Fc =
Fc =

17
ANOVA table for CRD
SV df SS MS Fc Ftab5% Ftab1%
Block
Treatment
Error
Total

Decision:

Conclusion:

Coefficient of Variation
𝐶𝑉 =
𝐶𝑉 =
𝐶𝑉 =

Standard error of a Treatment Means

𝑆𝑥 =
𝑆𝑥 =
𝑆𝑥 =

Standard error of the difference between treatment means

𝑆𝑑 =
𝑆𝑑 =
𝑆𝑑 =

References:

CMU Statistics Manual, 2014

APEC Agricultural and Technical Cooperation Working Group. (2013).
Agricultural Statistics Best Practice Methodology handbook
DAVIS, BOB.(2000). Introduction to Agricultural Statistics. Delmar Cengage
Learning; 1st Edition.
18
IDAIKKADAR M. N. (2001). Agricultural Statistics. A handbook for
Developing Countries. 1st Edition
RANGASWAMY.R. (2009). Agricultural Statistics. New Age
International Publisher. 8122425925

Prepared by:

JESSA D. PABILLORE
jessapabillore916@[Link]
09179869017

19