You are on page 1of 232

RESEARCH METHODOLOGY

(SFM 603)

Dr. Ambika P. Gautam


Research - What is it?
Research is a systematic scientific inquiry on an
issue of importance.
It generally involves application of systematic
method to:
(i) find solution to a problem.
(ii) investigate relationships among natural
phenomenon.
(iii) investigate scientific theories and
hypotheses.
Gautam/KAFCOL/2018 2
Objectives of Research

• To find out the truth, which has not yet been


found out (i.e. to fill the gap in existing
knowledge).
• Discover new information or relationships, or
to expand and verify the existing knowledge.
• To find answers to questions through the
application of scientific procedures.

Gautam/KAFCOL/2018 3
Philosophy of Research

• Human beings are curious to know something


new or the effects of various factors.
• Research is necessary for development. Or, in
other words, development is not possible until
the “truth” is known.

Gautam/KAFCOL/2018 4
How to know the “truth”?
1) Traditional sayings/beliefs/teaching
 Sound like they are the exact truth (whether
or not in fact they are!!)
 Such knowledge is stubborn (unwilling to
change; difficult to remove or deal with)
 Difficult to control
 No scientific way to test or verify

Gautam/KAFCOL/2018 5
How to know .. (cont.)
2) Authority (the right to command or control
other people).
• Some well respected or authoritative
persons/sources (e.g. religious books) say so
and it is taken as the truth.

Gautam/KAFCOL/2018 6
How to know .. (cont.)
3) Intuition
• It is the ability to understand something
instinctively, without the need for conscious
reasoning.
• An automatic, effortless feeling that often
quickly motivates you to act.
• A higher form of instinct or intelligence.
The conclusions differ among different people.
Gautam/KAFCOL/2018 7
How to know .. (cont.)
4) Scientific Method (systematic and organized
effort of knowing the “truth”). Science believes that:
• Truth are not ascertained by beliefs
• Ultimate conclusion from every individual shall be
the same
• Values, attitudes and emotions have no place in the
search for truth
• The search for truth is an ongoing and ever growing
process
• Requires expertise and resources
Gautam/KAFCOL/2018 8
Demands a Scientific Research Must Meet

1) Examine a significant issue.


2) Contribute something new to the level of
knowledge.
3) Have a serious purpose (i.e. the reason for
which it is done).
4) Follow the scientific method.

Gautam/KAFCOL/2018 9
Basic Characteristics of Research

Clear purpose  directed towards solving a


problem.
Rigor/systematic/carefully designed  based
on good theoretical base and sound
methodology; carefully designed.
Replicability  the same results under the
same design parameters.

Gautam/KAFCOL/2018 10
Basic Characteristics …(cont.)

Research is logical, objective and attempts to


minimize bias
Objective  conclusions based only on the
data/results.
Unbiased  results/conclusions must be free
from investigator’s own view or wishes.
Predictability (future forecasting)  results
can be predicted with sufficient level of
accuracy.
Gautam/KAFCOL/2018 11
Basic Characteristics .. (cont.)

Research requires expertise. The researcher


knows what is already known about the
problem under investigation.
Research involves accurate observation and
description.
Data/findings are carefully recorded and
reported.
Simplicity  in explaining the phenomenon
or problem; clear assumptions.
Gautam/KAFCOL/2018 12
Research Types
1. Deductive vs. Inductive
2. Quantitative vs. Qualitative
3. Basic vs. Applied
4. Social vs. Biophysical
5. Descriptive vs. Analytical

Gautam/KAFCOL/2018 13
Deductive vs. Inductive
This categorization is based on research
approach
• Deductive approach  to verify or test
theories for solving existing problems.
(theory: systematic explanation for observations
that relate to a particular behavior or aspect
of human life or natural phenomena)
• Inductive approach  to discover new facts
or theories.
Gautam/KAFCOL/2018 14
Deductive Approach to Research

Gautam/KAFCOL/2018 15
Inductive Approach to Research

Gautam/KAFCOL/2018 16
Quantitative vs Qualitative
• Quantitative research  based on quantitative
measurements. Involves systematic investigation of
quantitative properties and phenomena and their
relationships.
• Qualitative research  concerned with qualitative
phenomenon. Commonly used (but not limited) to
understand human behavior and the reasons that
govern such behavior.

Gautam/KAFCOL/2018 17
Difference Between Quantitative and
Qualitative Research

Quantitative Qualitative
Data expressed in Data expressed in words
numbers
Generally uses larger Uses small sample to get in-
sample that depth information. The
represents sample is often not
population (at least representative of the
theoretically) population
“Respondents” “Informants” “Participants”

Gautam/KAFCOL/2018 18
Difference Between .. (cont.)

Quantitative Qualitative
More structured Less structured research
research instruments instruments are used in
are used in data data collection
collection
Moving out from known Tries to find theory that
to unknown explains your data
Analysis – mostly Analysis – mostly subjective
statistical

Gautam/KAFCOL/2018 19
Difference Between .. (cont.)

Quantitative Qualitative
The research can The research can usually
usually be replicated not be replicated or
or repeated repeated
Reliability (probability of Reliability is lower (?)
getting same result in
repeated testing) is
higher (?)
Mechanical and static More flexible allowing
researcher to probe
further
Gautam/KAFCOL/2018 20
Basic (or fundamental or academic)
Research
• Conducted to (i) formulate theory (ii) increase our
knowledge of certain problems that commonly occur
in organizational setting, and find solutions.
• Objective is not finding solution to an immediate
problem at hand but to understand more about
certain phenomena and problems that occur widely.
e.g. investigating if informal institutions (local customs,
norms, values) play key role in the success of
community based forest management

Gautam/KAFCOL/2018 21
Basic Research (cont.)
• Basic research is usually undertaken by people
based in universities.
• Choice of topic and objectives are determined by
the researcher
• Usually have flexible time scales

Gautam/KAFCOL/2018 22
Applied (or Action) Research
• Aims at finding solution for an immediate problem
facing a society or organization.
• The finding can be implemented to rectify the
problem or situation (e.g. finding out ways to
improve productivity of forestland).
• Conducted to achieve client driven purpose .
• Objectives are negotiated with the client.
• Based more on real world situations - strict research
protocols (e.g. random sampling) may need to be
relaxed.
• Usually have tight/fixed time scales.
Gautam/KAFCOL/2018 23
Biophysical Research vs. Social Research

Biophysical Research Social Research


1 Fundamental elements are Concerned with society and human
physically related nature
2 Guided by biophysical laws in Guided by laws related to
natural phenomena human/societal behaviors
3 More precise/exact Less precise/exact
4 Predictability is higher Predictability is lower (because of
lower precision/exactness)
5 Objectivity is higher/more Objectivity is achieved with difficulty
easily achieved
6 Can be conducted in Difficult to create controlled
controlled conditions conditions

Gautam/KAFCOL/2018 24
Biophysical Research vs. Social Research (cont.)

• Social phenomena, which are usually ambiguous and


unique to particular society/place, play important
role in direction and depth of social research.
• Precise measurement of human behavior and social
phenomena is difficult because human beings do not
operate in controlled conditions.
• Even in such complexities, certain patterns are
discernable. Social research is based on analysis of
such patterns.

Gautam/KAFCOL/2018 25
Biophysical Research vs. Social Research (cont.)

• In many biophysical research, causal relations


between variables can be studied in controlled
conditions. Doing the same is difficult in social
research (in case of social science, controlled
condition means homogenous condition).

Gautam/KAFCOL/2018 26
Descriptive Research
Objective is to describe the state of affairs as it exists at
present (attempts to determine, describe, or
identify what is).
• Establishes associations between variables. For
example:
- how many times something occurs (i.e.
frequency)?
- what is the average number of occurrences (i.e.
central tendencies).

Gautam/KAFCOL/2018 27
Descriptive Research (cont.)
 Descriptive research cannot establish a causal
relationship between variables.
 No attempt is made to change behavior or conditions
i.e. you measure things as they are.
 Most common data collection methods: (i)
Observation, and (ii) Survey (e.g. interviews,
inventories, soil surveys etc).

Gautam/KAFCOL/2018 28
Analytical Research
• Analyzes the available facts (information, data) to
make a critical evaluation of the subject or
material.
• Attempts to establish why it is that way or how it
came to be.

Gautam/KAFCOL/2018 29
Some Other Categories of Research
One time vs time series
One time: examines (usually a wide range of)
subjects at one specific point in time. Most
commonly found in social research.
Time series: studying one (longitudinal) or a
variety of (cross-sectional) groups across time
Field based vs lab based

Gautam/KAFCOL/2018 30
Some Other Categories .. (cont.)

Primary vs Secondary
Primary  Original findings.
Secondary  Summary, collation and/or
synthesis of existing research.

