AE 58

Experimental Design for Agricultural

Engineering
Modules and Laboratory Exercises

0
 TABLE OF CONTENTS

I. Introducton To Experimental Design 2-5

II. Single Factor Experiment 6 - 20

A. Complete Block Design

1. CRD
2. RCBD
3. LSD

B. Incomplete Block Design

III. Two – Factorial Experiments 21 - 30

A. Complete Block Design
B. Split - Plot Design
C. Strip - Plot Design

IV. Comparison Between Treatment Means 31 - 41

A. Pair Comparison
B. LSD
C. DMRT
D. TUKEY HSD Test

V. Regression And Correlation Analysis 42 – 45

VI. Introduction to SPSS 46 - 47

VII. Laboratory Exercises 48 - 78

1
MODULE – I. Introduction to Design of Experiments
1.1 Four Eras in the History of DOE
Here's a quick timeline:

 The agricultural origins, 1918 – 1940s

o R. A. Fisher & his co-workers
o Profound impact on agricultural science
o Factorial designs, ANOVA
 The first industrial era, 1951 – late 1970s
o Box & Wilson, response surfaces
o Applications in the chemical & process industries
 The second industrial era, late 1970s – 1990
o Quality improvement initiatives in many companies
o Taguchi and robust parameter design, process robustness
 The modern era, beginning circa 1990, when economic competitiveness and globalization
is driving all sectors of the economy to be more competitive.

1.2 Engineering Experiments

If we had infinite budgets there probably wouldn't be a big fuss made over designing
experiments. In production and quality control we want to control the error and learn as much as
we can about the process or the underlying theory. From an engineering perspective we're trying
to do use experimentation for the following purposes:

 reduce time to design/develop new products & processes

 improve performance of existing processes
 improve reliability and performance of products
 achieve product & process robustness
 perform evaluation of materials, design alternatives, setting component & system
tolerances, etc.

1.2. a Most Common Purpose of Conducting Experiment

 To explore new technologies, new crops, and new areas of production
 To develop a basic understanding of the factors that control production
 To develop new technologies that are superior to existing technologies
 To study the effect of changes in the factors of production and to identify optimal
levels
 To demonstrate new knowledge to growers and get feedback from end – users about
the acceptability of new technologies

1.3 Steps for Planning, Conducting and Analyzing an Experiment

2
 The practical steps needed for planning and conducting an experiment include: recognizing
the goal of the experiment, choice of factors, choice of response, choice of the design, analysis
and then drawing conclusions. This pretty much covers the steps involved in the scientific
method.

1. Recognition of & statement of problem

2. Choice of factors, levels, and ranges
3. Selection of the response variable(s)
4. Choice of design
5. Conducting the experiment
6. Statistical analysis
7. Drawing conclusions, and making recommendations

1.4 Experimental Design/Designed Experiment

- It is a plan for the assignment of the treatments to the plots in the
experiment
- Treatments are imposed by investigator using standard protocols
- Designs differ primarily in the way the plots are grouped before
the treatments are applied
- May infer that the response was due the treatments

1.5 Designing a Research Experiments

Designing an experiment simply means planning an experiment so that information will
be collected is relevant to the problem under investigation. All too often data collected are of
little or no value in an attempted solution to the problem because little or no prior
consideration was given to the design of the experiment. Thus, the design of experiment is
the complete sequence of the steps taken ahead of time to ensure that the appropriate data
will be obtained in a way that permits an objective analysis leading to valid with respect to
the stated problem.

1.6 Two Key Considerations in Designing An Experiment;

1. Simplicity – choose the simplest experimental design among many possible
candidates to achieve the same proposed objective(s).

2. Efficiency – conduct the investigation as efficient as possible, that is, every effort
should be made to save time, money, personal and experimental materials. Fortunately,
most simple design is also efficient both statistically and economically.

Three Basic Principles in Achieving Optimal level of Simplicity and Efficiency;

1. Replication – repetition of treatments in an experiment.

2. Randomization – this process involves random allocation of treatments to the
experimental units. Thus, the process makes the law of chance applicable to

3
 our experimental data and ensures that the data are free from any systematical
error.
3. Blocking (Local Control) – refers to grouping of the experimental units in
such a way that the units within the group are more homogenous than are units
in different groups.

"All experiments are designed experiments, it is just that some are poorly designed and a some
are well-designed."

The Well – Planned Experiment

- Simplicity
- Degree of precision
- Absence of systematic error
- Range of validity of conclusion
- Calculation of degree of uncertainty

1.7 Two Basic Structures of Experimental Design;

1. Treatment Structures – consist of the set of treatments, treatment
combinations or set of treatments( one – way treatment structure) or a set of
treatment combinations(two – way factorial arrangement or high – order
factorial arrangement) plus any control or other standard treatments.
2. Design Structure – consists of the grouping of the experimental units into
homogenous groups or blocks (CRD, RCBD,LSD and etc.)

1.8 The Scientific Method for Conducting Experiment;

A. Formulation of Hypothesis
B. Planning an experiment to objectively test the hypothesis
 Once the hypothesis is framed, the next step is to design a procedure
for its verification. This experimental procedure, which usually
consists of four (4) Phases are as follows;
1. Selecting the appropriate materials to test
2. Specifying the characters to measure
3. Selecting the procedure to measure those characters
4. Specifying the procedure to determine whether the measurement made
support the hypothesis

 For 3 & 4, the procedures regarding how the measurements are to be

made and how the measurements can be used to prove or disprove a
hypothesis depend heavily on techniques developed by statisticians.
- These Two Tasks constitute much of what is generally
termed as the Designed of an Experiment, which has Three
Essential Components as follows;
1. Estimate of Error

4
 a. Replication
b. Randomization
2. Control of Error
- The ability to detect the existing differences among
treatments increases as the size of the experimental error
decreases.
Three Techniques of Controlling Error
a. Blocking
b. Proper Plot Technique
c. Data Analysis
3. Proper Interpretation of Results

1.8. a More Detailed Steps in Experimentation

1. Definition of Problems
2. Statement of the Objectives
3. Selection of Treatments
4. Selection of experimental materials
5. Selection of experimental design
6. Selection of the unit for observation and the number of replications
7. Control of the effects of the adjacent units on each other
8. Consideration of data to be collected
9. Outlining statistical analysis and summarization of results
10. Conducting the experiment
11. Analyzing data and interpreting results
12. Preparation of a complete, readable, and correct report of the research

1.9 Terminology
Experiment – planned inquiry
Treatment – procedure whose effect will be measured
Factor – class of related treatments
Levels – states of a factor
Variable – measurable characteristic of a plot
Experimental Unit – a unit which a treatment is applied(plot)
Sampling Unit – part of experimental unit that is measured
Experimental Error – variation among experimental units that are treated alike

5
MODULE – II. Single – Factor Experiments
Experiment in which only a single factor varies while all others are keep constant.

2.1 Type of Field Experiments

(Oriented toward applied research)
1. Variety Trials
2. Experiments with perennial crops
3. Pasture Experiments
4. Intercropping Experiments
5. Farming Systems Research

2.2 Two (2) Groups of Experimental Design that are applicable to a Single –
Factor Experiments:
1. Complete Block Design – small number of treatment characterized by blocks, each of
which contains at least one complete set of treatments.
Three (3) Complete Block Design
1. Completely Randomized Design(CRD)
2. Randomized Complete Block Design(RCBD)
3. Latin Square Design(LSD)

2. Incomplete Block Design – large number of treatments, characterized by blocks, each

of which contains only a fraction of the treatments to be tested.

