RESEARCH METHODS AND

SAMPLING DESIGN
[Document subtitle]
 Sampling designs
Universe or population. And sample
The whole pool from which a statistical survey is taken is the population. A whole community of
individuals, things, incidents, hospital stays, or measurements can be referred to by a population. For
making this clearer the example of workers in factory and cake is given. That all workers in the factory
and a whole cake is population. When you are talking about whole things you are talking about
population. Opposite to that when you take one worker from the factory for research or testing
purpose, or when you cut a small piece of cake for tasting you will get idea that how whole cake will
taste. By only testing a small piece. So that small piece from whole cake will be called sample.

Simply in video the whole thing or aggregate is population or universe. And when you take small portion
from that whole for testing purpose its sample. You get idea from that sample that how whole thing is.
What is average result.

Sampling frame
A listing of the objects or persons that create a targeted population from which a sample is taken.

Again, giving the example of workers in factory. If there are 14 workers in a factory. Any individual
worker will be called one unit. Making list of workers will be called sampling frame. Many times, the lost
is already available you don’t have to make it like telephone directory. So, you can take sample from
that sample.

Sample design
You select some of the units from population for research or testing purpose. But whom or what units
will you select. For that you make sample design it is called blue print for guiding to obtain a sample
from a sampling frame (the list).

There are two methods

 Probability/random sampling types

1. Simple random sampling

Subject or the population get equal chance of the selection mostly parametric test are used.

Lottery method of random sampling

The researcher gives a number to each member of the population in this process. To select
samples, researchers pull numbers from the box randomly. In this way each unit has equal
chance to be chosen as sample. The number chit will be taken randomly without looking at it. To
ignore the biasness and error.

2. Stratified sampling
Sampling from a group that can be split into subpopulations. Making different groups on
basis of sub populations within the population. On the bases of differences between each
 group. Every unit in one group will be with same characteristics. Then we will use simple
random sampling and assign numbers. And pick randomly. And then do your study.
This is mostly used when we divide population into groups. Units are different but
homogenously alike.
For example, dividing students into groups on the bases of course. Or dividing people intro
group on the bases of gender

3. Cluster sampling
The sampling of clusters involves splitting a population into clusters, such as counties, and
choosing a subset of these clusters randomly. But here the units in one group will not be
with same characteristics. And same numbering technique and then selecting will be
applied.

4. Systematic sampling
In systematic sampling, researchers at regular intervals pick representatives of the
population for their survey. For example, researcher say I will select every 2 nd person in the
list like 2nd, 4th, 6th unit will be selected like this systematic sampling works

 Non probability/non random sampling

There is no equal chance of selection non parametric test are mostly used.

1. Convenience sampling
A convenience sample is a type of method of non-probability sampling where the sample is
collected from a group of individuals who are easy to contact or meet.
For example, one researcher goes and ask people on metro bridge that how it is to travel on
metro. He does not take any difficult alternatives. He chooses one which was convenient.

2. Judgmental sampling or purpose sampling

Where respondents are deliberately made a part of sample. Based on his own existing
knowledge, or his professional judgement, the researcher selects units to be sampled.

3. Quota sampling
It is like stratified sampling but there is non random selection. Researchers produce a
survey that comprises people representing a community. According to particular
characteristics or abilities, researchers select these people.
Probability vs non probability

Census vs Sample
Whereas a survey is an approach to acquire statistics about each member of a population, sampling only
gathers data about a section, the sample, to represent the entire. Although a survey is just half of the
population, we can research it more comprehensively than other members of the population can do.

Doing study of all units of population is called census. And doing study of all units of population is called
sample.

Unit
Any individual object or thing from which the information is to be collected is called unit.
In factory there are 100 workers so one individual worker will be one unit.

Sampling or non-sampling error

When you are taking a sample, the errors you face is called sampling error. And beside that all the
errors are called non sampling errors.

Total error= Sampling error + non sampling error

Total errors

1. Sampling error
If the sample used in the analysis is not representative of the entire population, a sampling error occurs.
And it depends on nature of universe and sample design etc.

For example: taking 10 students from 500 students randomly and testing and then concluding that all
students are very intelligent. That is not possible. Because there might be a chance that you only took
intelligent students you didn’t took random students. So, error may occur that those students do no
represent the whole class. Because there are also weak students in that class

it has two types.