Gautam/KAFCOL/2018 31
TERMS COMMONLY USED IN
RESEARCH

Dr. Ambika P. Gautam


Concepts
• Concepts are abstract elements that represent
classes of phenomena, objects or facts, within a
field of study. A concept represents certain
aspects of reality.
• Applicable to both social and biophysical science.
• Usually need to be defined before they can be
measured.
Example concepts: poverty, forest condition,
impact, benefits, livelihood, participation etc.
Gautam/KAFCOL/2018
Defining Concept - Examples
“Effects of shifting cultivation on forest condition
and livelihoods of Chepang people in …….”
Is here any concept that needs to be defined? If yes,
what are these? Why they need to be defined?
Forest condition can refer to all or any of these
parameters: i) canopy cover, ii) structure, iii)
composition, iii) biodiversity, iv) soil fertility levels
etc. So, it needs to be defined.
Livelihoods: What aspect of the livelihood? (i)
subsistence, (ii) income ..?
Gautam/KAFCOL/2018
Characteristics of Concepts
• Concepts are the basic elements of scientific research.
• Concept is symbolic and short.
• Each field of science develops its own concepts.
• Each concept communicates an enormous amount of
experience and information to the specialists.
• A scientific concept avoids multiple meanings in the
same field.
• With the development of knowledge and passage of
time, a concept may change its meaning which may be
narrowed or broadened.

Gautam/KAFCOL/2018
Characteristics of Concepts (cont.)
• A good concept should:
– be clear and definite i.e. must be precise.
– be comprehensive in formulation and
understanding.
– avoid multiple meanings in the same field.
– be based on logic and law.

Gautam/KAFCOL/2018
Categories of Concepts
• Postulated Concept: It does not possess rigid
meaning, i.e. gives different meaning
according to time and context. (e.g. mass etc).
• Intuitive Concept: It has particular, rigid
meaning which does not change in different
contexts. (e.g. blue color, Nepalese etc.)

Gautam/KAFCOL/2018
Construct
• A construct is a concept that is deliberately
invented for a special scientific purpose.
• A construct cannot be seen, heard or felt. It is
inferred from behavior.
Example: anxiety, faith, social class, prestige

Gautam/KAFCOL/2018
Hypothesis – what is it?
 A testable prediction about a relationship under
study. (the prediction may be based on an educated
guess or a formal theory).
 A research hypothesis is a predictive statement,
capable of being tested by scientific methods.
 An idea which is suggested as a possible explanation
for a particular situation or condition, but which has
not yet been proved to be correct.

Gautam/KAFCOL/2018
Hypothesis (cont.)
 A hypothesis clearly states a relationship in such a
way that it can be tested statistically and proved or
disproved.
Example Hypotheses:
1) Compost fertilizer induces plant growth.
2) Homogenous FUGs are more successful in CF
management than heterogeneous communities.
3) Reduced exposure to sunlight causes reduced rate
of growth of seedlings.

Gautam/KAFCOL/2018
Types of Hypotheses

1. Null hypothesis (HO) – tested directly. It


must be falsifiable.
2. Research or alternative hypothesis (HA) –
supported when null hypothesis is
rejected.
HO is the logical opposite of your hypothesis
i.e. HA. Rejecting the HO means accepting
the HA.

Gautam/KAFCOL/2018
Is a Hypothesis Needed?

• Not all research must have hypotheses (e.g.


some descriptive research – evolution of CF in
Nepal).
• When the relationship between variables can
be statistically tested and therefore proved or
disproved, then a hypothesis is relevant and
useful.
• In other words, hypothesis formulation is a
prerequisite to the application of statistical tests.

Gautam/KAFCOL/2018
Usefulness of Hypothesis
• Helps decide what type of data to be
collected.
• Helps determine the focus of the research.
• Provides clear direction to the investigator in
which he/she is to move.
• Can help save time, money and energy.
• Helps drawing proper and reasonable
conclusions.
Gautam/KAFCOL/2018
Sources of Hypothesis
• Scientific theory: A theory gives us the basic idea of
what has been found to be correct, and leads us to
form further generalization or corollaries (idea,
argument, fact) from it. These generalizations and
corollaries form parts of a hypothesis.
• Analogies: Sometimes a hypothesis is formed from the
analogy (similarity between two phenomena or things)
at a circumstance. A hypothesis is then formed to test
whether two phenomenon or things are similar in
other circumstances too.

Gautam/KAFCOL/2018
Sources of Hypothesis (cont.)
• Personal Experiences/Quality: Sometimes facts
are there but only a right individual sets it in right
perspective and formulates a hypothesis.
• Culture: In social studies, a hypothesis may be
formulated with the help of general pattern of
culture. The culture has a great influence upon
the thinking process of people and hypothesis
may be formed to test one or more of these
ideas.

Gautam/KAFCOL/2018
Theory
– Systematic explanation for observations.
– It is the ordering of facts (i.e. empirically verifiable
observation) in some meaningful way.
– It summarizes facts into empirical generalizations.
– Every theory can be challenged and adapted by
other researchers.

Gautam/KAFCOL/2018
Relationship between Theory and Fact

• Theories cannot be constructed without facts and


facts cannot be understood without theories.
• While facts are concrete (i.e. definite, specific),
theories are abstraction (i.e. general idea, pattern).
• While facts are wide and scattered, theory is brief
and limited.
• In the scientific process of research, facts are
gathered first and then ordered through theories (i.e.
they provide input towards construction of theories).

Gautam/KAFCOL/2018
Variables
• Characteristics by which persons or objects
can be described.
• Variables can change from time to time for
given persons or objects; from person to
person; or from object to object.
Example social variables: age, gender,
occupation, social class etc.
Example biological variables: tree height, DBH
etc.
Gautam/KAFCOL/2018
Types of Variables
• Continuous Variable: Variable that can take
any values within a certain range. (e.g. age,
height, weight).
• Discrete Variable - Can only take on specific
values (e.g. sex – male or female; marital
status - married, single, divorced, widowed).
The categories are often assigned numerical
values (i.e. dummy variables) for use as labels
in analysis, e.g., 0 = male; 1 = female.

Gautam/KAFCOL/2018
Types of Variables (cont.)
• Independent Variable: The presumed cause.
• Dependent Variable: The presumed effect.

Gautam/KAFCOL/2018
Measurement of Variables – Scale of
Measurement
• Different measurement techniques are applied to
measure different types of variables.
1. Nominal scale
• Numbers (or symbols) are given to various objects
just to distinguish one object from other. The
numbers used have no numerical meanings and they
cannot be added or ordered.
Examples:
– Male, Female
– Hindu, Muslims, Christian, Buddhist
– CFOP exist, CFOP does not exist
– Dry season, wet season
Gautam/KAFCOL/2018
Measurement of Variables .. (cont.)
2) Ordinal (Ranking) Scale
• Ordinal values show the order in which things occur
within the range of scale.
• The numbers assigned to groups show one is more
(higher) than other but not how much greater.
• Examples:
– Low, medium, high, very high
– Infant, child, adult
– Not serious, serious, very serious
– Grades A, B+, B, C+, D
Gautam/KAFCOL/2018
Measurement of Variables .. (cont.)
• In ordinal scale, the numeric values indicate
relationship in terms of sequence or order, so the
numbers in this scale have more meaning than
nominal values.
• The ordinal values do not show the interval between
the values.

Gautam/KAFCOL/2018
Measurement of Variables .. (cont.)
3) Interval Scale
• Any measure, in general, can have a unique value.
• The numeric values show both order and magnitude.
• May not include the value of zero, or when they do,
the zero value does not really indicate the complete
absence of the thing measured.
– e.g. a temperature of zero degrees Celsius does not
represent the complete absence of heat
• The values in interval scale cannot be used to form
ratios.
Examples: number of trees; number of species; number
of people etc.

Gautam/KAFCOL/2018
Measurement of Variables .. (cont.)
4) Ratio Scale
• Similar to interval scale except that zero is absolute
in ratio scale. (i.e. a zero value does indicate
absence of the thing to be measured).
Example:
– distance in kilometers
– difference in age in years

Gautam/KAFCOL/2018
Nominal, Ordinal, Interval, and Ratio
Scales: a Comparison
Level of Clear Defined Defined unit Meaningful
measure catego order of zero,
ment ries measure ratios
ment
Nominal Yes No No No
Ordinal Yes Yes No No
Interval Yes Yes Yes No
Ratio Yes Yes Yes Yes

Gautam/KAFCOL/2018
Reliability
• Reliability means capable of being relied on;
dependable.
• It is the extent to which a questionnaire, test,
observation or any measurement procedure
produces the same results on repeated trials.

Gautam/KAFCOL/2018
Reliability (cont.)
• Reliability of an experiment is whether you could
replicate an experiment and get comparable results.
• Reliability of a statistical test means yielding the
same or compatible results in different statistical
trials.
• Reliability of a measuring instrument is its ability to
measure consistently (i.e. with same or similar
values) and accurately (i.e. true measure) the
phenomena as it is designed to measure.

Gautam/KAFCOL/2018
Validity
• Validity refers to whether a study measures or
examines what it claims to measure or examine (i.e.
are you measuring what you intended to measure?).
• Reliability is necessary but not sufficient for validity.
A measure can be perfectly reliable and yet not be
valid (i.e. for something to be valid it must be reliable
but it must also measure what it is intended to
measure).

Gautam/KAFCOL/2018
Accuracy and Precision
Accuracy refers to the success of estimating the true
value i.e. closeness of an estimated value to the
true value.
(how close the sample mean is to the true or
population mean)
Example: if the true height of a tree is 10 m, then 9 m is
more accurate than 8 m.