Two (2) Incomplete Block Design

1. Lattice Design(LD)
2. Group Balanced Block Design(GBBD)

2.3 Randomization Scheme:

1. By Table of Random Numbers( Appendix A)
Steps
1. Locate a starting point
2. From starting point read downward vertically up to n (n = total number
observation)
3. Rank the n random numbers
4. Divide the n ranks into t groups
5. Assign the t treatments to the n experimental plots

2. By Drawing Cards
Steps
1. Draw n cards
2. Rank n cards(2 = lowest, A = highest)
3. Assign the t treatments to the n plots

3. By Drawing Lot

6
2.4 Completely Randomized Design (CRD)
- Simplest and least restrictive
- Every plot is equally likely to be assigned to any treatment
- Appropriate for experiments with homogenous units, such as laboratory
experiments

2.4.1 Advantages of a CRD

1. Flexibility
- Any number of treatments and any number of replications
- Don’t have to have the same number of replications per treatment
2. Simple Statistical Analysis
3. Missing plots do not complicate the analysis
4. Maximum error degrees of freedom

2.4.2 Disadvantage of a CRD

1. Low precision if the plots are not uniform

2.4.3 Uses for the CRD

1. If the experimental site is relatively uniform

2. If a large fraction of the plots may not respond or may be lost
3. If the number of plots are limited

2.4.4 Randomization and Lay out of a CRD

Steps;
1. Determine n (n = rt), where r = replications, t = treatments
2. Use any of the randomization scheme
3. Assign plot number to each experimental plot
4. Assign the treatments to the experimental plots

2.4.5 Analysis of Variance of a CRD

Two (2) Sources of Variation

1. Treatments
2. Experimental Error

2.4.6 For CRD Equal Replication

Steps
1. Group the data by treatments and calculate the treatment totals (T) and grand total (G)
2. Construct an outline of the ANOVA

7
 Outline ANOVA with Corresponding Formula

Source of Degree of Sum of Mean Tabular F

Computed F
Variation Freedom Squares Squares 5% 1%

Treatment t–1 T 2

 CF
TrSS
TrMS
ErMS
r t -1
ErSS
Experimental ToSS – TrSS
t(r – 1) t (r-1)
Error

Total r(t) – 1 ΣX2 – C.F

3. Correction Factor(C.F) = G2
n
4. Obtain the tabular F values from Appendix E
f’1 = treatment df (t -1) (numerator df)
f’2 = error df (t(r -1)) (denominator df)
5. Enter all the values computed in step 2 using the formula in the outline ANOVA
6. Compare the computed F values with the Tabular F values

Where:
TrSS = Treatment Sum of Squares r = number of replications
ToSS = Total Sum of Squares t = number of treatments
TrMS = Treatment Mean Square T = treatment total
ErMS = Error Mean Square X = Variables (measurement of the ith plot)

2.4.7 Rules in Deciding the Significance of the Difference Among Treatments;

1. If the computed F value is larger than the tabular F value at the 1% level of
significance, the treatment difference is said to be highly significant such as a
result is generally indicated by placing two asterisks on the computed F value
in the ANOVA.
2. If the computed F value is larger than the tabular F value at the 5% level of
significant but smaller than or equal to the F value at the 1% level of
significance, the treatment is said to be significant, indicated by placing one
asterisk on the computed F value in the ANOVA.
3. If the computed F value is smaller than or equal to the tabular F value at the
5% level of significance, the treatment difference is said to be nonsignificant
and indicated by placing ns on the computed F value in the ANOVA.

8
2.4.8 Numerical Examples
Table 1. Plant height result for use of different treatments from a CRD experiment with
5 replications and 5 treatments.

Treatment Treatment
Treatment Plant Height (cm)
Total (T) Mean
A 8.2 8.7 9.4 9.2 9.0 44.5 8.9
B 7.7 8.4 8.6 8.0 8.1 40.8 8.2
C 6.9 5.8 7.2 6.8 7.4 34.1 6.8
D 6.8 7.3 6.3 6.9 7.1 34.4 6.9
Control 5.3 6.3 6.7 7.1 5.6 31.0 6.2
Grand Total
184.8
(G)
Grand
7.4
Mean

Table 1.a. Analysis of Variance of height result in Table 1.

Source of Degree of Sum of Mean Tabular F 5%

Computed F
Variation Freedom Squares Squares 1%
21**
Treatment 4 24.37 6.09 2.87 4.43
Experimental 0.29
20 5.71
Error
Total 24 30.08
b** = highly significant
cv = 7.28%

Table 2. Growth data recorded for the Plants that survived in a CRD experiments with 4
replications and 5 treatments.
Treatment Treatment
Treatment Plant Height (cm)
Total (T) Mean
A 8.2 8.7 9.2 9.0 35.1 8.78
B 7.7 8.4 8.0 8.1 32.2 8.05
C 6.9 5.8 6.8 7.4 26.9 6.73
D 6.8 6.9 7.1 20.8 6.93
Control 5.3 6.3 7.1 5.6 24.3 6.08
Grand Total
139.3
(G)
Grand Mean 7.31

9
2.4.9 for CRD with Unequal Replication

The same procedures with CRD equal replication, except that of the following;
t
T2
 TrSS =   CF
i 1 ri

 Todf = n -1 (n = total no. of remaining observations)

 Erdf = Todf - Trdf

Table 2a. Analysis of Variance of Growth data recorded in Table 2.

Source of Degree of Sum of Mean Tabular F
Computed F
Variation Freedom Squares Squares 5% 1%
15.57**
Treatment 4 18.70 4.68 3.11 5.03
Experimental 0.29
14 4.09
Error
Total 18 22.79
b** = highly significant
cv = 7.36 %

Note:
The cv indicates the degree of precision with which the treatments are compared and is good
index of reliability of the experiment.
- For rice 6 to 8 % for variety trials, 10 to 12 % for fertilizer trials, 13 to 15 % for herbicide and
insecticide trials.

10
2.5 Randomized Complete Block Design
- One of the most widely used experimental designs in Agricultural research.
- Suited for field experiments where the number of treatment is not large.
- Area has predictable productivity gradient.
- Blocks of equal size each contain all the treatments.

2.5.1 Factors for Arriving an Appropriate and Effective Blocking Technique

1. The selection of the source of variability to be used as the basis for blocking.

2.The selection of the block shape and orientation.

 Ideal source of variation to use as the basis for blocking is one that is large and
highly predictable.

2.5.2 Guidelines for the Shape and Size of Blocks

1. Unidirectional Gradient
 Long and narrow blocks
 Block is perpendicular to the direction of gradients.

2. Fertility gradient occurs in two direction with one direction much stronger than the
other, ignore the weaker gradient and follow the preceding guideline for the case of
the unidirectional gradient.

3. Fertility gradient occurs in two direction with both gradients equally strong and
perpendicular to each other.

Alternatives:
 Square blocks
 Long and narrow blocks with their length perpendicular to the direction of one
gradient use Covariance.
 Use Latin Square Design with two – way blockings, one for each gradient.

4. Pattern of variability is not predictable.

 Use square blocks

2.5.3 Randomization
Applied separately and independently to each of the blocks.

2.5.4 Analysis of Variance (ANOVA)

Steps:

1. Group the data by treatments and replications and calculate treatment totals (T),
replication totals (R), and grand total (G).

12
 2. Outline the analysis of Variance.

Outline ANOVA
Source of Degree of Means Tabular F
Sum of Squares Computed F
Variation Freedom Squares 5% 1%

Replication r–1 R 2

 CF
RSS
t r-1

T 2

 CF TSS TrMS f1 = Trdf

Treatment t –1 r t-1 ErMS f2 = Erdf

ErMS
Error (r -1)(t -1) ToSS –RSS - TrSS (r  1)(t  1)

ΣX2 – CF
Total rt - 1

G2
3. Compute C.F. =
rt
4. Compute for the degree of freedom, Sum of squares, Mean Squares and F – value.
Follow the formula in outline ANOVA.
5. Enter all values computed in the ANOVA.
6. Compare the computed F – value with the Tabular F – value.
7. Compute for cv.

