Sampling Design

Meaning of Sampling

• Sampling is a research technique in which a smaller group (sample) is selected from a

larger group (population or universe) to study and make conclusions about the whole.
• Since it is not always possible or practical to study every unit of the population, researchers
rely on sampling to save time, cost, and effort while still maintaining accuracy.
• The idea is that the sample should represent the characteristics of the entire population,
so that results from the sample can be generalized.

Illustration:
If a company wants to test the satisfaction level of its 50,000 customers, it is not feasible to ask all
of them. Instead, it may select 500 customers (sample) that represent different age groups, regions,
and income levels. By analyzing this sample, the company can estimate the satisfaction level of
all 50,000 customers (population).

Definitions of Sampling

• Kerlinger (1986):
“Sampling is taking any portion of a population or universe as representative of that
population or universe.”
• Yates (1960):
“Sampling is the selection of a part of an aggregate or totality on the basis of which a
judgment or inference about the aggregate or totality is made.”
• Cochran (1977):
“Sampling is the process of selecting some elements from a population so that by
studying them, we may fairly generalize our results back to the population from which
they were chosen.”

Types of Sampling Design

Sampling designs are broadly classified into two categories:

1
 1. Probability Sampling – Every unit has a known, non-zero chance of selection.
Considered more scientific and reliable.
2. Non-Probability Sampling – Units are selected based on researcher’s convenience or
judgment. Less scientific but often practical in exploratory studies.

A. Probability Sampling

In probability sampling, each element of the population has a known and calculable chance of
being selected. This ensures objectivity and allows the use of statistical tests.

1. Simple Random Sampling

• Meaning: Every unit of the population has an equal chance of being selected. Selection is
purely by chance.
• Techniques:
o Lottery method: Names/numbers placed in a box, drawn randomly.
o Random number tables: Numbers generated using statistical tables.
o Computer-generated random numbers.
• Example: From a list of 500 employees, randomly pick 50 employee IDs.
• Advantages: Simple, unbiased, and easy to understand.
• Limitations: Requires complete population list (sampling frame). May be costly if
population is large and scattered.

2. Systematic Sampling

• Meaning: Selects every kth unit from a list, after a random starting point.
• Formula:

𝑁
𝑘=
𝑛

Where N = population size, n = sample size.

• Example: From 1000 names, if sample size is 200, then 𝑘 = 1000/200 = 5. Select
every 5th name after a random start.
• Advantages: Simple, less time-consuming, ensures even spread of sample.

2
 • Limitations: Can lead to bias if list has hidden patterns (e.g., every 5th student is from
the same department).

3. Stratified Random Sampling

• Meaning: The population is divided into strata (groups) based on certain characteristics
such as gender, age, income, or region. Samples are drawn from each stratum, either
proportionally or equally.
• Example: A survey of students divided by gender (male/female) or course ([Link], BBA,
[Link]). Proportional numbers are picked from each group.
• Advantages: Ensures representation of all subgroups, reduces sampling error, improves
accuracy.
• Limitations: Requires prior knowledge of strata, may be complex to organize.

4. Cluster Sampling

• Meaning: Population is divided into clusters (usually based on geography, institutions, or

natural groups). A few clusters are randomly selected, and all units within those clusters
are studied.
• Example: Select 10 villages randomly from a district, and survey all households in those
villages.
• Advantages: Cost-effective, useful for large and scattered populations.
• Limitations: Less accurate, since all clusters may not be representative.

B. Non-Probability Sampling

In non-probability sampling, selection of units does not give every member of the population a
known chance of being included. Selection depends on judgment, convenience, or availability.
Useful in exploratory or qualitative research.

1. Convenience Sampling

• Meaning: Samples are selected based on availability and accessibility.

• Example: Interviewing people at a shopping mall or bus stop.

3
 • Advantages: Quick, easy, inexpensive.
• Limitations: High risk of bias, poor representativeness.

2. Judgmental / Purposive Sampling

• Meaning: Researcher selects respondents based on expertise, purpose, or relevance to the

study.
• Example: Selecting doctors to study medical ethics, or CEOs to study corporate
governance.
• Advantages: Useful for expert opinions, targeted research.
• Limitations: Subjective, prone to researcher bias.

3. Quota Sampling

• Meaning: Population divided into categories; researcher collects data until a quota for
each category is filled.
• Example: For a survey requiring 60% male and 40% female respondents, the researcher
continues to collect responses until quotas are achieved.
• Advantages: Ensures representation of specific categories.
• Limitations: Within each quota, selection is not random, so bias remains.

4. Snowball Sampling

• Meaning: Existing respondents refer new respondents. The sample "snowballs" as

referrals increase.
• Example: Studying drug users, HIV patients, or migrant workers—where respondents
are hard to locate.
• Advantages: Useful for hidden or hard-to-reach populations.
• Limitations: Risk of network bias (respondents may refer similar people), not
representative of general population.