Gautam/KAFCOL/2018
Accuracy and Precision (cont.)
• Precision refers to level of concentration of
estimated values to the true value or target. When
the distribution of estimated value is narrowly
concentrated around the target value, precision is
high, and when they are widely spread out around it
then precision is low.
• It is concerned with how finely you can measure
something of interest.
Example: If you are measuring and recording tree
diameter then 10.54 is more precise than 10.5.

Gautam/KAFCOL/2018
Accuracy and Precision (cont.)

High accuracy, but High precision, but


low precision low accuracy

Gautam/KAFCOL/2018
Bias
• Bias is a systematic distortion. It may be due to :
– flaw in measurement,
– flaw in the method of selecting sample, or
– flaw in the technique of estimating a
population parameter.
– personal bias

Gautam/KAFCOL/2018
Bias (cont.)
• Measurement bias examples:
– measuring diameter of trees with a tape from
which the first few cm has been removed
(broken, damaged).
– trees present at the plot boundary – some
researchers will nearly always include in the
count others will consistently exclude them.
Measurement biases can be minimized by continuous care
and check of instruments, good training.

Gautam/KAFCOL/2018
Bias (cont.)
Sampling bias:
• May arise when certain units of population are
given a greater or lesser representation in the
sample.
• Can be minimized by randomly selecting the
samples.

Gautam/KAFCOL/2018
Bias (cont.)
 Bias related to the technique of estimating a
population parameter
Examples:
– Calculating average income of households in a
village with unequal income using simple
average technique.
– Calculating average survival of planted trees in
two plots of different size using simple average
technique.
A better method in both the cases would be to
take weighted average.
Gautam/KAFCOL/2018
Bias (cont.)
Professional bias:
• Some researchers may try to “prove” their
preconceived notions through research (because
they take it as their so called obligation to prove
the outcomes positive – e.g. some people working
with some agency – NGO, govt.).
• Such bias can be introduced while selecting
samples, constructing survey questions (by
preparing “leading” or “framed” questions).

Gautam/KAFCOL/2018
Research Design

Dr. Ambika P. Gautam


Introduction
• Research design is the plan or arrangement of
methods and procedures to be used in the
research.
• It is a detailed blueprint used to guide the
conduct of research so that the research
questions are answered and the research
objectives are met.

Gautam/KAFCOL/2019 2
Introduction (cont.)
• A research design:
– provides the researcher a clear framework or outline for
addressing the research questions
– dictates boundaries of the research
– enables the researcher to expect potential problems in
the implementation of the study
• Research design is necessary to ensure that:
– the data collected are relevant to the research problem
and collected with optimum cost.
– the research is conducted efficiently.
– the reliability and validity of the data and procedure is
not questioned. Gautam/KAFCOL/2019 3
Research Types

Research

Exploratory Conclusive

Descriptive Experimental

Gautam/KAFCOL/2019 4
Exploratory Research Design
• Used to gather preliminary information that will help:
(i) define the problem and suggest hypotheses
(ii) formulation of the alternative courses of action to
address the problem
• Exploratory design is appropriate when the researcher is
involved in investigating a subject in which he/she has no
sufficient knowledge to formulate the hypothesis about
the problem.

Gautam/KAFCOL/2019 5
Exploratory Research Design (cont.)
• Common exploratory research methods
include:
– Expert surveys
– Pilot surveys
– Case studies
– Secondary data/information
– Qualitative research

Gautam/KAFCOL/2019 6
Characteristics of Exploratory
Research Design
• Objective is to provide insights and
understanding.
• Information need is defined only loosely.
• Research process is flexible and unstructured.
• Sample is usually small and non-representative.
• The results are tentative.
• Generally followed by further exploratory or
conclusive research.

Gautam/KAFCOL/2019 7
Descriptive Research Design
• Descriptive research design involves observing
and describing the behavior of a subject without
influencing it in any way.
• It is used when the objective is to provide a
systematic description that is as factual and
accurate as possible.
• One of the most widely-used research design
(particularly by students).

Gautam/KAFCOL/2019 8
Characteristics of Descriptive
Research Design
• Preplanned and structured design.
• Descriptive research establishes associations
between variables, but cannot establish causal
relationship between them.
• No attempt is made to change behavior or
conditions i.e. you measure things as they are.
• Most common methods for collection of
primary data include: (i) Observation, and (ii)
Survey (e.g. interviews, inventories).
Gautam/KAFCOL/2019 9
Experimental Research Design
• Objective is to determine cause and effect
relationships.
• Allows the researcher to manipulate a specific
independent variable in order to determine what effect
this manipulation would have on other dependent
variables.
• The experiment can take place in a laboratory
("artificial" setting) or in the field ("natural" setting).
• In an experimental study, you take measurements, try
some sort of intervention, then take measurements
again to see what happened.
Gautam/KAFCOL/2019 10
Experimental Studies with Control Group

• Experimental studies directly test hypothesis.


• In these designs, all subjects are measured, but only
some of them--the experimental or treatment
group--then receive the treatment.
• Control group (the group of subjects to which the
treatment is not applied) is allowed to remain
undisturbed and provide “baseline” data against
which the effect of the treatment is gauged.

Gautam/KAFCOL/2019 11
Some Basic Points to Remember While
Designing an Experiment
• What is your hypothesis?
• Why do you think your research hypothesis will be
supported?
• What is your independent variable? How will you
operationalize it?
• What is your dependent variable? How will you
operationalize it?
• How would you determine whether a “participant”
was in the treatment or no-treatment group?

Gautam/KAFCOL/2019 12
Three Minimum Requirements for
Experimental Research
1) Knowledge of initial conditions
• Are initial conditions similar or different?
• What is the previous history?

2) Adequate controls
Controls allow the researcher to examine exactly how
a population reacts to a treatment, as well as the
behavior of the population without the treatment.
With adequate controls, the researcher can be more
confident that an effect seen is the result of the
treatment rather than another environmental
factor.
Gautam/KAFCOL/2019 13
Three Minimum Requirements .. (cont.)
3) Replication
Replication is the random assignment of groups to either
the treatment or control group, and the repetition of
those groups over space (to take natural variability into
account).
Replication is important for the following reasons:
• to reduce biases (i.e. reproducing the same conditions
for each group)
• to account for natural variability
• to avoid possible systematic differences among
site/group locations
• to decrease chance of total loss due to catastrophe
• to meet requirement for statistical comparison

Gautam/KAFCOL/2019 14
Experiment, an Example: Effect of Thinning
on Tree Growth
Control group Treatment Treatment
group 1 group 2

Plot # II VII IX IV III XI V XIII X


Trees 25 25 25 25 25 25 25 25 25
(studied)
Treatment none none none G 1 G 1 G1 G2 G2 G2
(e.g.
thinning)
Measurement
(DBH in
cm)
G = Grade
Gautam/KAFCOL/2019 15
Experimental Studies Without a Control
Group (Time Series Studies)
• Measurements are taken repeatedly before and after
an intervention.
For example, studying the impact of community
forestry on forest cover. Measurements on the forest
are taken before and after the intervention.
Time series suffer from a major problem: any change
you see could be due to something other than the
treatment.

Gautam/KAFCOL/2019 16
Sampling Techniques
Sampling techniques can be broadly grouped into two:
1) Probability Sampling: every unit in the population
has a known chance of being selected. Used in most
research.
2) Non-probability Sampling: the chance of every unit
in the population being selected is unknown. The
selection of the subjects is non-random (arbitrary or
subjective), based on the researcher’s judgment.
Non-probability samples cannot depend upon the
rationale of probability theory.
Gautam/KAFCOL/2019 17
Probability Sampling
Four main types:
1) Simple random sampling
2) Stratified random sampling
3) Systematic sampling
4) Cluster sampling

Gautam/KAFCOL/2019 18
Simple Random Sampling
 Selection at random from the population.
 Each individual or location have the same
chance of being chosen as every other
individual or location.

Gautam/KAFCOL/2019 19
Stratified Random Sampling
 The units of the population are grouped together
on the basis of similarity of some characteristics.
 Random sampling is applied within groups of
population.
 The purpose of stratified random sampling is to
reduce the variability in the overall study
population by breaking it into groups that are
more homogeneous than the overall study
population.

Gautam/KAFCOL/2019 20
Stratified Random Sampling (cont.)
In sampling a human population, we may set up strata
corresponding to:
• gender – male and female
• age group
• family size
• wealth group
• rural and urban
• major occupation

Gautam/KAFCOL/2019 21
Stratified Random Sampling (cont.)
In sampling a forest, the strata could be based on:
• vegetation type
• forest origin e.g. natural and plantation
• tree size class (if distinctly different)
• governance type
• environmental factors – soil, moisture, elevation,
aspect, slope

Gautam/KAFCOL/2019 22
Sample Allocation in Stratified Random
Sampling
(A) Proportional Allocation
nh = (Nh/N) * n
Where,
nh = number of samples for stratum h
Nh = number of units in stratum h
N = total number of units
n = total number of samples

Gautam/KAFCOL/2019 23
Sample Allocation in Stratified .. (cont.)
Suppose, in an user group of 250 households you
found,
• Rich households (stratum I) = 25
• Middle income households (stratum II) = 140
• Poor households (stratum III) = 85

Number of sample households in:


Stratum I: nh = (Nh/N)*n or (25/250) * 50 = 5
Stratum II: (140/250) * 50 = 28
stratum III: (85/250) * 50 = 17
This is based on the sample size of 50 households
Gautam/KAFCOL/2019 24
Sample Allocation in Stratified .. (cont.)
Suppose, in a forest of 580 ha., you found,
• Stratum I (broadleaved) = 180 ha
• Stratum II (conifer) = 100 ha
• Stratum III (mixed) = 300 ha.