 Biased error
Bias is a deliberate misunderstanding that leads to an inaccurate result or relationship
calculation. Many variables will bias the findings of a research in such a way that a particular
effect you are attempting to explain is canceled, diminished or exaggerated.
Simple when researcher has committed any mistake when selecting a sample
This error can occur due to following reasons.
 Faulty process of selection
When researched has committed any mistake and has wrong method or process in selecting
sample.
For example, use of non-probability technique in selecting a sample
 Non response - for example when some of people who are part of the sample do not respond it
will lead to bias.
 biased due to the faulty collection of the data
 biased due to faulty analysis
 Choosing methods of analysis on convenience basis rather than on suitability bases will lead to
bias

Methods of reducing sampling error

 By increasing size of sample
 Use suitable sampling procedure
 By selective best representative

 Unbiased error
The unbiased errors arise because of a chance, i.e., the examiner has not deliberately tampered
with the data and there has been a chance of the discrepancy between the population and the
sample.
2. Non sampling error has two types
Errors which rise for the reason other then sampling error is called non sampling error for example
data collection error, editing error, instrument error. Wrong coding
1. Response error
Answer errors may result from the respondent's failure to report the correct value
(respondent error), the interviewer's failure to record the correctly recorded value
(interviewer error), or the instrument's failure to correctly measure the value (instrument
error).
 1. researcher error
when researcher is committing mistakes. Error from researcher side is call researcher
error

2. interviewer error
An interviewer error happens when an error occurs when an interviewer (or
administrator) reports a response. An interviewer can lead a respondent in
qualitative research to address a certain way. An interviewer can pose the question in
a different way in quantitative analysis, which leads to a different end result.

3. respondent error
the one who is giving interview. When he is providing inaccurate data, or wrong
answers then that error is called respondent error. These errors occur because of lack
of knowledge, fatigue, or shyness

2. Non response error

It arises when some of the respondent
do not respond. Nonresponse error in survey data arises from the lack of ability of the
entire sample to obtain a meaningful answer to all survey items. When that non-response
leads to biased estimates, it is a critical concern.
Non response could be due to failure of locating the respondent, the respondent has
unwillingness to respond

How to reduce non sampling error

3. By employing qualified and trained experienced personal
4. By using better equipment
5. By increasing efficiency of supervision
 2nd video
Research Design and its type
Research design
The research design refers to the general methodology that you use to combine the multiple
components of the analysis in a consistent and rational manner, meaning that the research
challenge is solved effectively. For example, if you are thinking to make a building then you have
visualization already in your mind that it will be 3 floor or 4 floors, how it will be designed the
blue print is same example as research design.
Another example you want to research how much stress is in elder population so you will first
have sample from elder population. Now how to find elderly population, which particular
segment you will research on. Next question you will make questionnaire or interview. All these
things that come in mind and we go to solution. This all planning is called research design.
Simply it is a structure, a design to better generalize and address our research.

Types of research designs

 Quantitative research methods
It is also called fixed design method because we do this to confirm our
hypothesis, we use proper method, and we use random sampling.
Quantitative analysis uses screening tools to gather information from current
and prospective clients and to carry out online surveys, internet polling,
questionnaires, etc.

1. Experimental
The method whereby a researcher designs a thesis is experimental
design. The best method of defining a cause-effect relationship between
two factors is a two-group architecture.

2. Quasi-experimental
a quasi-experimental design aims to establish a cause-and-effect
relationship between an independent and dependent variable. However,
unlike a true experiment, a quasi-experiment does not rely on random
assignment. Instead, subjects are assigned to groups based on non-
random criteria.

3. Descriptive
 Descriptive analysis attempts at defining a group, condition or
phenomena reliably and systematically. It will answer questions as to
what, where, when and how, but not questions about why. To analyze
one or even more factors, a descriptive research design may use a wide
range of research methods.

4. Comparative
Comparison study, simply put, is the process of contrasting two or more
items in order to find something from comparing one or more of the
things. ... For comparative analysis, the usual response of contrasting
items is the same as it is in our daily reference experience.

5. Correlational
Correlational research is a type of non-experimental study wherein two
variables are measured by the researcher and the statistical relation (i.e.,
the correlation) between them is calculated with little to no attempt to
monitor international variables.