4
Data
Meaning
Data refers to the collection of facts, figures, observations, and measurements that are used for
analysis in research. It is the raw material on which research conclusions are built. Data can be
quantitative (numerical, measurable, objective) or qualitative (descriptive, subjective,
interpretative). For example, the age and income of respondents are quantitative data, while
opinions, attitudes, and perceptions are qualitative data. In research, data serves as evidence to
support or reject hypotheses, and its accuracy directly influences the quality of findings.

Types of Data

Data in research is classified mainly into primary data and secondary data.

• Primary Data is collected directly from the source for the first time. It is original, specific
to the research problem, and reliable. Examples include surveys, experiments,
observations, and interviews.
• Secondary Data refers to data already collected and recorded by others. It is obtained from
published reports, government statistics, journals, websites, and databases. Secondary data
saves time and cost but may not always fit the specific requirements of the research.
Data may also be classified as quantitative (numerical, e.g., income, marks) and
qualitative (non-numerical, e.g., behavior, opinion).

Methods of Data Collection

• Researchers use different techniques to collect data depending on the nature of the study.
• Observation method involves recording behavior or events directly. It is often used in
psychology, anthropology, or consumer behavior studies.
• Survey method includes questionnaires and interviews to collect responses from
individuals. Surveys are common in social sciences and business research.

5
 • Interview method allows face-to-face, telephonic, or online interactions to gather
detailed information.
• Experimentation is widely used in scientific and applied research, where controlled
variables test cause-and-effect relationships.
• Documentary sources such as census reports, journals, books, and online databases
provide secondary data. In modern research,
• digital data collection through online surveys, mobile apps, and social media analytics
has gained prominence.

Introduction to Data Analysis

Data analysis refers to the process of systema8cally examining data to draw meaningful
conclusions. The process begins with data edi'ng and coding to ensure accuracy and consistency.
Quan8ta8ve data is analyzed using descrip8ve sta8s8cs such as mean, median, mode, standard
devia8on, and percentages, which summarize and simplify large data sets.

Introduction to Questionnaire Design

A questionnaire is a structured set of questions used to collect information from respondents in a

survey. The quality of the questionnaire determines the reliability of the data. A good questionnaire
must be clear, concise, unbiased, and logically ordered. The design process begins by identifying
research objectives and defining the information needed.

Questions may be closed-ended (multiple-choice, rating scales, yes/no) for easy analysis, or open-
ended to capture detailed opinions.

The wording of questions should be simple, avoiding technical jargon or leading statements that
may influence answers. The order should move from general to specific, starting with easy and
interesting questions to engage respondents. Pilot testing the questionnaire on a small group helps
identify ambiguities or problems before large-scale use.

Steps in Questionnaire Design

• Define Objectives: Clearly identify the purpose of the research and the type of information
required.
• Draft Questions: Frame questions that are simple, clear, and unbiased to avoid confusion.
6
 • Sequence of Questions: Arrange questions logically, starting with general and easy ones,
then moving to specific or sensitive ones.
• Include Demographic Questions: Collect background information like age, gender,
education, and income at the end.
• Select Question Types: Decide whether to use open-ended, closed-ended, or a mix
depending on research needs.
• Pilot Testing: Test the questionnaire on a small sample to identify flaws or ambiguities
and refine it before large-scale use.

Report Writing

Report writing is the process of presenting the outcomes of research in a systematic, organized,
and structured format. A research report communicates the purpose of the study, methodology,
data analysis, findings, and conclusions to the intended audience such as academic peers,
policymakers, or institutions.

A good research report should be clear, concise, objective, and logically organized. It should
follow a formal style and avoid personal bias or unnecessary complexity. Reports usually consist
of three main sections:

1. Preliminary Section
o Title page (topic, author, institution, date)
o Acknowledgements
o Abstract or executive summary
o Table of contents and list of tables/figures
2. Main Body
o Introduction: Background, research problem, objectives, hypotheses.
o Review of Literature: Previous studies and research gap.
o Methodology: Research design, sampling, data collection methods.
o Data Analysis and Interpretation: Presentation of results using charts, tables,
statistical tests.
o Discussion: Interpretation of results, implications, comparison with past research.
o Conclusion and Recommendations: Major findings and suggestions for practice
or future research.
3. End Section

7
 o References/Bibliography (using proper citation styles such as APA).
o Appendices (questionnaires, raw data, supplementary material).

A well-written report not only presents findings but also demonstrates the credibility of the
researcher and contributes to knowledge in the field.

Characteristics of a Good Report

1. Clarity

A good report must be written in a clear and simple language so that readers can easily understand
the ideas and findings. Ambiguity, technical jargon, and vague statements should be avoided. The
sentences should be short and precise to convey the message without confusion.