Number of sample plots in the 3 strata,


• Stratum I: (Nh/N)*n or (180/580) * 30 = 9.3 or 9
• Stratum II = (100/580) * 30 = 5.2 or 5
• Stratum III = (300/580) * 30 = 15.5 or 16

Assumption: 30 plots are enough to represent the forest


Gautam/KAFCOL/2019 25
Sample Allocation in Stratified .. (cont.)
(B) Optimum Allocation
If there are clear differences in variability among your
strata, sampling proportional to area will yield a
much less accurate estimate of your parameter.
In such a situation, optimum allocation is better.

Gautam/KAFCOL/2019 26
Sample Allocation in Stratified .. (cont.)
In optimum allocation, the units are allocated to the
strata in such a way that there is smallest standard
error possible.
nh = (Nhsh/ Nhsh) * n
Where,
sh = Standard error = Standard deviation among
estimates (s/√n) OR Square root of variance of
estimate (√(s2/n). As sample size increases, standard
error decreases.

Gautam/KAFCOL/2019 27
Sample Allocation in Stratified .. (cont.)

In the above example,


• NIsI (i.e. Nhsh in the formula) = 180 * sI = say, X
• NIIsII = 100 * sII = say, Y
• NIIIsIII = 300 * sIII = say, Z
 Nhsh = X + Y + Z
Apply these values in the above formula to get the
number of sample plots in each stratum.

Gautam/KAFCOL/2019 28
Systematic Sampling
 Involves arranging the target population according to
some ordering scheme and then selecting elements
at regular intervals through that ordered list.
 Selecting every nth case (individual, distance or
area).
 Begins at random.
Examples:
• Interviewing head of every 10th household from the
list of households.
• Measuring every 15th tree along a transect.

Sampling interval = population size/sample size.

Gautam/KAFCOL/2019 29
Cluster Sampling
• Population is divided into clusters, usually
based on geographical contiguity.
• A sample of such clusters is then selected.
• Sampling units are groups rather than
individuals.

Gautam/KAFCOL/2019 30
Cluster Sampling (cont.)
Two types of cluster sampling methods.
One-stage sampling. All of the elements within
selected clusters are included in the sample.
Two-stage sampling. A subset of elements
within selected clusters are randomly selected
for inclusion in the sample.

Gautam/KAFCOL/2019 31
Difference Between Strata and Clusters
• Although strata and clusters are both non-
overlapping subsets of the population, they differ in
several ways.
• All strata are represented in the sample; but only a
subset of clusters are in the sample.
• With stratified sampling, the best survey results
occur when elements within strata are internally
homogeneous. However, with cluster sampling, the
best results occur when elements within clusters are
internally heterogeneous.

Gautam/KAFCOL/2019 32
Non-Probability Sampling
Three main types:
1) Purposive Sampling
2) Quota Sampling
3) Convenience Sampling

Gautam/KAFCOL/2019 33
Purposive Sampling
• Handpicking typical or interesting cases.
• There must be some logic behind the selection (i.e.
we sample with a purpose in mind).
• We usually would have one or more specific
predefined groups we are seeking (targeted sample,
e.g. dalit members of a CFUG)

Gautam/KAFCOL/2019 34
Quota Sampling
• The population is first segmented into mutually
exclusive sub-groups (just as in stratified sampling).
Then judgment is used to select the subjects or units
from each segment non-randomly according to some
fixed quota.
• Two types: proportional and non proportional.
• Proportional quota sampling - you want to represent
the major characteristics of the population (e.g.
gender, age, education, religion etc) by sampling a
proportional amount of each.

Gautam/KAFCOL/2019 35
Quota Sampling (cont.)
• Non-proportional quota sampling - you specify the
minimum number of sample units you want in each
category, but you're not concerned with having
numbers that match the proportions in the
population.
• It is thus less restrictive than proportional.

Gautam/KAFCOL/2019 36
Convenience Sampling
• We have no evidence that the subjects selected are
representative of the populations we're interested in
generalizing to (and in many cases we would clearly
suspect that they are not).
• Also known as Accidental or Haphazard sampling.
• Example: interviews conducted frequently by
television news programs to get a quick (although
non-representative) reading of public opinion.

Gautam/KAFCOL/2019 37
How Many Samples Are Needed?
• Different methods are used for determining
sample size in different situation.
• Generally the following criteria should be
considered:
1) The Level of Precision: the range in which the
true value of the population is estimated to be.
This is related to sampling error the difference
between the sample and population values,
which is often estimated (rather than exact
measurement).

Gautam/KAFCOL/2019 38
How Many Samples … (cont.)
2) The Confidence Level: this is based on the
concept of normality of distribution.

Gautam/KAFCOL/2019 39
How Many Samples … (cont.)

• Concept of normality of distribution:


– When a distribution conforms closely to the standard
normal curve (many distributions do), a certain
proportion of cases will always fall within certain
ranges (determined by SD of the distribution).
– About 95% of the cases will be within ±2SD. This
means that, if a 95% confidence level is selected, 95
out of 100 samples will have the true population
value.
- Confidence Value at 95% level = 1.96

Gautam/KAFCOL/2019 40
How Many Samples … (cont.)
3) Degree of Variability in the attributes being
measured: the more variable (more
heterogeneous) a population, the larger the
sample size required.
4) Other Considerations:
(a) Variances of subpopulations or strata – if
stratified random sampling is used.
(b) Type of analysis planned – e.g. statistical tests
may need larger sample size compared to
descriptive statistics.

Gautam/KAFCOL/2019 41
Strategies for Determining Sample Size
There are several approaches:
• Using a census for small populations: eliminates
sampling error.
• Imitating a sample size of similar studies: in this
case you risk repeating errors that were made in
determining the sample size for another study.
• Using published tables
• Applying formula

Gautam/KAFCOL/2019 42
Strategies for Determining .. (cont.)
Application of Formula
Example 1 (Yamane, 1967):
n = N/[1+N (e2)]
n = sample size, N = total population, e = level of
precision
If using stratified sampling, then divide the sample size
proportionately. n1 = N1*n/N

Reference: Yamane, Taro (1967). Statistics: An Introductory


Analysis, 2nd Ed., New York: Harper and Row.

Gautam/KAFCOL/2019 43
Strategies for Determining .. (cont.)

• Example 2 (Cochran, 1977)


n = Nz2 x P(1-P)/Nd2 + z2 x P(1-P)
n = sample size
N= population size
P = proportion of sample on population estimate
d = level of precision
z = confidence level (at 95% confidence level, z = 1.96)
Reference: Cochran, W. G. 1977. Sampling Techniques, 3rd
Edition. John Willey and Sons, New York.
Gautam/KAFCOL/2019 44
Methods of Data Collection

Dr. Ambika P. Gautam


Planning the Field Work: Some Considerations

• Prioritize the activities.


– Is one phase of the research dependent on the
completion of another phase?
– How much time will it take for the various
activities?
• Are you relying on other people’s (e.g. local
people, field assistants, government
staff/offices, advisor etc) capacity? If so you
must realize its possible implications for your
work.

Gautam/KAFCOL/2019
Planning the Field Work: Some .. (cont.)

• Improving/maximizing reliability is crucial,


particularly with local people and assistants.
How is this done?
– Incentives
– Participation
– Information
– A good time (?)
– ………..

Gautam/KAFCOL/2019
Planning the Field Work: Some .. (cont.)

Be prepared for the “inevitable” !!!!


• Some thing may go wrong in the field – need to
have a “contingency plan” for that.
– Know the interconnections of various activities
in your research, and what you can
drop/change without changing the whole
research project.
– Keep in contact with your advisor.

Gautam/KAFCOL/2019
Data Collection and Organization

• Be organized
– have a list of goals for each week or day.
– equipment, notebooks, and other accessories
should be sufficient.
• Organize a good field team
– selection of members
– evaluation of their capacity
– specify duties and delegate authority for each
member
– conducive environment for team work

Gautam/KAFCOL/2019
Data Collection and .. (cont.)

• Collect data in the field using the same framework


you used to design your research questions
– beginning with general to specific.
– field notes – as much information as possible
related to your research.
– where appropriate, make drawings and label
various features in your study site.

