CONTENTS

I. Introduction 3

II. Conducting an Investigatory Project:

Problem Identification and Clarification 6

III. Formulating the Objectives 9

IV. Choosing the Research Design 10

V. Basic Experimental Designs 12

VI. Collecting, Analyzing and Interpreting Data 28

VII. Making Conclusions and Recommendations 30

VIII. Writing the Abstract 31

IX. Writing the Report 32

2
 I. INTRODUCTION

A. Science Investigatory Projects

1. Definitions

There are two central themes around which most views of Science may be grouped:
Science is basically a method of inquiry or of investigation; and Science is the result of an
inquiry. The first is focused on method and the other, or knowledge itself.

Focused on method, Science is defined as an objective, accurate, systematic analysis

of a determinate body of empirical data, in order to discover recurring relationships among
phenomena. The keywords in the definition are:

1.1 Objective which requires that the inquiry must be unbiased, unprejudiced, detached
and impersonal. The investigator should not become ego-involved in the research
activities.

1.2 Accurate which requires that variables and measurements should be definite, precise
and exact.

1.3 Systematic implies that science is methodical, thorough and regular in its procedures.
This is characterized by the interrelations of an events various parts and these parts
form a coherent whole and not merely a collection of facts.

1.4 Analysis implies that science is concerned with the identification and study of the
component parts or elements which make up the things under study.

1.5 Determinate means that the investigation should identify the traits or boundaries of
things one studies.

2. Aims of Science

The aims of science may be categorized into three domains.

2.1 Description this directs the investigator to detail what he is talking about. He should
know what the facts are, what the case is and what is out there.

2.2 Explanation this denotes that the investigator should show how things and events
come about. It brings meaning to description. The investigator should be able to show
what causes events and, if possible, to identify factors which determine the
occurrence of these events.

2.3 Prediction this is the inference from facts, observations or laws. This domain
implies a sort of cause and effect relationship like If this is so what will happen?

3
 3. The Scope of Science Investigation

Science covers all known disciplines: Natural Science, Social Science, Engineering,
etc. Focused on method of investigation, the scope may be grouped under the following
categories:

3.1 Generalizations Those which can be verified through actual observation

3.2 Controlled investigation or inquiry That which demonstrates the relation between
theory and reality
3.3 Prediction That which demonstrates conditional relationship that consistently hold
within stated limits
3.4 Causality That which seeks single and multiple causes
3.5 Naturalism and determinism
3.5.1 Phenomena and aspects of nature. Things can be made according to the order of
nature
3.5.2 Phenomena can be determined and all orders can be accounted for or explained

B. The Method of Science

As defined earlier the method of science involves an objective, accurate and systematic
analysis of a determinate body of knowledge. Whatever the discipline maybe the method of
science is what most of us call the Scientific Method.

These are implied steps attributed to the Scientific Method:

1. Identifying the problem

2. Defining the problem
3. Formulating the hypothesis
4. Deductive Reasoning
5. Testing the hypotheses

Given other names and whether implied or stated, the above elements would comprise the
procedures for conducting investigatory projects or of conducting research.

C. The Investigatory Project

The investigatory project is a careful, diligent, and exhaustive investigation of a specific

subject matter, having as its aim the advancement of knowledge. It is also a systematic,
controlled, empirical and critical investigation of hypothetical propositions about the
presumed relations among natural phenomena.

4
Conducting investigatory projects involve various stages. They are:

1. Problem Identification and Clarification

2. Formulating the Objectives
3. Choosing the Research Design
4. Collection of Data
5. Analysis and interpretation of data
6. Making conclusions and recommendations
7. Writing the Abstract
8. Writing the Report

5
 II. Conducting the Investigatory Project: Problem Identification and
Clarification

A. Identifying the Research Problems

The investigatory project starts with a research problem. A research problem is nothing
but the issue to be settled or the question to be answered. This might arise from the nature or
characteristics of an event or phenomenon or from the relationship between two or more
variables. The research problem may come from:

1. Intellectual curiosity

This may involve the analysis of needs, practices, and observations. For example two
neighbors are taking care of orchids of the same variety and species. Both the orchids are
of the same age but one is already flowering after two months. The neighbor will ask
herself and her neighbor why their orchids did not flower at the same time. Asking the
question is being curious and it is a starting point of an investigation.

2. Serendipity

Somebody with research interest just come up with so many ideas about things and
possibilities. The testing of any of these ideas can become an investigatory project.