2. Conciseness

The report should be brief but comprehensive. It must include all essential facts, analysis, and
conclusions but avoid unnecessary details and repetition. A concise report saves the reader’s time
and makes the document more impactful.

3. Objectivity

Reports must present facts and analysis without personal bias, emotions, or opinions of the
researcher. The conclusions should be based on evidence, not assumptions or personal beliefs.
Objectivity enhances credibility.

4. Logical Organization

The content of the report should be arranged in a systematic order, usually starting with the
introduction, literature review, methodology, analysis, results, and ending with conclusions and
recommendations. Logical flow helps the reader follow the argument step by step.

5. Accuracy

Accuracy in data collection, analysis, and presentation is essential. Figures, tables, charts, and
references should be correct and reliable. Even minor errors may mislead readers or weaken the
value of the report.

8
6. Completeness

A good report should provide all necessary information related to the research objectives. It must
answer the questions raised by the problem statement and include sufficient background, data, and
evidence to support the findings.

7. Proper Presentation

The report should be attractive and well-formatted. This includes a neat layout, proper headings
and subheadings, correct numbering of tables and figures, consistent fonts, and adequate spacing.
Good presentation improves readability and professionalism.

8. Consistency in Style

Reports should follow a consistent writing style throughout, including uniform tense, voice, and
referencing style (e.g., APA). Inconsistencies distract readers and reduce quality.

Plagiarism

Plagiarism means using another person’s words, ideas, or creative expressions without giving
them proper credit and presenting them as your own. It is not just limited to copying entire
paragraphs or articles—it also includes:

• Copying sentences or phrases directly from books, articles, or websites without citation.
• Paraphrasing someone else’s work (changing a few words) but keeping the original idea
and not citing the source.
• Using data, charts, figures, or images created by others without acknowledgment.
• Submitting someone else’s assignment, research paper, or thesis as your own work.
• Reusing your own previously published work (self-plagiarism) without mentioning it.

Prevention

• Cite every source of ideas, words, data, images, instruments, code (follow the required
style, e.g., APA).
• Quote & paraphrase correctly: Use quotation marks for exact text; when paraphrasing,
change structure and wording substantially and cite the source.

9
 • Track sources diligently: Keep meticulous notes; separate your words from source notes
to avoid accidental copying.
• Use reference managers: Zotero/Mendeley for accurate citations; maintain a complete
bibliography from day one.
• Acknowledge reuse: Disclose prior publication when reusing your own text, figures, or
data; get permission where needed.
• Run a similarity check (before submission) and resolve flagged overlaps ethically
(rewrite, quote, or cite).

Falsification

Falsification means changing, manipulating, or misrepresenting real data, methods, or results

in research so that the findings look different from reality. Unlike plagiarism (stealing ideas),
falsification involves tampering with genuine research materials to mislead others. This can
happen in many ways, such as:

• Editing survey responses to make results appear stronger.

• Deleting “unfavorable” data points (e.g., ignoring responses that do not fit the hypothesis).
• Misrepresenting research methods (claiming a random sample was used when it was not).
• Altering images, graphs, or charts to exaggerate the findings.
• Changing the time, place, or conditions of data collection and reporting it differently.

Prevention

• Pre-register hypotheses, primary outcomes, exclusion rules, and analysis plans (where
feasible).
• Keep an audit trail: Preserve raw data, code, and version history; document every
transformation.
• Use reproducible workflows: Scripted analyses (R/Python), code review by a
colleague, and shared repositories (OSF, Git).
• State all analyses transparently: Distinguish clearly between confirmatory (pre-
specified) and exploratory (post hoc).
• Define data cleaning rules in advance (e.g., outlier thresholds) and report all
exclusions with reasons.

10
 • Ethical figure handling: Only apply global adjustments; never add/remove features;
keep original images.

Fabrication

Fabrication is the most serious form of research misconduct. It means making up data, results,
participants, or sources that never existed at all and presenting them as if they were real. Unlike
falsification (where data exists but is altered), fabrication involves completely inventing
information. Examples include:

• Creating fake survey responses instead of actually collecting them.

• Making up numbers in an experiment without conducting it.
• Citing references or sources that do not exist.
• Inventing interviews or case studies that were never carried out.

Prevention

• Realistic planning: Set feasible timelines and sample sizes; avoid perverse incentives that
pressure results.
• Data governance: Maintain signed consent forms, timestamps, device logs, and data
provenance records.
• Independent verification: Random spot-checks by supervisors; double-entry of key
variables; audit trails.
• Mandatory data deposition: Share de-identified data and code where policy allows;
enable replication.
• Team culture: Ethics training, open discussion of null results, and valuing rigor over
“positive” findings.

11