Gautam/KAFCOL/2019
Sketch Map Example (Source:
IFRI Data Collection Manual)

Gautam/KAFCOL/2019
Data Collection and .. (cont.)
• Review your notes after every field session or end
of the day
– check for any incomplete/unclear writing.
– add any observation that you were unable to
make notes in the field.
– discuss with your team members if you are not
sure whether your note on a particular aspect of
the survey is correct. Did other team member(s)
noted/observed the same or different?
• Make copies of raw data
Gautam/KAFCOL/2019
Social Methods of Data Collection -
Commonly Used Tools
1. Questionnaire Surveys and Interviews
2. Observation
3. Group meeting
4. Rapid Rural Appraisal (RRA) and
Participatory Rural Appraisal (PRA)

Gautam/KAFCOL/2019
Questionnaire Surveys and Interviews
• Involves collection of primary data, usually by
selecting a representative sample of the
population under study.
• Common types include:
• Unstructured interviews
• Semi-structured interviews
• Structured interviews
• Key-informants interviews

Gautam/KAFCOL/2019
Unstructured Interviews

• Interviewer simply facilitates the subject talking at


length.
• Prompts, such as photographs, can be useful for
stimulating discussion.
• Recording of the interview can be done in a variety
of ways, such as:
– taping
– taking notes during or after the interview
– one taking notes while the other conducts the
interview

Gautam/KAFCOL/2019
Advantages of Unstructured Interviews

• Respondents can answer questions in as much


detail/in-depth as they want.
• More valid information about respondents’ values,
attitude, and opinions can be obtained, particularly
how people contextualize and explain these issues.
• The informal atmosphere can encourage the
respondent to be honest and open.
• It is flexible method – the interviewer can adjust
questions and change direction as the interview is
taking place.

Gautam/KAFCOL/2019
Disadvantages of Unstructured Interviews

• Comparison of data will be difficult because each


respondent can have his/her unique view
• Can take long time (which means less number of
interviews in the available time)
• The sample may not represent the population
(because of small sample size)

Gautam/KAFCOL/2019
Structured Interviews or Questionnaire Surveys

• One of the most widely employed techniques in


social research (mainly, in quantitative research).
• Questionnaire can be administered in a number
of ways - face-to-face; sending by post to the
intended respondents; over the telephone.
• In case of face-to-face or telephone, the
interviewer reads the questions exactly as they
appear on the survey questionnaire.
• Questionnaire can be close-ended or open-
ended.

Gautam/KAFCOL/2019
Structured Interviews .. (cont.)
• Questions can be asked in a number of ways:
– Simple information – e.g. what is your main
occupation?
– Category
– Multiple choice
– Scale
– Complex grid or table
– Open-ended etc.

Gautam/KAFCOL/2019
Structured Interviews .. (cont.)
• Categorical question: example
Q. Do you collect firewood from forest?
Yes/No
• Multiple choice question: example
Q. What do you think about the general trends of changes in the
level of soil erosion over the last 20 years?
(a) Increasing
(b) Decreasing
© No Change
(d) Don’t know

Gautam/KAFCOL/2019
Structured Interviews .. (cont.)
• Scale question: example
Q. How do you rate the problem of human-wildlife
conflict in your area?
(a) Very severe
(b) Moderately severe
© Not severe
(d) Negligible

Gautam/KAFCOL/2019
Structured Interviews .. (cont.)
• Complex question – example
Q. How do you best describe the supply of following
forest products in your area at present?
Availability
Surplus Adequate Shortage Critical
Product shortage
Firewood
Fodder
Timber

Gautam/KAFCOL/2019
Advantages of Structured Interviews

• Can reach a large sample (because of relatively less


time consumed in each interview).
• Results more likely represent the population
(because of large sample size)
• Comparison is possible/easier (because questions
are structured and asked in the same way)
• Can ensure understanding of the question by the
respondent.

Gautam/KAFCOL/2019
Disadvantages of Structured Interviews

• Respondents are forced to choose between the


alternative answers the interviewer gives them.
• It will be difficult to obtain data on attitudes,
opinions and values.
• Rigid – interviewer has to stick to the pre-
prepared/agreed questions even though interesting
lines of inquiry might emerge in an interview.

Gautam/KAFCOL/2019
Advantages of Interview Surveys

• Allows for standardization and uniformity making


comparisons easier (particularly, in case of structured
interviews).
• Ensures higher reliability than some other
techniques.
• Results can be provided relatively quickly.
• Relatively inexpensive.

Gautam/KAFCOL/2019
Disadvantages of Interview Surveys

• Response error or bias - the information provided


may not be valid.
The respondent may wish to impress the researcher
(e.g. by attributing her/himself a higher income or
education level) or please her/him by providing the
kind of response s/he believes the researcher is
looking for.

Gautam/KAFCOL/2019
Disadvantages of Interview Surveys (cont.)

• Willingness or ability to reply -


– the information could be considered sensitive or
intrusive (e.g. information about household
income) leading to a high rate of refusal.
– the question is so specific that the respondent is
unable to answer, even though willingness is
there.

Gautam/KAFCOL/2019
Disadvantages of Interview Surveys (cont.)

• Interviewer error or bias- interviewer can influence


the response or introduce bias through:
– comments made or by stressing certain words in
the question itself.
– facial expressions or body language.

Gautam/KAFCOL/2019
Disadvantages of Interview Surveys (cont.)

• Response rate – can be influenced by many factors:


– the length of the questionnaire,
– the type and/or motivation of the respondent,
– the type of questions and/or subject matter, the
time of day or place,
– Incentive.
Proper questionnaire design and question wording can
help increase response rate.

Gautam/KAFCOL/2019
Caution!!
• Choice of appropriate words is important in
interview surveys.
Try to avoid:
– ambiguous questions
– questions requiring specialist knowledge
– questions which may be offensive or sensitive

Gautam/KAFCOL/2019
Observation
• The researcher may or may not have direct
contact or communication with the subject
(people or thing) under investigation.
• Data can be collected by human or electronic
means.

Gautam/KAFCOL/2019
Observation Methods
 Participant and nonparticipant observation:
depending on whether the researcher chooses to be
part of the situation s/he is studying.
 Obtrusive and unobtrusive observation: depending on
whether the subjects being studied can detect the
observation (e.g. hidden microphones or cameras
observing behavior)
 Structured and unstructured: whether guidelines or a
checklist being used for the aspects of the behavior
that are to be recorded.
 Direct and indirect: depending on whether the
behavior is being observed as it occurs or after the fact
(e.g. video records).

Gautam/KAFCOL/2019
Advantage and Disadvantage of Observation

• Advantage: records actual behavior.


• Disadvantage: does not provide us with any
insights into what the person may be thinking or
what might motivate a given behavior/comment.
Ethical issue – observing people when they are not
aware of it (e.g. use of cameras and
microphones) can be considered an invasion of
privacy or abusive.

Gautam/KAFCOL/2019
Key-informant Interviews

• Interviewing “experts”.
• A simple checklist can be useful.

Gautam/KAFCOL/2019
Group Meeting

• Coming together of people for specific purpose.


• Can involve a large number of people or smaller
(e.g. focus group or people with common concerns
– women, herders, carpenters, blacksmiths).
• Meetings generally have a facilitator who
encourages communications.

Gautam/KAFCOL/2019
Group Meeting (cont.)

• Major advantages of Group Meeting:


– A large number of people can be reached in a
relatively short period of time.
– Open invitations open possibility of participation
by all interested.
– Can help bring problems/concerns of
“peripherals” on the “surface”.
– Help get community insight into the “root” of the
problem.

Gautam/KAFCOL/2019
Group Meeting (cont.)
• In Group Meeting:
– Careful planning and a clear purpose of the
meeting are necessary.
– Beware of hidden agendas – some individuals or
the group might use the meeting to bring up their
own concerns – role of facilitator becomes
important here

Gautam/KAFCOL/2019
Rapid Rural Appraisal (RRA) and
Participatory Rural Appraisal (PRA)
 A family of approaches and methods.
Some common PRA methods:
• Participatory analysis of secondary sources, e.g.
maps, aerial photographs, satellite images
• Participatory mapping
• Transect walks
• Time lines and trend and change analysis –
chronologies of events with dates
• Oral histories – e.g. of a crop, a tree species, a
weed.

Gautam/KAFCOL/2019
RRA and PRA (cont.)
• Seasonal calendars – e.g. distribution of rain, crops,
food type, migration, animal fodder
• Venn diagram – e.g. to identify important
institutions or individuals in the community
• Matrix scoring and ranking – e.g. comparing
different tree species in terms of their value to the
community.
• Probing – e.g. what species of seedlings they like to
plant?
etc.

Gautam/KAFCOL/2019
Main Difference Between RRA and PRA

RRA PRA
Extractive Facilitating
Outsiders learning from Outsiders facilitate to
local people’s enhance local people’s
knowledge analytical capability
Ultimate outcomes – Ultimate outcomes -
plans, publications Sustainable local action
and institutions

Gautam/KAFCOL/2019
Biophysical Methods of Data Collection –
some commonly used tools

1 Forest Inventory
2. Soil Survey
3. Experiments (e.g. nursery research)
4. Wildlife research

Gautam/KAFCOL/2019
Forest Inventory – International Forestry
Resources and Institutions (IFRI) Method

Gautam/KAFCOL/2019
IFRI Method

• Randomly selected plots composed of three


concentric circles of 1-meter, 3-meter, and 10-
meter in radius
• In the innermost circle (i.e. 1-m in radius),
woody seedlings and herbaceous ground
cover is recorded.