3. Expert opinion

In view of experiences and exposures of experts, their statements or opinions can

form part of the sources of research problem. These opinions can be sought in written
communication or in oral communication. For example, you ask the question on why the
frequency of flooding in Tuguegarao has increased. Experts will have varied opinions
like:

Massive denudation of forest

Silting of downstream Cagayan
La Nia season, etc.

4. Literature review

This can be the richest source of research problems. In fact the literature can help in
validating issues. Other uses are that literature can:

Further define the problem

Provide theoretical basis of the research question
Help in the interpretation of results
Help point out the implication of the study

6
B. Formulating The Research Problem

The first thing to consider before formulating the research problem is determine whether it
is researchable, can data be collected to solve the problem or the research question? After
deciding that the issue is researchable you can next consider the following:

1. Significance This implies answers to the following:

Is the potential result worth the effort, time and money that will be put into the work?
Will it fill an important gap in knowledge?
Will it have possible practical application?
Will it improve the quality of life?

2. Feasibility
Is the required methodology feasible?
Are there enough subjects for study?
Are resources available?
Will the potential result be
Politically significant?
Economically viable?
Socially acceptable?
Technically feasible?
Environmentally friendly?

C. Defining the Research Problem

The bottom line in defining a research problem is to ensure that depth of the coverage
rather than the scope is achieved. To ensure this, one must

1. Dissect and Clarify the Problem

1.1. Break down the problem into smaller components
1.2. Each component should be researchable
1.3. Clarify the problem through consultation with experts or through a review of
literature

2. State of the problems and sub-problems clearly

2.1. Look at the possible relations between two or more variables
2.2. State the problem in question form
2.3. State the problem to imply possibilities for empirical testing

3. Answer the basic questions

3.1. Is the problem new?
3.2. Is the problem interesting?
3.3. Will the study add to knowledge?
3.4. Has anyone else studied it?
3.5. Do I have the resources to study it?

D. Formulating the Hypothesis

7
 The hypothesis is a tentative solution to the problem which is going to be tested in the
study. This is a statement of an idea which could answer the problem or the sub-problem.

The hypothesis then guides the researcher in identifying, categorizing and defining the
pertinent variables involved in the study. It will help him establish the frame of reference with
which to approach the problem.

1. Types of hypotheses

There are two types of hypotheses. The null (H0) and the alternative (H1) hypotheses. The
null hypothesis is a statement of no difference. For example:

H0: There is no difference in yield of plants given 10 kg Nitrogen and yield of plants not
given Nitrogen.

On the other hand, alternative hypothesis may take the form of a directional statement. As
in the example above:
H1: The yield of plants given 10 kg Nitrogen is higher than the yield of plants not given
fertilizer.

2. Additional notes on hypotheses

2.1 The main purpose of a hypothesis is to deduce consequences.
2.2 Hypotheses are not supposed to be confirmed.
2.3 Hypotheses can not verify, they make possibility statements.
2.4 Hypotheses should be explained simply. The simpler the hypothesis the easier it is to test.

III. Formulating The Objectives

8
 Objectives are the goals of action. They are the outcome the researcher hopes to come up to
attain. Being such, they must be specific and achievable. They must fit properly into the problem
stated. When stated clearly, the objectives serve the following uses

1. Give an indication of the relevant variables to be considered in the study

2. Guide the researcher in the choice of an appropriate research design
3. Guide the researcher on what data will be collected
4. Provide basis for the analysis and interpretation of data

A. Statement of objectives

1. The general objective This is the over all purpose of the research undertaking. It is
essentially a statement of the hypothesis as expected outcome

2. The specific objectives These are statements of the sub-problems. Some guides or
reminders on specific objectives are:
The attainment of specific objectives collectively should achieve the general objective
Specific objectives should be written clearly as statements which incorporate the specific
hypotheses
Specific objectives should specify the variables or indicators of the events to be observed,
measured or controlled
Specific objectives should be expressed in measurable terms.

B. Objectives and Variables

Variables are events, factors or characteristics to be studied which can take on several
values in response to interventions. They are the indicators used to detect the events or changes
in the events under investigation. They are the object of measurement.