6. Survey
Survey Research is defined as the method of conducting research using
surveys submitted to survey respondents by researchers.

 Qualitative research methods

It is also called flexible research design. For data collection, sampling method,
and framework you get flexible design. You can bring changes to process so
this is flexible in this manner. The design is flexible, iterative and evolving and
thus requires the willingness of the researcher to modify and adjust the
testing methodology in line with emerging findings. Consequently, qualitative
analysis differs from quantitative studies because it facilitates flexibility in the
research process.

1. Ethnography
Ethnographic research is a qualitative methodology in which researchers
observe the subjects of a sample in their real-life context and/or
communicate with them. ... The aim of an ethnographic analysis inside a
functionality project is to get a design problem 'underneath the skin' (and
all its associated issues).

2. Phenomenology
 Phenomenology can be defined as a science methodology that attempts
to explain the nature of a phenomenon from the point of view of those
who have encountered it.

3. Case study
A case study is a research technique in the social and life sciences
requiring an up-close, most of, and systematic analysis of a particular
case. A case study in medicine, for example, might investigate a single
patient treated by a physician, and a case study in industry may study the
policy of a particular organization.

4. Grounded theory
The Grounded Theory encompasses data processing and interpretation.
The hypothesis is "dependent" on actual data, which means that once
you have gathered the data, the study and creation of hypotheses exists.

5. Historical analysis
Historical analysis is a technique for analyzing facts to get an
interpretation of the past. Grounded theory requires historical research.
While it can be applicable to all artefacts, it is especially applied to
evidence found in records. Second, the historian tries to obtain some
assurance about the truth of the past.
Basic Vs applied
Basic focus on new knowledge creation. We can do detailed work on old work. It has no direct
implication to change the society. It focusses on how good we can generalize human behavior
Simply Basic research is a fully analytical approach to research aimed at developing or
extending the knowledge base of a given field of study. It emphasizes on "information for the
sake of it" and is motivated mainly by curiosity and the desire to investigate the mysterious.
It is also described as fundamental or pure analysis and is a comprehensive investigation
intended not to solve a particular problem, but to obtain a deeper and more thorough
understanding of the research topic or phenomena.

Applied research
It is solution focused. We take a scenario. We do research. And we want to bring change to that
situation. Simultaneously intervention driven, solution focused action research. We go to
solution.
Applied analysis is designed to concentrate on presenting a real issue with realistic solutions. It
is a type of analysis that includes solution-oriented research into a phenomenon, a field of
study or research topic that typically uses observational methodologies.
it further investigates the outcomes of pure or basic research in order to validate these findings
and apply them to create innovative solutions to specific problems.

Inductive vs deductive
Qualitative is inductive in nature. Data which is available you gather it and then you move to
specific thing.
Quantitative is deductive you have hypothesis and a research problem and you create a specific
design and then you move according to that design. That your hypothesis is proved or not.

Experimental vs non experimental

Experimental is structural and causal and we divide it inro two groups, a controlled group and
an experimental group
Non experimental is explorative, descriptive historical in nature

Confirmatory vs exploratory
Exploratory research is the step of the research process that attempts to connect ideas and
unveil the "why" of possible interactions between cause and effect. ... Confirmatory research
(a.k.a. checking of hypotheses) is where scientists get a pretty clear idea of what is expected.

Explanatory research
Currently, explanatory research is a form of study design that focuses on illustrating the aspects
of the research. The researcher starts with a broad understanding and uses analysis as a
method that could contribute to the issues being discussed in the future.

Causal research
The investigation of (research into) cause-and-effect interactions is causal research, also called
explanatory research. It is necessary to analyze variance in the variable assumed to cause the
change in the other variable(s) to determine causality, and then calculate the changes in the
other variable(s) (s).
Descriptive research
Descriptive research is described as a procedure of research that explains the features of the
studied population or phenomena. ... The approach of descriptive analysis focuses mainly on
explaining the existence of a population segment, without concentrating on "why" a given
phenomenon happens.

Correlational research
Correlational analysis is a type of non-experimental research approach in which two variables
are measured, interpreted and analyzed by a researcher without any external variable affecting
the statistical relationship between them.
So that was all about how research design can vary with what type of research you are doing
and what is the situation.