Gautam/KAFCOL/2019
IFRI Method
• In the next circle (3-m in radius), shrubs, saplings,
and climbers are identified and counted and also
the diameter at breast height (DBH) and heights
of woody stems between 2.5 and 10 cm in
diameter was recorded.
• In the largest circle (10-m in radius), stems of
equal or greater than 10 cm in DBH are identified,
counted and DBH and height measured

Gautam/KAFCOL/2019
DBH Measurements
(Source: IFRI Data
Collection Manual)

Gautam/KAFCOL/2019
IFRI Method
• Each forest plot is also described in terms of
its topography (elevation, slope, aspect) and
soil condition (soil depth, color, texture,
drainage).

Gautam/KAFCOL/2019
Soil Surveys
Depends on your research question.
May involve analyses of humus layer, depth of A and
B horizons, texture, color, drainage, and/or nutrients
analyses. Analysis can be done in the field using
simple techniques or in laboratory.

Gautam/KAFCOL/2019
INSTRUCTIONAL DIAGRAM FOR
DETERMINING SOIL TEXTURE BY FEEL
(Source: IFRI Manual)

Gautam/KAFCOL/2019
Secondary Data: Common Sources

• Office records
• Published materials (journal articles, books, etc)
• Management plans
• Maps
– topographic
– thematic
• Aerial photographs – B/W, color
• Satellite image – digital, hardcopy
• Digital GIS data
• Internet

Gautam/KAFCOL/2019
DATA ANALYSIS AND
INTERPRETATION

Dr. Ambika P. Gautam


Classification and Organization of Data

• Classification of data is the process of


organizing data into groups according to various
parameters (such as scales of measurement)
• Data organization refers to the method of
classifying and organizing data sets to make them
more useful. It can involve:
– Organizing all data forms/questionnaire in one place
– Checking completeness and accuracy
– Assigning unique identifier as per the need
etc

Gautam/KAFCOL/2019
Data Analysis
• Analysis can be qualitative or quantitative depending
upon the nature of your data.
• Quantitative analysis involves turning raw numbers
into meaningful data or information through the
application of rational and critical thinking. It
generally involves the following:
– Preparing and checking the data.
– Input of data into computer.
– Selecting appropriate tables and diagrams to use according
to your research objectives.
– Selecting the most appropriate statistics to describe,
examine relationships and trends in your data.
Gautam/KAFCOL/2019
Data Analysis (cont.)

• Usually, statistical software is used in analysis of


quantitative data.
• Excel Spreadsheet, SPSS, and/or Microsoft Access are
some of the commonly used software.
• Quantitative analysis can be descriptive (mean,
median, mode, SD, variance, percentage,
frequencies, ranking, cross-tabulation etc) or
statistical tests.

Gautam/KAFCOL/2019
Statistical Tests
• Can be grouped into two broad types:
1) Parametric
2) Non-parametric
• Choice of test is mainly determined by:
– kind of data that you have collected and its
distribution
– research objective(s)

Gautam/KAFCOL/2019
Parametric Tests
• We use parametric tests for interval and ratio
scale data (continuous measures), when the
following assumptions are considered valid:
– Independence of data within and among
samples.
– Homogeneity of variances (i.e. dispersion
around the mean is similar).
– Normal distribution of the data.
• Some common parametric tests include: t-test,
ANOVA, Correlation, Regression.

Gautam/KAFCOL/2019
t-test
• A technique used to measure the statistical
significance of the difference between the means
of two populations.
• HO: there is no difference between the sample
means of two populations.
Note: although normality of data is a requirement,
the t-test does not fail if the data are somewhat
non-normal.

Gautam/KAFCOL/2019
Analysis of Variance (ANOVA)
• A technique used to measure the statistical
significance of the difference between the means
of more than two populations.
• HO: the means of the three (or more) groups are
equal.
Note: ANOVA only responds with the probability
that there is statistical difference among the
means. It does not respond as to specifically
which means are different.

Gautam/KAFCOL/2019
Correlation
• Measures the degree of association between two
variables (i.e. the degree to which two variables vary
together).
(A) Pearson’s Correlation: the coefficient is a measure of
linear association.
HO in Pearson’s Correlation: there is no linear relationship
between the variables.
(B) Spearman Rank Correlation: this is more appropriate if
you don’t know whether there is a linear relationship
among the variables.
HO in Spearman Correlation: there is no uniformly positive
or negative relationship between the variables.

Gautam/KAFCOL/2019
Regression
• Regression expresses the value of one variable (Y) in
terms of a function of another variable (X).
• A bivariate linear regression analysis will give you the
correlation information, but it will also calculate the
equation of the straight line that best fits the data – the
regression line.
Y = a + bX, where Y is the dependent variable, X is the
independent variable, ‘a’ is the intercept (the value at
which it crosses the Y axis), and ‘b’ is the slope of
regression line (regression coefficient).
(e.g. you can try to predict the height of a tree by knowing
its DBH.)

Gautam/KAFCOL/2019
Correlation vs. Regression
• In Correlation,
– we are concerned with whether the two variables are
interdependent and vary together
– we do not express one variable as a function of
another
– there is no distinction between dependent and
independent variables
– there is no prediction
• In Regression,
– we intend to describe the dependence of variable Y
on the independent variable X
– we can predict the dependent variable based on the
value of independent variable
Gautam/KAFCOL/2019
Non-parametric Tests
• Nonparametric tests do not require assumptions
about the shape of the underlying distribution.
• Example non-parametric tests:
– Chi-square Test
– Mann-Whitney U Test
– Kruskal-Wallis test
– Wilcoxon signed rank test

Gautam/KAFCOL/2019
Chi-square Test
• When data are frequencies and you wish to compare
these frequencies with an expected set of
frequencies, then you can use Chi-Square test.
X2 = ∑ (fA – fE )2/ fE
Where, fA is the actual frequency and fE is expected
frequency.
fE = (row total x col total)/sum total
• It tells you whether or not two populations are
different enough in some characteristics.
• The data should be raw frequencies (not
percentages)
• The data are assumed to be a random sample from
the population but does not assume any particular
distribution for the population.
Gautam/KAFCOL/2019
Computing Expected Frequency – an example
Men and women preferences of forest products
Timber Fuelwood Fodder Other Total
Men 6 17 9 5 37
Women 13 5 16 9 43
Total 19 22 25 14 80

fE of men preferring timber = (19x37)/80 = say “x”


fE of women preferring timber = (19x43)/80 = …..
………………….
X2, men/timber = (6-x)2/x = …………………
X2, men/firewood = ……………..
…………………………
Sum of all X2 above is the X 2 value for this table
Gautam/KAFCOL/2019
Mann-Whitney U Test
• The test still assumes that the general distributions
of the two groups are similar, but they don’t have to
be normal.
• The objective is to determine whether one
population is “higher” or “lower” than the other.
• HO: the Medians of the two groups are equal.

Gautam/KAFCOL/2019
Kruskal-Wallis test
• Compares three or more sample medians.
• HO: the medians of all the groups are equal.

Gautam/KAFCOL/2019
Choosing Appropriate Statistical Test

• Choice mainly depends upon:


– Type of dependent (outcome) variable (continuous or
categorical)
– Distribution of your data (normal or
abnormal/free/skewed)
• Continuous variable (quantitative data i.e.
interval or ratio scale), and data distribution
normal  can use parametric tests, such as t-
test, ANOVA, Pearson Correlation, linear
regression)

Gautam/KAFCOL/2019
Choosing Appropriate Statistical Test (cont.)

• Continuous variable, data distribution non-normal


 use non-parametric tests, such as Mann-Whitney
U test; Kruskal Wallis test; Spearman Correlation;
linear regression etc).
• Categorical binary (i.e. two categories) variable
(qualitative/nominal/discrete)  use Chi-square test
etc.

Gautam/KAFCOL/2019
Interpretation of Findings
• Interpretation is the process of attaching meaning to the
data (data do not speak themselves)
• Interpretation of findings means “making sense” of
the outputs of your statistical analyses and presentations.
• It involves identifying and abstracting important
understandings from the detail and complexity of the
findings.
• It is about answering the “why?” and “why not?” questions
around the findings, which you include in the discussion
where you compare your findings with findings of similar
past researches.
• Predispositions, assumptions, and biases of the researcher
(consciously or unconsciously) is a barrier to credible
Gautam/KAFCOL/2019
Understanding the Concept of Probability
• A probability is a value extracted from
comparative analysis of two or more sets of data.
• Such comparisons can be between or among:
– parameters among treatment and control
populations
– natural populations that you think differ in some
respects
– population parameters that you think are
related somehow

Gautam/KAFCOL/2019
Concept of Probability (cont.)

 Any difference that you found from a comparative


analysis (e.g. between the control and treatment
groups in an experimental study) may be:
– due to actual effect of your treatment, or
– due to the influence of some factor you didn't
account for in planning your research, or
– by chance (the result of internal or random
variation in your samples).

Gautam/KAFCOL/2019
Concept of Probability (cont.)

• When you run statistical test, it will give you the


probability that the “difference” you found is by
chance rather than a true difference.
• The probability is denoted as “p”.