The types of variables are

1. Qualitative and quantitative. Qualitative variables are not numerically expressed
whereas quantitative variables are expressed numerically
2. Independent and dependent. The independent variable affects the value of the
dependent variables whereas the dependent variable is influenced by the independent
variable. In other terms, the independent is the casual variable and the dependent is
the result variable
3. Extraneous variables. These are variables beyond the control of the researcher.

IV. Choosing the Research Design

9
The research design is the plan, structure and strategy of the investigation.

The plan is the overall scheme to be followed

The structure is the more specific outline of how the variables are operationalized
The strategy is the description of details such as methods to be employed in collecting
and analyzing data

The main target of the research design is to attain the objectives of the study.

A. The Basic Research Designs

1. Descriptive Design
The descriptive design is used for inquires into the nature of an event or a phenomenon.

It seeks to identify the characteristics of the event and categorize it into some
descriptive variables.

An example of an investigatory project of this type is the Identification of the

optimum level of N for Crop yield

The object event is crop yield

The independent variable is N level
The dependent variable is yield
The objective is which N level gives the highest yield

The design consists of the treatments and the number of replicates. Thus we have:

Treatments: 0 fertilizer
5 kg
10 kg
Replicate 3 or 4 of each treatment
Analysis technique simple CRD
The findings answer which level of N gives the highest yield

2. Analytic Design
The analytic design tests hypothesis of relationship. There are two groups of this
design.

Observational Involves test of hypothesis without manipulating the study factors. This is
usually based on records of past events.
Experimental The researcher manipulates the study factors. An important pre requisite of this is
the use of randomization or random sampling.

10
 3. Experiments

Characteristics of well planned experiments

Simplicity the selection of treatments and the experimental arrangement should be as simple as
possible, consistent with the objectives of the study.

Degree of precision The probability should be high that the experiment will be able to measure
treatment differences. This implies an appropriate experimental design and sufficient replication
or number of observations.

Absence of systematic error The administration of treatment should minimize experimental

errors.

Range of validity of conclusion.

Calculation of degree of uncertainty.

4. Basic Requirements of Experiments

Random assignment of treatments

Treatment and Control (no-treatment) group
Observations during the experiment and other factors which may affect the
experimental treatments.

11
 V. Basic Experimental Designs

Experimental design is a system of procedures regulating the assignment of treatments to

experimental plots. The basic principles involved in these designs are replication, randomization,
and error control. These three are aimed towards increasing precision of the experiment.

Replication. The precision of an experiment can always be increased by additional replication

but the degree of improvement may fall off as the number of replication increases.

Randomization. The procedure involved in the selection of samples or the assignment of

treatments in a manner that the probability of drawing a sample or assigning a treatment is the
same as any other sample or treatment.

Error Control. The various experimental techniques administered to increase precision. In

field experiments these may take the form of managing one or more or all of the following:

1. Plot size, shape, orientation

2. Replication
3. Blocking
4. Randomization
5. Border Effects
6. Residual effects
7. Missing data/observations
8. Treatment and experimental materials
9. Sampling techniques

A. Completely Randomized Design

The Completely Randomized Design (CRD) is used when the experimental units are
essentially homogenous. For example, one wishes to find out the effect of hormones on
weight gain of pigs. Three hormones are available. The experimental units (pigs) may be
obtained from a piggery. Homogeneity can come in the form of:

1. piglets from a litter, if sufficient

2. piglets of equal age from litters of the same breed
3. piglets with the equivalent weights

The assignment of treatments to the experimental units will have no restrictions in a CRD.

Suppose these are 16 experimental pigs selected on the basis of homogeneity. There are
three hormones to be tested but considering the treatment and no-treatment basic principle in
experimentation, there can be 4 treatments: A-control; B-Hormone 1; C-Hormone 2; D-
Hormone 3. Now we have 4 treatments and can allocate 4 experimental units each to a
treatment. We therefore have 4 replications.

12
 In the assignment of the treatments to the experimental units there is no restriction in
randomization. By this we mean that it is possible that many of the piglets may be designated
to receive any of the treatments. Suppose that our resources especially the pigpens can allow
the floor plan below, we can devise a random procedure for the allocation of treatments.

Floor plan of pigpens

1 2 3 4

5 6 7 8

9 10 11 12

13 14 15 16

Randomization Procedure
1. Assign numbers to the pigpens and place one piglet into each.
2. Cut two pieces of bond paper into 8 pieces. On the pieces write the letters
corresponding to the treatments so that there will be four pieces labeled A, four
labeled B, four labeled C and four labeled D.
3. Fold the pieces separately and place them in a container for drawing lots.
4. Get one and assign to pen 1; the next will be pen 2 and so on until the last piece is
drawn for pen 16.