Where:
TrSS = Treatment Sum of Squares r = number of replications
ToSS = Total Sum of Squares t = number of treatments
TrMS = Treatment Mean Square T = treatment total
ErMS = Error Mean Square X = Variables (measurement of the ith plot)
RSS = Replication Sum of Squares R = Replication Total

2.5.5 Block Efficiency

1. Determine the level of significance of the replication variation.

F( Replication ) = RMS ∕ ErMS

2. Relative Efficiency

13
 RE = ( r – 1 ) Eb + r( t – 1 ) Ee
( rt – 1 ) Ee

Where:
Eb = replication mean square
Ee = error mean square
Note:
If the error df is less than 20, the RE value should be multiplied by adjustment factor k
defined as:

k
(r  1)(t  1)  1t (r  1)  3
(r  1)(t  1)  3t (r  1)  1

2.5.6 Numerical Examples

Table 2.6 Data on Stem and Root Rot of Cucumber under Trichoderma inoculation
with Different Levels of Inoculumns (%).
Replication
Trichoderma Level Mean
Total
(GM/16L H2O) 1 2 3

Control(H2O only) 24.17 21.67 15.00 60.84

20.28
T1(10g) 19.17 25.00 26.67 70.84
23.61
T2(20g)
22.50 25.00 20.83 68.33 22.78
T3(30g)
20.00 19.17 23.33 62.50 20.83
T4(40g)
14.17 21.67 20.83 56.67 18.89

Replication Total 100.00 112.51 106.66

Grand Total 21.28

319.18
Grand Mean

Table 2.6a Analysis of Variance on Stem and Root Rot of Cucumber under
Trichoderma inoculation with Different Levels of Inoculumns (%).

14
 Source of Degree of Sum of Mean of Computed Tabular F
Variation Freedom Squares Squares F 5% 1%
Replication 2 15.64 7.82
Treatment 4 43.78 10.95 0.78ns 3.84 7.01
Error 8 112.59 14.07
Total 14 172.01
ns = not significant
cv = 18 %

2.6 Latin Square Design

- More restrictive than the RCBD
- The total number of plots is the square of the number of treatments
- Each treatment appears once and only once in each row and column
- If you can block on two (perpendicular) sources of variation (rows x columns) you can
reduce experimental error when compared to the RCBD

2.6.1 Advantages and Disadvantages

Advantage:
– Allows the experimenter to control two sources of variation

Disadvantages:
– Error df is small if there are only a few treatments
– The experiment becomes very large if the number of treatments is large
– The statistical analysis is complicated by missing plots and misassigned
treatments

2.6.2 Useful in Animal Nutrition Studies

 Suppose you had four feeds you wanted to test on dairy cows. The feeds would be tested
over time during the lactation period
 This experiment would require 4 animals (think of these as the rows)
 There would be 4 feeding periods at even intervals during the lactation period beginning
early in lactation (these would be the columns)
 The treatments would be the four feeds. Each animal receives each treatment one time
only.
2.6.3 Uses in Field Experiments

 When two sources of variation must be controlled

– Slope and fertility
– Furrow irrigation and shading
– If you must plant your plots perpendicular to a linear gradient

15
  Practically speaking, use only when you have more than four but fewer than ten
treatments
– a minimum of 12 df for error

2.6.4 Randomization and Lay out

Randomize row and column treatments using the LS plan in Appendix K

Steps :
1. Select a LS plan from Appendix K
2. Randomize the row arrangement of the plan
3. Rank from lowest to highest
4. Rank as existing row numbers and sequence as row number of the new plan.
5. Randomize the column arrangement

2.6.5 Analysis

1. Set up a two-way table and compute the row and column means and deviations
2. Compute a table of treatment means and deviations
3. Set up an ANOVA table divided into sources of variation
– Rows
– Columns
– Treatments
– Error
4. Significance tests
– FT tests difference among treatment means
– FR and FC test if row and column groupings are effect

The LSD ANOVA

Source of Degree of Sum of Tabular F

Mean squares Computed F
Variation Freedom Squares 5% 1%
ΣR2 – CF
Row t-1 RSS/(t-1) RMS/ErMS
t
ΣC2 – CF
Column t-1 CSS/(t-1) CMS/ErMS
t
ΣT2 - CF
Treatment t-1 t TrSS/(t-1) TrMS/ErMS

Error (t-1)(t-2) ToSS - SASS ErSS/(t-1)(t-2)

ΣX2 - CF
Total t2-1

Note : SASS = Sum of All other Sum of Squares ( SASS = RSS + CSS + TrSS )

16
CSS = Column Sum of Squares
RSS = Row Sum of Squares

2.6.6 Relative Efficiency

 To compare with RCBD using columns as blocks

RMS  (t  1) ErMS
RE 
t ( ErMS )
Where:

RMS = Row Mean Square

CMS = Column Mean Square
ErMS = Error Mean Square

 To compare with RCBD using rows as blocks

RE 
CMS  (t  1) ErMS
k
(t  1)(t  2)  1(t  1) 2  3
t ( ErMS ) (t  1)(t  2)  3(t  1) 2  1
 To compare with CRD

RMS  CMS  (t  1) ErMS

RE 
(t  1)( ErMS )

2.6.7 Numerical Examples

1. To determine the effect of four different sources of seed inoculums, A, B, C, and D, and a
control, E, on the dry matter yield of irrigated alfalfa in pounds per plot. The plots were furrow
irrigated and there was a line of trees that might form a shading gradient.

A B D C E
33.8 33.7 30.4 32.7 24.4
D E B A C
37.0 28.8 33.5 34.6 33.4
C D A E B
35.8 35.6 36.9 26.7 35.1
E A C B D
33.2 37.1 37.4 38.1 34.1
B C E D A
34.8 39.1 32.7 37.4 36.4

Table 1.0 Dry Matter Yield of irrigated alfalfa from LSD Experiment

17
 Row Row Total
Yield, lb/plot
Number
(R)
A B D C E
1 155
33.8 33.7 30.4 32.7 24.4
D E B A C
2 167.3
37 28.8 33.5 34.6 33.4
C D A E B
3 170.1
35.8 35.6 36.9 26.7 35.1
E A C B D
4 179.9
33.2 37.1 37.4 38.1 34.1
B C E D A
5 180.4
34.8 39.1 32.7 37.4 36.4
Column
174.6 174.3 170.9 169.5 163.4
Total(C)
Grand Total
(G) 852.7

Table 1.1a Treatment Totals and Treatment Means

Treatment Total(T) Treatment Mean
A 178.8 35.76
B 175.2 35.04
C 178.4 35.68
D 174.5 34.9
E 145.8 29.16
Grand Total(G) 852.7

Grand Mean (Gm) 34.11

Table 1.1 ANOVA of Dry Matter Yield

18
 Source of Degree of Sum of Tabular F
Mean squares Computed F
Variation Freedom Squares 5% 1%
21.85
Row 4 87.40 7.12** 3.26 5.41

Column 4 16.56 4.14 1.35 ns 3.26 5.41

155.89
Treatment 4 38.97 12.69** 3.26 5.41

Error 12 36.80 3.07

296.66
Total 24
ns
b** = highly significant, = nonsignificant
cv = 5.14%

2.7 Incomplete Block Design

- An alternative set of designs for single – factor having a large number of treatments.
- It is enhanced by an increased variability in the experimental material.
- Data Analysis for an incomplete block design is more complex than that for a complete
block design.