Gautam/KAFCOL/2019
“p” Values and Statistical Significance
• The traditional approach to reporting a result of
statistical test requires you to say whether the
difference is statistically significant. You are supposed
to do it by generating a p value from a test statistic.
• In general, statisticians and scientists have come to
the general conclusion that an acceptable level of
risk of getting the “difference” by chance is 5% (or p
= .05).
So we indicate a significant result with "p<0.05” or “p
= 0.05”.

Gautam/KAFCOL/2019
“p” Values and Statistical Significance (cont.)

• p = 0.05 means that the probability of this result


occurring by chance is 5%.
• This means that we are 95% confident that the
results are not by chance.
• This is also expressed as: the level of significance is
5%.

Gautam/KAFCOL/2019
DATA ANALYSIS AND
INTERPRETATION

Dr. Ambika P. Gautam


Classification and Organization of Data

• The type and scale of data largely dictates the


type of statistical techniques that can be used
in data analysis.
• Before actually beginning the statistical
analyses, it is important to know the scale and
type of your data.

Gautam/KAFCOL/2015 2
Concept of Probability
• A probability is a value extracted from
comparative analysis of two or more sets of data.
• Such comparisons can be between or among:
– parameters among treatment and control
populations
– natural populations that you think differ in some
respects
– population parameters that you think are
related somehow

Gautam/KAFCOL/2015 3
Concept of Probability (cont.)

 Any difference that you found from a comparative


analysis (e.g. between the control and treatment
groups in an experimental study) may be:
– due to actual effect of your treatment, or
– due to the influence of some factor you didn't
account for in planning your research, or
– by chance (the result of internal or random
variation in your samples).

Gautam/KAFCOL/2015 4
Concept of Probability (cont.)

• When you run statistical test, it will give you the


probability that the “difference” you found is by
chance rather than a true difference.
• The probability is denoted as “p”.

Gautam/KAFCOL/2015 5
“p” Values and Statistical Significance
• The traditional approach to reporting a result of
statistical test requires you to say whether the
difference is statistically significant. You are supposed
to do it by generating a p value from a test statistic.
• In general, statisticians and scientists have come to
the general conclusion that an acceptable level of
risk of getting the “difference” by chance is 5% (or p
= .05).
So we indicate a significant result with "p<0.05” or “p
= 0.05”.

Gautam/KAFCOL/2015 6
“p” Values and Statistical Significance (cont.)

• p = 0.05 means that the probability of this result


occurring by chance is 5%.
• This means that we are 95% confident that the
results are not by chance.
• This is also expressed as: the level of significance is
5%.

Gautam/KAFCOL/2015 7
Statistical Tests
• Can be grouped into two broad types:
1) Parametric
2) Non-parametric
• Choice of test is mainly determined by:
– kind of data that you have collected
– research objective(s)

Gautam/KAFCOL/2015 8
Parametric Tests
• We use parametric tests for interval and ratio
scale data (continuous measures), when the
following assumptions are considered valid:
– Independence of data within and among
samples.
– Homogeneity of variances (i.e. dispersion
around the mean is similar).
– Normal distribution of the data.
• Some common parametric tests include: t-test,
ANOVA, Correlation, Regression.

Gautam/KAFCOL/2015 9
t-test
• A technique used to measure the statistical
significance of the difference between the means
of two populations.
• HO: there is no difference between the sample
means of two populations.
Note: although normality of data is a requirement,
the t-test does not fail if the data are somewhat
non-normal.

Gautam/KAFCOL/2015 10
Analysis of Variance (ANOVA)
• A technique used to measure the statistical
significance of the difference between the means
of more than two populations.
• HO: the means of the three (or more) groups are
equal.
Note: ANOVA only responds with the probability
that there is statistical difference among the
means. It does not respond as to specifically
which means are different.

Gautam/KAFCOL/2015 11
Correlation
• Measures the degree of association between two
variables (i.e. the degree to which two variables vary
together).
(A) Pearson’s Correlation: the coefficient is a measure of
linear association.
HO in Pearson’s Correlation: there is no linear relationship
between the variables.
(B) Spearman Rank Correlation: this is more appropriate if
you don’t know whether there is a linear relationship
among the variables.
HO in Spearman Correlation: there is no uniformly positive
or negative relationship between the variables.

Gautam/KAFCOL/2015 12
Regression
• Regression expresses the value of one variable (Y) in
terms of a function of another variable (X).
• A bivariate linear regression analysis will give you the
correlation information, but it will also calculate the
equation of the straight line that best fits the data – the
regression line.
Y = a + bX, where Y is the dependent variable, X is the
independent variable, ‘a’ is the intercept (the value at
which it crosses the Y axis), and ‘b’ is the slope of
regression line (regression coefficient).
(e.g. you can try to predict the height of a tree by knowing
its DBH.)

Gautam/KAFCOL/2015 13
Correlation vs. Regression
• In Correlation,
– we are concerned with whether the two variables are
interdependent and vary together
– we do not express one variable as a function of
another
– there is no distinction between dependent and
independent variables
– there is no prediction
• In Regression,
– we intend to describe the dependence of variable Y
on the independent variable X
– we can predict the dependent variable based on the
value of independent variable
Gautam/KAFCOL/2015 14
Non-parametric Tests
• Nonparametric tests do not require assumptions
about the shape of the underlying distribution.
• Example non-parametric tests:
– Chi-square Test
– Mann-Whitney Test
– Kruskal-Wallis test

Gautam/KAFCOL/2015 15
Chi-square Test
• When data are frequencies and you wish to compare
these frequencies with an expected set of frequencies,
then you can use Chi-Square test.
X2 = ∑ (fA – fE )2/ fE
Where, fA is the actual frequency and fE is expected
frequency.
• It tells you whether or not two populations are different
enough in some characteristics.
• The data are assumed to be a random sample from the
population but does not assume any particular
distribution for the population.

Gautam/KAFCOL/2015 16
Mann-Whitney test
• The test still assumes that the general distributions
of the two groups are similar, but they don’t have to
be normal.
• The objective is to determine whether one
population is “higher” or “lower” than the other.
• HO: the Medians of the two groups are equal.

Gautam/KAFCOL/2015 17
Kruskal-Wallis test
• Compares three or more sample medians.
• HO: the medians of all the groups are equal.

Gautam/KAFCOL/2015 18
REPORT WRITING AND
PRESENTATION
Writing Research Proposals

A. Thesis Research Proposal


B. Proposal for Seeking Grant

Gautam-2019 2
Parts of a Proposal
• Content of proposals vary depending upon
whether you are writing a thesis research
proposal; proposal for seeking support for a
research grant; project implementation etc.

Gautam/KAFCOL/2019
Thesis Research Proposal – two key
dimensions

A) Relevance: Your research question or idea should


be relevant to society, a social group, and scholarly
literature. This means that the research addresses
an important concern in your field.
B) Answerability/Practicality: The research
question should be answerable or addressable
through data collection within a reasonable (in
your case, available) time frame.

Gautam-2019 4
Proposal Format – Thesis Research
• Title (short and catching)
• Introduction (1-2 pages)
• Statement of the Research Problem (1/2 - 1 page)
• Objectives (General objectives, and Specific
Objectives)
• Research questions
• Hypotheses (if any)
• Literature review
• Conceptual framework
• Limitations (if any; max ½ page)

Gautam/KAFCOL/2019
Proposal Format – Thesis Research
• Methodology
– Study area
– Data collection methods
• Sampling Strategy
• Methods of primary data collection
• Secondary data
• Methods of Data Analysis
• References (in alphabetical order by surname of
authors)
• Appendices (work plan, data collection
instruments, budget etc)

Gautam/KAFCOL/2019
Proposals for Seeking Grants
1) Solicited proposals - Submitted in response to a
specific solicitation issued by a sponsor. Such
solicitations, which are typically called Request for
Proposals (RFP), are often made through a bidding
process.
2) Unsolicited proposals - Submitted to a sponsor that
has not issued a specific solicitation but is believed
(by you) to have an interest in the subject.

Gautam/KAFCOL/2019
Proposals for Seeking Grants..(cont.)

3) Preproposals - Requested when a sponsor wishes to


minimize an applicant's effort in preparing a full
proposal. Also called sometimes an Expression of
Interest (EOI).
4) Continuation proposals- relate to multi-year project
for which the sponsor has already provided funding
for an initial period. Continued support is usually
contingent on satisfactory work progress and the
availability of funds.