13
 Suppose the lots were drawn and you have:

1 2 3 4
C B A D

5 6 7 8
A D B C

9 10 11 12
B A D C

13 14 15 16
B D C A

The above is now your experimental layout. The features:

Four treatments: A, B, C, D
Four replicates per treatment

You now administer the treatment.

Maybe you can mix the hormone with the feeds such that

A feed only, no hormone

B Hormone 1
C Hormone 2
D Hormone 3

After 100 days of tender loving care and giving the same attention to the piglets except for
the hormone treatment (Experimental variables) you weigh them. You get the weight in
kilograms.

57 50 47 54

52 65 54 53

67 62 74 69

57 59 57 51

14
 Transfer the result into an observation table ready for analysis.

Table 1. Effect of hormones on weight gain of pigs (kg/pig)

Replication
Treatment 1 2 3 4 Total
A 47 52 62 51 212
B 50 54 67 57 228
C 57 53 69 57 236
D 54 65 74 59 252
Ho: TA = TB = TC = TD

The data are ready for statistical analysis using the CRD

The model for analysis is the linear additive model:

X=u+T+E
X = any given observation in weight
T = Treatment effect (Hormones)
E = Experimental error effect
u = The theoretical population mean

In the above model it is assumed that if there is any difference in the weight of the
pigs, this due to the hormones (T) and experimental error (E) since the experiment started
with homogeneous experimental materials (equal weights).

The CRD Analysis of Variance

ANOV Table

Source of Degree of Sum of Mean

Variance Freedom Squares Squares F-ratio
Treatment t-1 TrSS TrMS TrMS/EMS
Error [(N-1)-(t-1)] ESS EMS
Total N-1 ToSS

Procedures (From data in Table 1)

1. Compute Treatment Totals/Treatment means

1.1 TA = 212 XA = 53
1.2 TB = 228 XB = 57
1.3 TC = 236 XC = 59
1.4 TD = 252 XD = 63

15
2. Compute Grand Total

GT = TA + TB + TC + TD
= 212 + 228 + 236 + 252 = 928

3. Compute Correction Factor (CF)

(GT)2 (928)2
CF = =
N 16

CF = 53824

4. Compute Total Sum of Squares = ToSS

ToSS = [472 + 522 + 622 + 512 + +592] CF

= 54678 53824 = 854

ToSS = 854

5. Compute Treatment Sum of Squares = TrSS

2 2 2 2
TrSS = TA + TB + TC + TD - CF
r

(212)2 + (228)2 + (236)2 + (252)2

= - 53824
4

= 54032-53824 = 208

TrSS = 208

6. Compute Error Sum of Squares = ESS

ESS = ToSS TrSS

= 854 208
ESS = 646

7. Accomplish ANOV Table

16
 Source of Degree of Sum of Mean F-ratio F .05
Variance Freedom Squares Squares
Treatment 3 208 69.3 1.29 n.s. 3.49
Error 12 646 53.8
Total 15 854
TrSS
TrMS =
df

ESS
EMS =
df

F-ratio = TrMS
EMS

8. Compare F ratio computed with standard (Tabular value at P = 0.05 and error d.f.)

Fc > F .05 ; significant

Fc> F .01 ; highly significant

Fc < F .05 ; not significant

9. Since Fc < F .05, the difference among treatment means are not significant. Therefore
Ho is accepted.

Advantages of CRD

1. The design is flexible. The number of treatments and replicates are limited only by
the availability of experimental units.
2. Number of replicate may vary from one treatment to another.
3. Simple analysis.
4. Simplicity of analysis is not lost if some experimental or entire treatments are lost or
rejected.
5. Error degree of freedom is maximum and improves precision.
6. Loss of information due to missing data is small relative to loses with other designs.

Disadvantages of CRD

1. Often inefficient.
2. Experimental error includes entire variation among experimental units except that due
to treatment. This is due to unrestricted randomization.

B. Randomized Complete Block Design

17
 The randomized Complete Block Design (RCBD) allows experimental units to be
meaningfully grouped. The number of units in a group is equal to the number of treatments or
multiples of it. The object grouping is to have the units in a block as uniform as possible so
that observed differences would be largely due to treatments.