2.7.1 Balanced Lattice Design

2.7.1a Basic Features of BLD

1. The number of treatments (t) must be a perfect square( t = K 2 ).
2. The block size (k) is equal to the square root of the number of treatments
( k = t ).
3. The number of replications ( r) is one more than the block size ( r = k +1 )

2.7.1b Randomization and Lay out

Steps:
1. Divide the experimental into replication(r).
2. Divide each replication into k incomplete blocks, each containing k
experimental plots.
3. Select from Appendix L a Basic Balanced Lattice plan corresponding to the
number of treatments to be tested.
4. Randomize the replication arrangement of the selected plan.
5. Randomize the incomplete blocks within each replication.
6. Randomize the treatment arrangement within each incomplete block.
7. Apply the final outcome of the randomization process of step 6 to the field lay
out.

19
 2.7.1.c Analysis of Variance

Steps:
1. Calculate the block totals (B) and replication(R ).
2. Calculate the treatment totals (T) and the grand total (G).
3. Calculate Bt for each treatments.
4. Calculate W = kT – (k+1)Bt + G
5. Construct the Outline ANOVA
G2
6. CF  2
(k )(k  1)
7. Compute for the Sum of Squares and Mean Squares as indicated in the
ANOVA
Note:
Computations for Mean Squares are equal to its corresponding Sum of Squares divided by its
degree of freedom except for Treatment Adjusted MS and Effective Error MS.

ANOVA of BLD with Corresponding Formula

Source of Degree of Sum of Tabular F
Mean Square Computed F
Variation Freedom Squares 5% 1%
Replication K  R2  CF
k2
Treatment 2
k -1
 T 2  CF
(unadj) (k  1)
t

Block (adj) 2 W 2

k -1 i 1

(k 3 )(k  1)
ToSS – RSS –
Intrablock Error 2
(k – 1)(k -1) TrunadjSS –
BSS
 1  2 G 
2
TradjMS
  T '  2 
2
Treatment(adj) k -1 ------------- 
 (k  1)(k  1)  
2
k  EfERMS
2
Effective Error (k – 1)(k -1) -------------- IntraMS(1+kµ)
Total 2
k (k+ 1) – 1  x 2  CF

20
MODULE – III. Two – Factor Experiment

3.1 Three Points to Consider For Interaction Effect

1. Two factors can be measured only if the two factors are tested together in the
same experiment.
2. When interaction is absent – simple effect is same for all levels of other factors
and equals the main effect.
3. When interaction is present the simple effect of factor changes as the level of
other factor changes.

 Factorial Experiment – treatments consist of all possible combinations in

selected levels in two or more factor.
 Complete Factorial Experiment – treatments include all combinations of the
selected levels of the variable factors.

3.2 Advantages and Disadvantages of a Factorial Experiment

Advantages – If the factors are independent
- Results can be described in terms of the main effects
- Hidden Replication – the other factors become replications of the main
effect
Disadvantages
- As the number of factors increase, the experiment becomes very large.
- Can be difficult to interpret when there are interactions.

3.3 Uses of Factorial Experiments

1. When you are charting new ground and you want to discover which factors are
important and which are not
2. When you want to study the relationship among a number of factors
3. When you want to be able to make recommendations over a wide range of conditions

3.4 How to set up a Factorial Experiment

1. The Field Plan
- Choose an appropriate experimental design
* Experimental Designs for Factorial Experiment
1. Any of the CBD (CRD, RCBD,& LSD)
2. Split – Plot Design
3. Strip – Plot Design
- Make sure treatments include combinations of all factors at levels
- Set up randomization appropriate to the chosen design

2. Data Analysis

21
 - Construct tables of means and deviation
- Complete an ANOVA table
- Perform significance tests
- Compute appropriate means and standard errors
- Interpret the analysis and report the results

3.5 RCBD Factorial

The procedures for randomization and lay out for RCB single factor experiments is
directly applicable to RCB factorial experiments by simply ignoring the factor composition of
the factorial treatments and considering all the treatments as if they were unrelated. For the
ANOVA, the computations discussed for RCB single factor is the same for RCB factorial except
that in the RCB factorial there is an additional computational steps are required to partition the
treatment sum of squares into factorial components corresponding to the main effects of
individual factors and to their interactions.

3.5.1 ANOVA for RCBD Factorial

Tabular F
Source Of Degree Of Sum of
Mean Squares Computed F 5% 1%
Variation Freedom Squares
ΣR2 - C.F RSS
Replication (r – 1)
ab (r – 1)
Σ T2 - C.F TrSS TrMS
Treatment (t -1)
r (t – 1) ErMS
Σ A2 - C.F ASS AMS
Factor (A) (a – 1)
rb (a – 1) ErMS
Σ B2 - C.F BSS BMS
Factor (B) (b – 1)
ra (b 1) ErMS
TrSS – ASS – BSS
(AXB)SS
AXB (a-1)(b-1)
(a – 1)(b -1)
ToSS – RSS – TrSS
ErSS
Error (r-1)(ab -1)
(r -1)(ab-1)
Total (rab – 1) ΣX2 – C.F

22
3.5.1a Numerical Example

Table 3.0 Grain Yield of Three Rice Varieties Tested with Five Fertilizer in a RCB Design.
Grain Yield, Treatment
Kg/ha Total
(T)
Fertilizer Rep I Rep II Rep III Rep IV
V1
F1 3.85 2.61 3.14 2.89 12.50
F2 4.79 4.94 4.56 4.61 18.89
F3 4.58 4.45 4.88 3.92 17.84
F4 6.03 5.28 5.91 5.65 22.87
F5 5.84 5.92 5.98 5.52 23.29
V2
F1 2.85 3.79 4.11 3.44 14.19
F2 4.96 5.13 4.15 4.99 19.22
F3 5.93 5.70 5.81 4.31 21.74
F4 5.66 5.36 6.46 5.47 22.96
F5 5.46 5.55 5.79 5.93 22.72
V3
F1 4.19 3.75 3.74 3.43 15.11
F2 5.25 4.58 4.90 4.29 19.01
F3 5.82 4.85 5.68 4.93 21.28
F4 5.89 5.52 6.04 4.76 22.21
F5 5.86 6.26 6.06 5.36 23.55
Rep Total(R) 76.99 73.69 77.20 69.51
Grand Total ( G) 297.39

Construct the factor A and factor B two – way Table of Totals

Table 3.1 The Variety X Fertilizer Table of Totals from Data in Table 3.3
Yield Total
Fertilizer Total (
(AB)
B)
Fertilizer V1 V2 V3
F1 12.50 14.19 15.11 41.80
F2 18.89 19.22 19.01 57.13
F3 17.84 21.74 21.28 60.86
F4 22.87 22.96 22.21 68.04
F5 23.29 22.72 23.55 69.56
Variety Total (A) 95.39 100.83 101.16 297.39

23
Table 3.2 Analysis of Variance of Data in Table 3.3 from a 3 x 5 Factorial Experiment in
RCBD
Source of Degree of Sum of Tabular F
Mean Squares Computed F
Variation Freedom Squares 5% 1%
Replication 3 2.60 0.87 5.74** 2.83 4.29
Treatment 14 44.49 3.18 21.09** 1.94 2.54
Variety(A) (2) 0.96 0.48 3.48* 3.22 5.15
Fertilizer(B) (4) 41.25 10.31 68.26** 2.59 3.80
AXB (8) 2.28 0.29 1.89ns 2.17 2.96
Error 42 6.10 0.15
Total 59 53.19
b** = highly significant, * = significant, ns =nonsignificant
cv = 7.8%

3.6 SPLIT – PLOT DESIGN

- It is a design which allows the levels of one factor to be applied to large plots
while the levels of another factor are applied to small plots
- Large plots are whole plots or main plots
- Smaller plots are split plots or subplot
- Suited for a two – factor experiment that has more treatments than can be
Accommodated by CBD

