Sampling
The Population: All the units (people, events, Sampling Element:
objects or institutions) from which data are The most basic unit about which information is
collected collected (e.g., an individual employee).
The census: A survey in which information is Sampling Frame:
gathered about all members of a population The list or source from which potential
respondents are selected.
Sample
• A Sample is a smaller collection of units from a Example:
population used to determine truth about that If you are researching working conditions at
population Company X:
• A proportion or subset of a population • Population: All 1,000 employees of the
Why use a sample? company
• It is very difficult to study each unit of the • Sampling Element: Each individual
population employee
• Population units are heterogeneous, so • Sampling Frame: The HR database
studying all is impractical containing names and contact details of all
employees
• Limited resources such as time, money, and
manpower Types of Sample
• Population analysis is cumbersome and
complicated, while samples are easier to
analyze
Sampling:
Sampling is the process of selecting a small number
of units from a larger population to estimate or
predict characteristics, situations, or outcomes of
the entire population.
Population Vs Sample 1. Probability Sampling
o Every member of the population has a
• Population: The entire group of interest. Its known and equal chance of selection.
characteristics are called parameters. o When all elements have the same
• Sample: A smaller part taken from the probability, it is called Equal Probability
population. Its characteristics are called of Selection Design (EPSED) or self-
statistics. weighting.
• We measure the sample using statistics to ❖ Simple Random Sampling
make inferences about the population and • Every member has an equal chance of
its parameters. being selected.
• Population Parameter • Requires a complete sampling frame.
• Sample Statistic
• Best for small, homogeneous, and easily ❖ Judgement (Purposive) Sampling
accessible populations. • Researcher intentionally selects participants
Example: Number employees 1–1000 and considered most suitable.
randomly select 100 numbers. • Useful for specialized or targeted studies.
Example: Selecting disabled students to
❖ Systematic Sampling study their experiences with student
services.
• Population is listed and members are
selected at regular intervals (k) after a
❖ Quota Sampling
random start.
• Population is divided into subgroups, and
Example: Select every 10th employee
samples are selected based on fixed
starting from number 6 (6, 16, 26, …).
proportions.
❖ Stratified Sampling • Like stratified sampling, but selection is not
• Population is divided into subgroups (strata) random.
based on key characteristics. Example: Selecting respondents from age
• Ensures proper representation of all groups (21–30, 31–40, etc.) based on
subgroups. quotas.
Example: From 800 females and 200 males,
select 80 females and 20 males. ❖ Snowball Sampling
• Used when the population is hard to
❖ Cluster Sampling identify or sensitive.
• Population is divided into naturally • Existing participants refer other
occurring clusters. participants.
• Entire clusters are randomly selected, not Example: Researching homelessness
individuals. through participant referrals.
Example: Randomly select 3 offices out of
Determining the Sample Size
10 city offices and survey all employees in
them. o Sample size refers to the number of units
selected for a study.
2. Non-Probability Sampling o A large sample: n ≥ 30
• Sampling methods where not all population o A small sample: n ≤ 30
elements have a known chance of selection. o The sample should be large enough to give
• Used in qualitative research, pilot studies, and reliable results.
exploratory research.
Techniques for Determining Sample Size
❖ Convenience Sampling ▪ Roscoe’s Rule (1975):
• Also called grab / opportunity / accidental Sample size between 30 and 500 is suitable
sampling. for most studies.
• Sample is chosen because it is easy and
▪ G*Power Technique:
readily available.
Used for power analysis in statistical testing.
• Cannot generalize results to the whole
population. ▪ Morgan Table (Krejcie & Morgan, 1970):
Example: Surveying people available at a Provides recommended sample sizes based
shopping mall in the morning. on population size.
▪ 50–50 Method:
Assumes maximum variability when
population proportion is unknown.
▪ Sample Size Estimation Formula:
Used for finite and infinite populations.
Factors Affecting the Choice of Technique
• Number of variables in the study
• Statistical tools used
• Statistical software/package adopted
Sampling Error
• Sampling error is the difference between
sample statistics and population parameters
caused only by selecting a sample instead of
the whole population.
• Increasing sample size reduces sampling
error.
Types of Sampling Errors
1. Sample Errors
• Occur due to the process of sampling.
• Cause sample results to differ from
population results.
• Statistical errors are sample errors.
2. Non-Sample Errors
• Not controlled by sample size.
• Cannot be avoided by using a large sample.
• Non-response Error: Selected respondents
do not respond fully or partially.
• Response Error (Data Error): Bias during
data collection or analysis due to
o Respondent errors
o Interviewer bias
o Recording errors
o Poorly designed questionnaires
o Measurement errors