Suppose we go back to our example on the effect of hormones on weight gain in pigs
presented earlier on a CRD. We have the same floor plan as before but this time we consider
the rows of 4 pens each as blocks. There will now be 4 treatments and RCBD imposes that all
treatments are in each block.

1 2 3 4
I

5 6 7 8
II

9 10 11 12
III

13 14 15 16
IV

Randomization procedure randomization will now be done in each block. The four
treatments are assigned to the four pens at random. For this we need just four pieces of paper
with one treatment written in each paper (A, B, C, D) each treatment in a piece, fold and
draw lots for Block I, then for Block II, Block III and Block IV.

Suppose the result of randomization is as follows

18
I A B C D

B A D C
II

III B D A C

IV C B D A

The treatments are imposed on the piglets and after 100 days the weights are obtained

I 47 54 57 50

II 54 52 65 53

III 67 74 62 69

IV 57 57 59 51

Transfer the result into a standard table in preparation for analysis.

Table 2. Effect of hormones on weight gain in pigs (kg/pig)

19
 Blocks
Treatment I II III IV Total
A 47 52 62 51 212
B 50 54 67 57 228
C 57 53 69 57 236
D 54 65 74 59 252
Total 208 224 272 224 928

Ho = TA +TB + TC + TD
Model : X = u + T + B + E
X = any given observation
T = treatment effect
B = Block effect
E = experimental error
u = population mean

The RCBD Analysis of Variance

ANOV Table
Source of Degree of Sum of Mean
Variance Freedom Squares Squares F-ratio
Treatment t-1 TrSS TrMS TrMS/EMS
Block b1 BSS BMS BMS/EMS
Error (t 1)(b 1) ESS EMS
Total bt - 1 ToSS

Procedures (from data in Table 2)

1. Compute Treatment Totals as before.
2. Compute Block Totals.
BI = 208
BII = 224
BIII = 272
BIV = 224

3. Compute Grand Total as before.

4. Compute Correction Factor as before.
(GT)2
CF =
N
CF = 53824
5. Compute Total Sum of Squares = ToSS

ToSS = [472 + 522 + 622 + 512 + +592] CF

= 54678 53824 = 854

ToSS = 854

6. Compute Treatment Sum of Squares = TrSS

2 2 2 2
TrSS = TA + TB + TC + TD - CF
r
20
 (212)2 + (228)2 + (236)2 + (252)2
= - 53824
4

= 54032-53824 = 208
TrSS = 208

7. Compute Block Sum of Squares

BI2 + BII2 + BIII2 + BIV2

BSS = - CF
t

2 2 2 2
= (208) + (224) + (272) + (224) - 53824
4
BSS = 576

8. Compute Error Sum of squares

ESS = ToSS TrSS BSS

= 854 208 576

ESS = 70

9. Accomplish ANOV Table

Source of Degree of Sum of Mean F-ratio F

Variance Freedom Squares Squares .05 .01
Treatment 3 208 69.30 8.91 ** 3.86 6.99
Block 3 576 192.00 24.69 ***
Error 9 70 7.78
Total 15 854

Since Fc > F .05, conduct mean separation using Least Significant Difference (LSD)

D = 63 C = 59
B = 57 A = 53

Mean Comp Difference Significance

D vs C 4 n.s.
vs B 6 sign.
vs A 10 sign.
21
 C vs B 2 n.s.
vs A 6 sign.
B vs A 4 n.s.

LSD = t.05 2EMS

4

= 2.262 2(7.78)
4

= (2.262) (1.97)
LSD = 4.46

Interpretations
1. Treatments C & D increased weight gain significantly compared to the control.
2. There was no significant difference between the weight gain of treatments C and D.
3. Treatment D was better than Treatment B.
4. There was no significant difference between Treatments B and Treatment C; likewise
with Treatment A and Treatment B.

Advantages of RCBD

1. If blocking is effective the variability within block will be less than the variability
between blocks.
2. Variability among blocks does not affect the treatment differences.
3. RCBD is often more precise.

C. Latin Square Design

The Latin Square design involves a layout or arrangement in which treatments occur only
once in every row and once in every column in response to two directional gradients. Since
its main feature calls for the number of rows being equal to the number of treatments and to
the number of columns, its use is limited to 4-8 treatments. The design is able to remove
probable experimental errors bought about by the directional gradients.