3.6.1 Two (2) Factors of Split – Plot Design

1. Main – plot Factor
2. Subplot Factor

3.6.2 Randomization of a Split –Plot Design

- Levels of the whole – plot factor are randomly assigned to the main plots, using
different randomization for each block
- Levels of the subplots are randomly assigned within each main plot using a
Separate randomization for each main plot

3.6.3 Guidelines for Assignment of a particular factor to either the main plot or subplot:

1. Degree of precision
2. Relative size of the main effects
3. Management Practices

3.6.4 Advantages and Disadvantages of Split – Plot Design

Disadvantages
1. Main plot factor is estimated with less precision so larger differences are required for
significance – may be difficult to obtain adequate df for the main plot factor
2. Statistical analysis is more complex because different standard errors are required for
different comparisons

24
Advantages
1. Permits the efficient use of some factors that require different sizes of plot for their
application
2. Permits the introduction of new treatments into an experiment that is already in
progress

3.6.5 Data Analysis

- Estimate and test the appropriate main effects and interactions
- Analysis proceeds as follows;
1. Construct tables of means
2. Complete an analysis of variance
3. Perform significance tests
4. Compute means and standard errors
5. Interpret the analysis

3.6.6 Analysis of Variance for Split – Plot Design

Tabular F
Source of Degree of Sum of
Mean Squares Computed F 5% 1%
Variation Freedom Squares
ΣR2 - C.F RSS RMS
Replication (r – 1)
ab (r -1) Er(a)MS
Main-Plot ΣA2 - C.F ASS AMS
(a -1)
Factor(A) rb (a – 1 ) Er(a)MS
Σ(RA)2/b - CF - Er(a)SS
Error(a) (r- 1)(a – 1) RSS - ASS (r -1)(a – 1 )
Subplot ( b- 1) ΣB2 - C.F BSS BMS
Factor(B) ra (b -1) Er(b)MS
Σ(AB)2 /r - C.F – (AXB)SS (AXB)MS
AXB (a – 1)(b-1) BSS – ASS (a – 1)( b -1) Er(b)MS
Er(b)SS a(
Error(b) a(r-1)(b-1) ToSS - SASS
r – 1)(b-1)
Total (rab – 1) ΣX2 – C.F

Note : SASS = Sum of All other Sum of Squares ( SASS = RSS + ASS + Er(a)SS + BSS + ABSS )

25
Table A. Grain Yield Data of Four Rice Varieties Grown with 4 levels of Nitrogen in Split –
Plot Design with Three(3) replications

Grain Yield, t/ha

Variety
Rep I Rep II Rep III
N0(0 kg/ha)
V1 4.4 5.0 4.0
V2 4.0 5.3 3.6
V3 3.5 3.0 3.1
V4 4.0 4.5 4.8
N1(60 kg/ha)
V1 5.4 5.2 6.4
V2 6.5 6.0 5.6
V3 4.7 6.0 5.5
V4 5.2 4.6 4.6
N2( 90 kg/ha )
V1 6.0 6.4 6.7
V2 6.0 6.1 6.6
V3 6.2 5.7 6.0
V4 4.5 5.7 4.2
N3(120 kg/ha)
V1 6.4 7.1 6.7
V2 7.1 7.0 6.5
V3 6.0 5.9 6.4
V4 2.7 5.0 3.6

Table A.1 The Replication X Nitrogen Table of Yield Totals Computed from data in Table A

Yield Total (RA) Nitrogen

Nitrogen
Total(A)
Rep I Rep II Rep III
N0 15.9 17.8 15.5 49.2
N1 21.8 21.8 22.1 65.7
N2 22.7 23.9 23.5 70.1
N3 22.2 25.0 23.2 70.4
Rep Total (R ) 82.6 88.5 84.3
Grand Total
255.4
(G)

26
Table A.2 The Nitrogen X Variety Table of Yield Totals Computed from data in Table A

Yield Total (AB)

Nitrogen
V1 V2 V3 V4
N0 13.4 12.9 9.6 13.3
N1 17.0 18.1 16.2 14.4
N2 19.1 18.7 17.9 14.4
N3 20.2 20.6 18.3 11.3
Variety Total (B) 69.7 70.3 62.0 53.4

Table A.3 Analysis of Variance of data in Table A from 4X4 Factorial Experiment in Split Plot
Design
Source of Degree of Sum of Tabular F
Mean Squares Computed F
Variation Freedom Squares 5% 1%
Replication 2 1.15 0.575
Nitrogen (A) 3 25.00 8.33 59.5** 4.76 9.78
Error(a) 6 0.82 0.14
Variety (B) 3 15.71 5.24 15.41** 3.01 4.72
AXB 9 13.24 1.47 4.32** 2.30 3.25
Error(b) 24 8.05 0.34
Total 47 63.96
b** = highly significant
cv(a) = 7.09%
cv(b) = 10.96%

3.7 STRIP – PLOT DESIGN

- Sometimes called split – block design
- For experiments involving factors that are difficult to apply to small plots
- Three sizes of plots so there are three experimental errors
- The interaction is measured with greater precision than the main effects

3.7.1 Three (3) Plot Size

1. Vertical – Strip Plot - Vertical Factor
2. Horizontal – Strip Plot - Horizontal Factor
3. Intersection Plot - interaction between the two factor

3.7.2 Advantages and Disadvantages

 Advantages
- Permits efficient application of factors that would be difficult to apply to small plots
 Disadvantages
- Differential precision in the estimation of interaction and the main effects
27
 - Complicated statistical analysis

3.7.3 Randomization
- Randomize the vertical factors and horizontal factors separately and independently.

3.7.4 Analysis of Variance for Strip – Plot Design

Source of Degree of Sum of Squares Mean Computed Tabular F
Variation Freedom Squares F 5% 1%
Replication (r – 1) ΣR2 - C.F RSS RMS
ab (r -1) Er(a)MS
Horizontal (a -1) ΣA -C.F ASS AMS
Factor(A) rb (a – 1 ) Er(a)MS
Error(a) r- 1)(a – 1) Σ( RA)2 / b –CF-RSS- Er(a)SS
ASS (r -1)(a – 1)
2
Vertical ( b- 1) Σ B - C.F BSS BMS
Factor(B) Ra (b -1) Er(b)MS
Error (b) (r – 1)(b – 1) (Σ( RB)2 )/a – CF-RSS-
BSS
AXB (a – 1)(b-1) (Σ (AB)2)/r–CF-ASS- (AXB)SS (AXB)MS
BSS (a – 1)( b -1) Er(c)MS
Error(c) (r-1)(a – 1) ToSS - SASS Er(b)SS
(b-1) ( r – 1)(a-1)(b-1)

Total (rab – 1) Σ X2 – C.F

Note : SASS = Sum of All other Sum of Squares ( SASS = RSS + ASS + EraSS + BSS + ErbSS +
ABSS )

28
3.7.5 Numerical Examples
Table 1.0 Data on Grain Yield of 3 Varieties of Rice grown with Four Nitrogen Rates in a Strip
– Plot Design with Three Replications

Grain Yield, t/ha

Nitrogen Rate,kg/ha
Rep I Rep II Rep III
V1
N1 2.0 4.0 4.3
N2 4.0 6.4 5.0
N3 7.3 6.8 8.6
N4 4.0 6.0 5.0
V2
N1 5.6 4.5 5.6
N2 7.0 6.6 7.0
N3 2.6 7.3 8.6
N4 4.6 5.6 3.8
V3
N1 4.8 7.0 4.8
N2 6.8 7.7 6.7
N3 4.4 3.3 4.6
N4 5.5 5.3 6.0

Table 1.1 The Replication X Variety Table of Yield Totals Computed from Data in Table 1.0