Gautam/KAFCOL/2019
Parts of a Grant Proposal
• Usually a proposal prepared to apply for a grant will
have the following components:
Title Page
Abstract
Table of Contents
Introduction -including, a brief Background,
Statement of the Problem, Objectives (i.e. purpose of
research), and Significance of Research

Gautam/KAFCOL/2019
Parts of a Grant Proposal (cont.)
 Literature Review (detail background): this is required (i)
to make clear what has been accomplished in the past;
(ii) to give evidence of your own competence in the field;
and (iii) to show why the previous work needs to be
continued.
 Methodology: This section is the heart of the proposal
and is the primary concern of the technical reviewers. Try
to be:
(a) realistic about what can be accomplished
(b) clear about the focus of the research
© specific about the means of evaluating the data or the
conclusions
Gautam/KAFCOL/2019
Parts of a Grant Proposal (cont.)
Description of Relevant Institutional Resources (if
applying for a grant by an institution)
List of References
Personnel – including (i) an explanation of the
proposed personnel arrangements, and (ii) CVs of
each of the main contributors to the project (if
applying for a grant by an institution).
Budget - the costs to be met by the funding source,
including personnel, administrative, and overhead
expenses. The budget also specifies items paid for by
other funding sources. Should include justifications
for requested expenditures.
Gautam/KAFCOL/2019
Good Proposals Seeking Grants Should Quickly and
Easily Answer the Following Questions

• What do you want to do, how much will it cost, and how
much time will it take?
• How does the proposed project relate to the sponsor's
interests?
• What difference will the project make to the community,
district, country, your institution etc.?
• What has already been done in the area of your project?
• How do you plan to do it?
• How will the results be evaluated?
• Why should you, rather than someone else, do this
project?
Gautam/KAFCOL/2019
Writing Thesis
A thesis usually has the following main components:
1. INTRODUCTION
– Background (of the problem)
– Statement of the problem
– Hypothesis
– Objectives of the study
– Research questions
– Significance/rationale of the study
– Limitations of the study (if any)

Gautam/KAFCOL/2019
Writing Thesis (cont.)
2. LITERATURE REVIEW
– Organization of the review
– Analysis and synthesis
– Summarize, strengths, weaknesses, and gaps in
the literature
– Definition of variables and their relationships
– etc.

Gautam/KAFCOL/2019
Writing Thesis (cont.)
3. METHODOLOGY
– Description of the study area
– Research design
– Data collection methods
– Description of the data collection instruments
– Sampling (strategy and selection)
– Data analysis (including statistical tests if used)

Gautam/KAFCOL/2019
Writing Thesis (cont.)
4. RESULTS
– Present the data (including tables, figures, graph
etc.)
– Summary of findings (detail, if necessary should
be included in Annex).
– Interpretation of the results

Gautam/KAFCOL/2019
Writing Thesis (cont.)
5. DISCUSSION
– Further interpretation of the results, if needed
– Compare how the study does or does not support the
literature

6. CONCLUSIONS AND RECOMMENDATIONS

REFERENCES (arranged in alphabetical order by the


author’s surnames)

ANNEX
Gautam/KAFCOL/2019
Writing Scientific Papers
• Publishing papers in a scientific journal is one
of the best ways to disseminate your research
work. Scientific papers constitute valuable,
lasting references for other scientists.
• Writing high-quality scientific papers takes
time, but it is worth.

Gautam/KAFCOL/2019
Steps involved in Writing Scientific Papers
1) Structuring Scientific Paper –
– Abstract
– Introduction: includes background information,
clarifies the motivation for the work presented and
informs the readers about structure of the paper
– Materials and Methods: provides sufficient detail for
other researchers to replicate the research.
– Results - present the results
– Discussion - discuss the results
– Conclusion - presents the outcome of the work by
interpreting the findings

Gautam/KAFCOL/2019
Steps involved in Writing Scientific.. (cont.)
2) Drafting the Paper: effective writing is clear, accurate,
and concise, which the audience will understand
effortlessly, unambiguously, and rapidly. The following
are some considerations:
– Choosing an appropriate verb and using it well.
– Using the right tense
– Choosing between active and passive voice
– Using abbreviations – beware of overusing
– Writing numbers - in general, write single-digit
numbers (zero to nine) in words
– Capitalizing words – students have tendency of
overusing capital letters

Gautam/KAFCOL/2019
Steps involved in Writing Scientific.. (cont.)

3) Revising Your Scientific Paper - writing is an iterative


process. Do not hope to write a perfect paper in one
pass. Check also the numbering of figures and tables,
the validity of cross-references, the accuracy of
dates, etc. Be aware that translation of Nepali words
in English may distort the meaning.

Gautam/KAFCOL/2019
Miscellaneous Writing Tips
 Presentation  general to specific
 Consistent (not contradictory sections/sentences)
and evidence-based
 Focused on the theme
 Citation (especially for an assertive statement)
 Concise (say everything that is necessary without
using unnecessary words) and succinct (express
ideas and facts clearly and in few words)
 Avoid plagiarism  copy-and-paste from others’
work is not acceptable
 References  in alphabetical order by surname
of the author(s)
Gautam/KAFCOL/2019
Plagiarism
• Plagiarism is presentation of someone else’s ideas or
words as your own.
All of the following are considered plagiarism:
• turning in someone else's work as your own
• copying words or ideas from someone else without
giving credit
• failing to put quotation marks in a quotation
• giving incorrect information about the source of a
quotation
• changing words but copying the sentence structure
of a source without giving credit
Gautam/KAFCOL/2019
How to Avoid Plagiarism?

– Most cases of plagiarism can be avoided by


citing sources.
– Avoiding cutting and pasting
– Enclosing quoted materials within
quotation marks.
– Rewriting in your own style and language.

Gautam/KAFCOL/2019
Citing References
• When citing references in the text:
– Use the author’s last name only.
– If there are two authors: Karki and Tamang (1996).
– If there are more than two authors:
Karki et al. (1996).
– Do not CAPITALIZE or boldface the author’s name.
– Do not use ‘&’ as an abbreviation for ‘and’.
• Citation can be at the beginning or at the end of a
sentence.
Example 1: It has been found that …… ........ (Gautam, 2011).
Example 2: Gautam (2011) found, for example, that …. ………

Gautam/KAFCOL/2019
List of References
• Referencing involves detail listing of the works cited
in the text.
• In alphabetical order of the authors’ surnames.
• There are many acceptable variations in referencing
style.
Some commonly used styles:
• APA (American Psychological Association)
• Chicago Manual of Style - more flexible
• Harvard referencing
Whatever style is chosen, the key is consistency.
Gautam/KAFCOL/2019
List of References

Books Examples:
• Bruffee, K. (1993). Collaborative Learning.
Baltimore, USA, John Hopkins University.
• Gibson, C., M. A. McKean and E. Ostrom (Eds.)
(2000). People and Forests: Communities,
Institutions, and Governance. Cambridge,
Massachusetts, The MIT Press.
• Inoue, M. and Shivakoti, G. (Editors) (2015).
Multi-level Forest Governance in Asia: Concepts,
Challenges and the Way Forward. SAGE India.
Gautam/KAFCOL/2019
List of References
• Journal article, example:
• Gautam, A. P. and Shivakoti, G. P. (2005). Conditions
for successful local collective action in forestry: Some
evidence from the hills of Nepal. Society and Natural
Resources 18(2): 153-171.
• Gautam, A. P., Shivakoti, G. P. and Webb, E. L. (2004).
A review of forest policies, institutions, and changes
in the resource condition in Nepal. International
Forestry Review 6(2): 136-148.
• Metz, J. J. (1994). Forest product use at an upper
elevation village in Nepal. Environmental
Management, 18(3):Gautam/KAFCOL/2019
371-390.
• Internet source, example:
ICIMOD (2013). Kailash Sacred Landscape. International
Centre for Integrated Mountain Development
(ICIMOD), Kathmandu. Available online at:
http://www.icimod.org/?q=9457 (accessed on 15
October 2015).
FAO, 2013. Climate-Smart Agriculture: Sourcebook.
Food and Agriculture Organization of the United
Nations (FAO). Available online at:
http://www.fao.org/docrep/018/i3325e/i3325e00.ht
m (Accessed on 27 February 2016).
Gautam/KAFCOL/2019
Multiple publications by the same agency or author in
a year, example:
• NPC, 2015a. Nepal Earthquake 2015: Post Disaster
Needs Assessment. Government of Nepal, National
Planning Commission (NPC), Kathmandu.
• NPC, 2015b. Nature Conservation National Strategic
Framework for Sustainable Development (2015-
2030). Government of Nepal, National Planning
Commission (NPC), Kathmandu.

Gautam/KAFCOL/2019
Tips for Creating Good Presentation

Number of slides
Font – style, size
Sentences – simple, short
Use of charts, graphs, photographs etc.
Slide background – contrast between
background and text
Grammar, spelling

Gautam/KAFCOL/2019
Tips for Delivering Effective Presentation

 Timely presence
Checking computer, projector etc
Eye contact with audience (do not read all)
Pointer use
Time management
Monitoring the audience
Answering questions

Gautam/KAFCOL/2019
Assignment -2
• Divide the class into 10 groups based on similarity of research
interests
• Each group will find a peer reviewed journal article relevant to
their research interest
• The group will review the article based on the skills learned
• Each group will present the review findings in the class. The
presentation should include the following components: (i) the
title of the article and research problem addressed, (ii)
research objectives, research questions, and hypothesis, (iii)
type of data used and collection method(s), (iv) sampling
strategy, (v) main findings, (vi) conclusions (and
recommendations, if any), (vii) was the review useful for
designing your own research? If yes, how?

Gautam/KAFCOL/2019 33
Assignment - 3

Using the methods suggested by Cochran (1977), and


Yamane (1967), compute separately sample size for a
population of 350 households at a confidence level
of 95%, and precision level of 4%. Assume P = 95%.

Gautam/KAFCOL/2019 34

You might also like