S
uLet us go back to our CRD example and consider the floor plan and site of the pens.
n
l 1 2 3 4
i
g
h 22
t
 5 6 7 8

9 10 11 12

13 14 15 16

Slope Drainage

Suppose we have reason to believe that the direction of sunlight will cause some heating
differentials on the pens and could affect sanitation. The same could be said about drainage
especially when the daily chore of cleaning is considered. So we decide using the Latin square.

Randomization

The common practice in Latin square is to adopt a systematic randomization procedure.

1. As in RCBD, write out the treatments on a piece of paper each A, B, C, D. Fold

paper and place in container.
2. Draw lots.
Assign first drawn to pen 1, the second to pen 2, the third to pen 3 and the last to pen 4. Suppose
that the drawn lots correspond to B D A C.
B D A C will now be base of the next step.
Assign D to pen 5, A to pen 6, C to pen 7 and B to pen 8.
Assign A to pen 9, C to pen 10, B to pen 11 and D to pen 12.
Assign C to pen 13, B to pen 14, D to pen 15 and A to pen 16.

B D A C

23
 D A C B

A C B D

C B D A

The above readily complies with the requirement that a treatment should occur only once
in a particular row in a particular column.

Given the design, we can administer the treatments as before:

A Control
B Hormone 1
C Hormone 2
D Hormone 3

After 100 days get the weight of the pigs

50 54 47 57

65 52 53 54

62 69 67 74

57 57 59 51

Transfer the observation on standard tables prior to analysis.

Table 3. Effect of hormones on weight gain in pigs (kg/pig)

Columns
Rows 1 2 3 4 Total
1 50 54 47 57 208

24
 2 65 52 53 54 224
3 62 69 67 74 272
4 57 57 59 51 224
Total 234 232 226 236 928

Replicates
Treatment I II III IV Total
A 47 52 62 51 212
B 50 54 67 57 228
C 57 53 69 57 236
D 54 65 74 59 252

Ho = TA = TB = TC = TD

The model for analysis

X=u+T+R+C+E

X = any given observation

T = treatment effect
R = Row effect
C = column effect
E = Error effect

The Latin Square analysis of variance

Source of Degree of Sum of Mean F-ratio F

Variance Freedom Squares Squares .05 .01
Treatment t-1 TrSS TrMS TrMS/EMS
Row r-1 RSS RMS RMS/EMS
Column c-1 CSS CMS CMS/EMS
Error (r-1)(c-2) ESS EMS
Total rc-1 ToSS

Procedures (from Table 3 data)

1. Compute Row, Column and Treatment totals.
2. Compute Grand Total; GT = 928
3. Compute Correction Factor; CF = 53824
4. Compute Total Sum of Squares
ToSS = 854 (as before)
5. Compute Treatment Sum of Squares
TrSS = 208 (as before)

6. Compute Column Sum of Squares (CSS)

2 2 2 2
CSS = (234) + (232) + (226) + (236) - CF
4

= 53838 53824
25
 CSS = 14

7. Compute Row Sum of Squares

2 2 2 2
RSS = (208) + (224) + (272) + (224) - CF
4
= 54400 - 53824

RSS = 576

8. Compute Error Sum of Squares

ESS = ToSS TrSS RSS CSS

= 854 208 576 14
ESS = 56

9. Accomplish ANOV Table

Source of Degree of Sum of Mean F-ratio F

Variance Freedom Squares Squares .05 .01
Treatment 3 208 69.33 7.43* 4.76 9.78
Row 3 576 192.00 20.58***
Column 3 14 4.67 0.50
Error 6 56 9.33
Total 15 854

Since Fc > F .05, conduct mean separation using Least Significant Difference (LSD)

LSD = t.05 2EMS

4

= 2.447 2(9.33)
4

= (2.447) (2.1599)
LSD = 5.29

Mean Comp Difference Significance

D vs C 4 n.s.
vs B 6 sign.
vs A 10 sign.
C vs B 2 n.s.
vs A 6 sign.
B vs A 4 n.s.
26
 Comparison on the Basic Designs

1. Errors that can be detected

CRD All extraneous errors are lumped in the ESS.
RCB Separates BSS from ESS.
LS Separates RSS and CSS from ESS

2. Error Sum Of Squares (as in the example)

CRD 646 with 12 d.f.
RCBD 70 with 9 d.f.
LS 56 with 6 d.f.

3. Error Mean Squares

53.83
7.78
9.33

27