Yield Total (RA)

Variety Variety Total (A)
Rep I Rep II Rep III
V1 17.3 23.2 22.9 63.4
V2 19.8 24 25 68.8
V3 21.5 23.3 22.1 66.9
Rep Total(R) 58.6 70.5 70
199.1
Grand Total (G)

Table 1.2 The Replication X Nitrogen Table of Yield Totals Computed from Data in Table
1.0

Yield Total (RB) Nitrogen Total

Nitrogen
(B)
Rep I Rep II Rep III
N1 12.4 15.5 14.7 42.6
N2 17.8 20.7 18.7 57.2
N3 14.3 17.4 21.8 53.5
N4 14.1 16.9 14.8 45.8

29
Table 1.3 The Variety X Nitrogen Table of Yield Totals Computed from Data in Table 1.0
Yield Total (AB)
Variety
N1 N2 N3 N4
V1 10.3 15.4 22.7 15.0
V2 15.7 20.6 18.5 14.0
V3 16.6 21.2 12.3 16.8

Table 1.4 Analysis of Variance of Data in Table 1.0 from 3 x 4 Factorial Experiment in a Strip –
Plot Design.
Tabular F
Source of Degree of Sum of Mean
Computed F 5% 1%
Variation Freedom Squares Squares
Replication 2 7.55 3.78
Variety (A) 2 1.25 0.63 1.14ns 6.94 18
Error(a) 4 2.20 0.55
Nitrogen (B) 3 15.14 5.04 4.67ns 4.76 9.78
Error(b) 6 6.53 1.08
AXB 6 32.87 5.47 3.12* 3.00 4.82
Error(c) 12 21.02 1.75
Total 35 86.56
b* = significant, ns = nonsignificant
cv(a) = 7.09%
cv(b) = 10.96%

30
MODULE – IV. COMPARISON BETWEEN TREATMENT
MEANS
4.1 PAIR COMPARISON
4.1.2 Two Types :

1. Planned Pair Comparison

2. Unplanned Pair Comparison

4.1.3 Most commonly Used Test :

1. Least Significant Difference ( LSD ) – suited for a planned pair comparison

2. Duncan’s Multiple Range Test ( DMRT ) – applicable for unplanned pair comparison

4.2 LSD Test

- Suited for planned pair comparisons not valid for comparing all possible pairs of
means
– When no. of treatments is large.

Steps :

1. Compute the difference between the ith and jth treatments :

dij = Xi – Xj

2. Compute LSD value at δ level of significance

LSDδ = tδ ( Sd ˉ )

Where :
Sdˉ = standard error
Tδ = tabular value at appendix C( n = Error df )

3 Compare the mean difference computed in step 1 and step 2.

- declare significantly at the level of significance if the absolute value of dij is greater that
the LSD value.

31
4.3 STANDARD ERROR (Sdˉ)
A. For CBD
- Only one error term is involved CRD, RCBD and LSD
- Standard error of the mean difference for any pair of treatment means
2S 2
1. Sdˉ = for Equal Replication
r

1 1
2. Sdˉ = S2   for Unequal Replication
r r 
 i j 

B. For Split – Plot Design

1. Differences between 2 A means

Sdˉ = 2Ea / rb with (r-1)(a-1) df

2. Differences between 2 B mean

Sdˉ = 2Eb / ra with a(r-1)(b-1) df

3. Differences between B means at same level of A

Sdˉ = 2 Eb / r with a(r-1)(b-1) df

4. Difference between A means at same or different level of B

2b  1Eb  Ea  / rb
Sdˉ =

with weighted Tabular t – value

t’ = (b-1) Ebtb + Eata

(b-1) Eb + Ea

Where:
Ea = Error(a) MS
Eb = Error(b) MS
r = no. or replication
a = no. of mainplots
b = no. of sub-plot

32
 C. For Strip – Plot Design

1. Differences between 2 A means

2 Ea
Sdˉ =
rb
2. Differences between 2 B means
2 Eb
Sdˉ =
ra

3. Differences between A means at same level of B

2b 1Ec  Ea 
Sdˉ =
rb
with t’ = (b-1)Ectc + Eata
4. Difference between B means at same level of A (b-1)Ec + Ea
2a 1Ec  Eb 
Sdˉ =
ra
with t’= (a-1)Ectc + Ebtb
5. Differences between A and B means at diff. levels (a-1)Ec + Ea
2ab  a  b Eab  a( Ea )  b( Eb)
Sdˉ =
rab

4.3.1 Numerical Examples:

Table B. Comparison between Mean Yields of Treatment D and each of the four Treatments
A,B,C and E using LSD Test Data from a CRD experiments.
Dij(Treatment Difference), b
Treatment Mean Yield, a kg/ha
kg/ha
A 5.2 2.4ns
B 7.8 5.0**
C 4.0 1.2ns
D 2.8 ---
E 6.6 3.8*
a =
Average of five Replication
b**
= significant at 1% level, * = significant at 5% level, ns = not significant

From Appendix C, n = 20 = Erdf, t0.01 = 2.845, t0.05 = 2.086

Sdˉ = 1.52, LSD0.05 = 2.086(1.52) = 3.17 kg/ha, LSD0.01 = 2.845(1.52) = 4.32 kg/h

33
34
4.3.2 LSD Test for Split – Plot Design

Four (4) Different Types of Pair Comparison

1. Comparison between to main-plot treatment means averaged over all subplots

treatment
2. Comparison between two subplot treatment means averaged over all main plot
treatment
3. Comparison between subplot treatment means at the same main-plot treatment
4. Comparison between two main-plot treatments means at the same or different
Subplot treatments (means of any two treatments combinations).

Table A. 4 Comparison between Mean Yields of Two Variety Means with same Nitrogen Rate

Nitrogen Rate, Mean Yield , kg/ha

kg/ha
V1 V2 V V4
N0 4.47 4.30 3.20 4.43
N1 5.67 6.03 5.40 4.80
N2 6.37 6.23 5.97 4.80
N3 6.73 6.87 6.10 3.77

Steps (Data from Table A.3)

2 Eb
1. Sdˉ = = 0.48
r

2. tδ at n = Eb = 24
t0.01 =2.797
t0.05 =2.064

3. LSD0.05 = 2.064(0.48) = 0.98 kg/ha

LSD0.01 = 2.797(0.48) = 1.33 kg/ha

35
4.4 DUNCAN’S MULTIPLE RANGE TEST
- Experiments that require the evaluation of all possible pairs of treatment means.

4.4.1 Steps for Computing DMRT Values :

1. Rank all the treatment means in decreasing or increasing order

- Rank according to preference
2. Compute the Sdˉ
3. Compute the (t -1) values of the shortest significant ranges as:

Rp = (rp )(Sdˉ), for p = 2,3….t, (rp ) at Erdf at Appendix F

√
4. Identify and group together all treatment means that do not differ
Significantly from each other.
A. Compute the difference between the largest treatment mean and the largest Rp
value

dTrRp = HTr - HRp

A.1 Compare dTrRp to the remaining Treatment Mean all less than dTrRp are
declared significantly different from the treatment mean being compared.
A.2. Compute the difference between the treatment group
dTi.Tj = Ti - Tj
A.3 Compare the dTi.Tj to the Rp value at p = m, if dTi.Tj is less than the
Rp value at p = m,declare not significantly different
B. Continue the process in A until all the treatments are being compared.
5. Present the test results in one of the following ways:

1. Line Notation – if the treatments is arrange in rank

2. Alphabet Notation – if the treatments is arrange in sequence

36
DMRT for C.B.D ( Complete Block Design )

Rank
Treatment Mean Yield, Kg / ha
T2 : Dolmix ( 2 kg ) 2,678 1
T3 : DDT + δ – BHC 2,552 2
T4 /: AZODRIN 2,128 3
T1 : Dol – mix ( 1 kg ) 2,127 4
T5 : Dimecron – Boom 1,796 5
T6 : Dimecron ÷ Knap 1,681 6
T7 Control 1,316 7
Control

Rp = ( r p )( Sdˉ ) for p = 2,3…t

√
At error df. = 21 at the 5% level at Appendix F

P rp ( 0.05 )
2 2.94
3 3.09
4 3.18
5 3.24
6 3.30
7 3.33

The ( t – 1 ) Rp Values :

p Rp = ( r p b )( Sdˉ )
√

2 ( 2.94 ) ( 217.68 ) = 453

√
3 ( 3.09 ) ( 217.68 ) = 476
√
4 ( 3.18 ) ( 217.68 ) = 489
√
5 ( 3.24 ) (217.68 ) = 499
√
6 ( 3.30 ) ( 217.68) = 508
√
7 ( 3.33 ) ( 217.68 ) = 513
√

37
Table 5.2 DMRT for Comparing all possible Pairs of Treatments Means, from a CRD
Experiment Involving Seven Treatments, Using the Line notation ( Data Table 2.1

Treatment
Mean yield, kg/ haa DMRTb
T2 2,678
T3 2,552
T4 2,128
T1 2,127
T5 1,796
T6 1,681
T7 1,316
a = Average of four replication
b = any two means connected by one same vertical line are not significantly different at 5% level
of significance

Table 5.2 DMRT for Comparing all possible Pairs of Treatments Means, from a CRD
Experiment Involving Seven Treatments, Using the Alphabet notation ( Data Table 2.1

Treatment
Mean yield, kg/ haa DMRTb
T1 2,127 bc
T2 2,678 a
T3 2,552 ab
T4 2,128 bc
T5 1,796 c
T6 1,681 cd
T7 1,316 d
a = Average of four Replication
b = any two means having a common letters are not significantly different at the 5%

38
4.5 Tukey HSD Post – hoc test
It is needed after an ANOVA is completed and it should be significant in order to
determine which groups differ from each other.

Steps For Conducting Tukey Test

1. Find the differences between the means of all groups.
2. Compare the differences to critical values using the formula below;

Where :

3. Reject if step 1. value is greater than step 2 value.

Numerical Example:

Valencia oranges were tested for juiciness at 4 different orchards. Eight oranges were
sampled from each orchard, and the total ml of juice per 20 gms of orange was
calculated given below:

a. Test for a difference in Orchards using alpha = .05

b. Perform all the pairwise comparisons using Tukey's Test and an overall risk level of
5%.

39
 Source of Degree of Sum of Mean Tabular F
Computed F
Variation Freedom Squares Squares 5% 1%
Treatment
Experimental
Error
Total

40
41
MODULE – V. REGRESSION AND CORRELATION ANALYSIS

5.1 Three Groups of variables that are normally recorded in crop experiments:

1. Treatments – such as Fertilizer rates, varieties and etc.

2. Environmental Factors – such as rainfall and solar radiation

3. Responses – Biological and physical features of the experimental units that expected to be
affected by the treatments being tested.

- Response to treatments such as:

1. Crop Response – changes in biological features such as grain yield and plant
height.
2. Noncrop Response – changes due to the surrounding environment

5.2 Four Classification of Regression and Correlation Analysis:

1. Simple Linear Regression and Correlation

2. Multiple Linear Regression and Correlation
3. Simple Nonlinear Regression and Correlation
4. Multiple Nonlinear Regression and Correlation

5.3 LINEAR RELATIONSHIP

- If the change is constant throughout the whole range
- It is represented by a straight line in the graph

5.4 SIMPLE LINEAR REGRESSION AND CORRELATION

5.4.1 Two Conditions to consider before applying SLRC:

1. There is only one independent variable X affecting the dependent variable Y.
2. The relationship between Y and X is known, or can be assumed, to be linear.

5.4.2 SIMPLE LINEAR REGRESSION

Y    X

Where :
  intercept of Y axis
 = linear regression coefficient or slope of the line

42
5.4.3 Step by step procedures for solving Regression Analysis:

1. Compute for the means, sum of squares and cross – product of x and y.
2. Compute the estimates of the regression parameters  and 
Note: x and y – deviation from means

a = y – bx

 xy
b=
 x2

3. Test the significance of 

a. Compute the residual mean square:

  xy 2 
x  2 
  x2 
s y2. x   
n2

b. Compute tb

b
tb 
s y2. x
x 2

c. Compare tb to tabular t value of Appendix C, with df = n -2

If tb> t tabular,  is significant at a level of significance.

Note: Linear Regression is only useful if there is evidence of a linear relationship between x and
y. If the correlation is very low, then the line will fit horribly and will be useless.

43
5.4 SIMPLE LINEAR CORRELATION ANALYSIS

5.4.1 The Correlation Coefficient

The linear correlation coefficient (denoted r) measures the strength of the linear
relationship between two variables x and y in a sample. It is sometimes called a Pearson
Correlation Coefficient, after a famous statistician named Karl Pearson. The formula for this
statistic is as follows:

∑ ∑ ∑
√[ ∑ ∑ ][ ∑ ∑ ]
Where:
x = the observed data for the independent variable
y = the observed data for the dependent variable
n = is the number of observation or size of the sample
r = the degree of the relationship between x and y

Note: If absolute value of computed r is greater than tabular r values from Appendix H with df =
n – 2 , then, r is significant at a level of significance.

5.4.2 Properties of the Correlation Coefficient

Due to the standardization that takes place in the formula, there are a couple of interesting
properties of r
1. If the values of either variable are converted to a different scale, r will be the same.
2. If the variables x and y are interchanged, r will be the same.
3. The correlation coefficient r will only measure the strength of a linear relationship. It says
nothing about other kinds of relationships.

The sign of the correlation coefficient tells whether the line is sloping upward (positive) or
downward (negative). When r is close to 0, there seems to be no linear pattern at all. Here is a
general guide for how to interpret the value of r :

Range of Values Strength of Linear Relationship Direction of Linear Relationship

-1 to -0.8 Strong Negative

-0.8 to -0.6 Moderate Negative
-0.6 to -0.3 Weak Negative
-0.3 to 0.3 None None
0.3 to 0.6 Weak Positive
0.6 to 0.8 Moderate Positive
0.8 to 1 Strong Positive

44
45
6.0 INTRODUCTION TO SPSS

6.1 What is SPSS?

SPSS (Statistical Package for Social Stduies) is a windows based program that can be
used to perform data entry and analysis and to create tables and graphs. It is capable of handling
large amounts of data and can perform all of the analyses covered in the text and much more.

Layout of SPSS

The data editor window has two views that can be selected from the lower left hand side
of the screen. Data view is where you see the data you are using. Variable view is where you can
specify the format of your data when you are creating a file or where you can check the format of
a pre-existing file. The data in the Data Editor is saved in a file with the extension. Sav.

SPSS Menus and Icons

File
Includes all the options typically use in other programs, such as open, save, exit and etc.

Edit
Includes the typical cut, copy and paste commands and allows you to specify various options
for displaying data and output.

View
It allows you to select which tool bars you want to show.

Transform
It includes several options to change the current variables.

Analyze
Includes all the commands to carry out statistical analyses and to calculate descriptive
statistics.

Graphs
Includes commands to create various types of graphs including box, histograms and etc.

Utilities
It allows you to list file information which is a list of all variables, there are labels, values ,
locations in the data file and type.

Adds – on
Programs that can be added to the base SPSS package.

46
Help
It includes link to the SPSS homepage, statistics coach, and syntax guide. Using topics, you
can use the index option to type in any key word and get a list of options.

47
 Laboratory Exercises

Part I. Randomize and lay out a CRD experiment with 4 replication and 5 treatments
(T1, T2, T3, T4, and T5) using random numbers in calculator.

48
Part II. Randomize and lay out a RCBD experiment with 6 replication and 4 treatments
using random numbers in calculator.

49
 Part III. Provide what is required for the following:

Table 1.0 Germination percentage of abaca seeds under different plant residues with carabao manure as
based compost from CRD experiment with 3 ( r ) replications and 6 ( t ) treatments.
Treatment Replication
T1(ordinary garden soil) 76.67 81.67 85.00
T2(Ordinary Soil+Carabao Manure) 73.33 75.00 95.00
T3(Ordinary Garden Soil T Carabao Manure+Abaca Leaves) 68.33 93.33 85.00
T4(Ordinary Garden Soil+Carabao Manure+Abaca Corms) 86.67 80.00 68.33
T5(Ordinary Garden Soil+Carabao Manure+Abaca Bracts) 73.33 85.00 61.67
T6(Ordinary Garden Soil + Carabao Manure+Coconut Husk) 88.33 60.00 86.67

Required:
1. Filled Anova

2. Make Decision

50
51
Table 3.0 Data on the average weight in grams of filled grains with different treatments
from RCBD Experiment with four replication.

Yield in grams (g)

Treatment

T1 9.77 7.60 7.83 7.83

T2 9.73 7.55 7.97 7.97
T3 9.46 7.53 7.26 7.78
T4 9.06 9.45 8.75 7.60
T5 8.96 8.96 9.73 8.55
Control 8.70 8.70 8.46 7.53

Required:
1. Filled Anova
2. Make Decision

52
53
Table 4.0 Yield of three promising maize hybrids ( A,B, and D ) and a check
variety ( C )from an experiment with LSD.

Yield, Kg/plot
Row Number
Col. 1 Col. 2 Col. 3 Col. 4
1
165 ( B ) 131 ( D ) 152 ( C ) 144 ( A )
2
148 ( D ) 128 ( A ) 176( B ) 139 ( C )
3
168 ( C ) 100 ( B ) 175 ( A ) 128 ( D )
4
157 ( A ) 139 ( C ) 150 ( D ) 76 ( B )

Required:
1. Filled Anova
2. Make Decision

54
55
56
Table 5.0 Data on Different Sizes (Fine, Medium and Coarse)
of Materials in grams as Affected by the Five different Number
of Passes in a Split - Plot Design with three Replication

Size
Weight in Grams
Rep I Rep II Rep III
T1(One pass)
Fine 35 40 32
Medium 55 70 55
Coarse 198 180 200

T2( Two passes)

Fine 60 62 68
Medium 125 115 103
Coarse 100 110 113

T3(Three passes)
Fine 86 93 79
Medium 135 130 135
Coarse 60 60 68

T4( Four passes)

Fine 95 99 94
Medium 130 128 140
Coarse 50 50 45

T5(Five Passes)
Fine 97 109 101
Medium 135 127 127
Coarse 40 38 48
Control
Fine 9 12 18
Medium 12 18 18
Coarse 270 267 261

Required:
1. Filled Anova
2. Make Decision

57
58
59
60
Table 6.0. Grain Yield Data of four Rice variety Grown with 4 levels of
Nitrogen in a Strip - Plot Design with Four Replication

Variety
Grain Yiled , t/ha
Rep I Rep II Rep III Rep IV
N0(0 kg/ha)
V1 16 16 15 15
V2 15 16 14 16
V3 14 14 14 14
V4 15 15 15 15
N1(60 kg/ha)
V1 16 15 15 15
V2 15 16 14 15
V3 15 15 15 14
V4 16 15 15 16
N2(90 kg/ha)
V1 17 16 16 15
V2 15 16 15 16
V3 14 13 14 14
V4 15 15 15 15
N3(120 kg/ha)
V1 18 17 16 16
V2 15 16 14 16
V3 15 14 15 14
V4 15 15 14 14

Required:
1. Filled Anova
2. Make Decision

61
62
63
Table 7.0. Grain Yield Data of four Rice variety Grown with 4 levels of
Nitrogen in a RCB Design with Four Replication

Variety Grain Yiled , t/ha

Rep I Rep II Rep III Rep IV
N0(0 kg/ha)
V1 16 16 15 15
V2 15 16 14 16
V3 14 14 14 14
V4 15 15 15 15
N1(60 kg/ha)
V1 16 15 15 15
V2 15 16 14 15
V3 15 15 15 14
V4 16 15 15 16
N2(90 kg/ha)
V1 17 16 16 15
V2 15 16 15 16
V3 14 13 14 14
V4 15 15 15 15
N3(120 kg/ha)
V1 8 7 6 6
V2 5 6 4 6
V3 5 4 5 4
V4 5 5 4 4

Required:
1. Filled Anova
2. Make Decision

64
65
66
Table 8.0 Plant heigt (cm) Data Resulting from the use Different treatment
( t,1 t,2 t,3 t,4 and t 5 ( control ) CRD experiment with 5 replications.

Treatment Treatment Total

T1 44.5
T2 40.8
T3 34.1
T4 34.4
T 5 (control) 31

Analysis of Variance of Table G1.

Source of Variation df SS MS Comp. F Tabular F

5% 1%
Treatment 4 24 6 21** 2.87 4.43
Experimental Error 20 6 0.3
Total 24 30

Required:
1. LSD test
2. Present the final result and interpret

67
10. From the given below compare the following treatment means using DMRT and interpret the
result. ANOVA Result from RCBD is highly significant and with four (4) replications and error mean
square is 97,774.
Table 9.0 Analysis of Variance on the height gained of plants at 60 to 90 days.

Source of variation df SS MS Comp. F Tabular F

5% 1%
Replication 2 14.49 7.25
Treatment 2 67.28 33.64 6.94 18
Error 4 51.16 12.79
Total 8 132.93

Required:
1.LSD test
2 Present the final result and interpret.

68
 Treatment Total
T1 28,508
T2 20,712
T3 20,208
T4 18,512
T5 27,184
T6 16,724
T7 15,264

Note:
Present the Final DMRT Result in Alphabet and Line Notation.

69
70
71
Table 11.0 Mean Yield Of 10 Treatments and the Control Treatment with Unequal Replication

Treatment Replication Treatment

Number No. Total
1 4 14,576
2 3 9,040
3 4 11,793
4 4 11,638
5 3 27,703
6 3 27,694
7 4 19,934
8 3 6,618
9 4 8,164
10 4 11,192
11(control) 4 4,949

Note: ANOVA result is highly significant and error mean square

is 176,532.

Required :
1. LSD test

72
73
74
Table 13.0 Grain Yield and Nitrogen Rate using the Data in Rice Experiments

Nitrogen Rate, Kg/ha Grain Yield, Kg/ha

25 15,124
50 15,070
75 15,304
100 14,848
125 14,708
150 14,703

Required:

1. Estimated Simple Linear Regression Equation

2. Draw a graphical representation of the estimated regression equation

75
76
Part IV. SPSS

Answer the following;

1. Look up ANOVA in Help/help topics. What kind of information did you find?

2. Look up “compare groups for significant differences” in Help/Statistics Coach. What did you
learn?

77
78
79
80
81
82
83
References:

1. Statistical Procedure for Ag. Research by K. Gomez and A. Gomez

2. Statistics for Engineers and Scientist 3rd Ed. by Walpople

3. Design of Experiments for Agriculture and the Natural Sciences by A. Reza

Hoshmand

4. Agricultural Field Experiments : Design and Analysis by Roger G. Petersen

5. http://www.stat.psu.edu/online/courses/stat.503/01_intro_intro_html

6. http://www.books.google.com.ph/books?id

84