Population & Sampling
Dr Mohd Radzi Tarmizi Bin A Halim
Dr Mohd Aris Safree Bin Mohd Yasin
Populations Versus Samples
• Population: A population is a complete set of elements (people, items, or
objects) with specific characteristics you want to study and make inferences.
• Sample: A subset of a population. This is the particular set of elements that you
will collect data from.
• Sampling frame: A sampling frame is the source material or the complete list
from which a sample is drawn. It is a comprehensive compilation of all the
elements in your population.
• Sample size: The sample size is the number of population members surveyed,
measured, or observed for a particular experiment. The sample size determines
the amount of data you have, which further influences the precision of your
study and the reliability of your results. Generally, larger sample sizes produce
better and more accurate results.
• All of the math teachers teaching in a particular school are the
population.
• The sampling frame is the school record (list) that names each of the
teachers employed in the school.
• Suppose there are 40 teachers teaching math. The sample size could
vary from 1 to 40 depending on the purpose of your research and
how many teachers you select.
• The sample is the actual subset of math teachers that you choose to
study.
Probability Sampling
Simple random sampling
• Simple random sampling is a technique wherein each member of the
population has an equal chance and probability of being selected.
Equipment such as random number tables or random number
calculators can be utilized to perform simple random sampling. For
example:
• Let’s say you wanted to select a simple random sample of 50
passengers in a train and interview them about their age. You
randomly pick people based on their seat numbers and interview
them.
Systematic sampling
• Systematic sampling involves selecting sample members from a
random starting point but then choosing sample members from that
starting point at fixed and periodic intervals. Systematic sampling is
straightforward compared to simple random sampling and allows a
wide study area to be covered. For example:
• The seat numbers of the train are numbered 1-500. You select a
random seat number such as ten as the starting point. Starting from
10, every fifth person (i.e., 10, 15, 20, 25, and so on) would be
included in the sample.
Stratified sampling
• Stratified sampling involves dividing the population into subpopulations or homogeneous
subgroups (strata) based on specific characteristics. It allows for more improved precision
than simple random sampling as it enables the selection of all the subgroups which may
be of interest within a population.
• The population is first divided into subgroups, or strata, according to various
characteristics such as age, height, weight, occupation, salary, etc. Then, depending on
the percentage of a stratum within a population, the number of people who need to be
sampled from a particular subgroup can be calculated. The selection of the sample from
a subgroup can then be conducted through random or systematic sampling. For example:
• The train consists of 200 passengers in the age group 1-30 years, 200 passengers in the
age group 31-60 years, and 100 passengers in the age group above 60. The age balance
in the sample can be achieved by sorting the population into three strata based on age
groups. Then, using random sampling, you can select 20 passengers each from the first
two age groups and ten passengers from the third age group. Therefore, you would be
able to obtain a representative sample of 50 passengers, based on the age makeup of
passengers on the train.
Cluster sampling
• Cluster sampling involves dividing a population into subgroups with the same characteristics as the whole sample. Then,
rather than selecting individual people from the population, the subgroups, or clusters, themselves are randomly
selected. For example, let’s say a researcher divides their population into 20 equal size clusters. They then may decide to
randomly select only ten of those clusters- the members of those particular clusters are then the only people who may be
selected to be a part of the sample.
• The data collection from clusters can be done in a few ways:
• Single-stage sampling: Data is collected from every element or member in a selected cluster.
• Double-stage sampling: Data is only collected from random members of a selected cluster.
• Multi-stage sampling: Another layer of random selection is added to shrink the size of the sample further. Larger clusters are
broken into smaller clusters, smaller clusters are randomly selected from within the larger cluster, and then members of the
smaller cluster are selected.
• Cluster sampling can be convenient for researchers, is easy to implement, and provides accuracy. Each cluster can contain a
large sample that compensates for the loss of accuracy in information about each individual. However, there is a chance for
error, as clusters might be slightly different from each other and may not be representative of the whole population. For
example:
• Ten trains are running on the same route with the same passenger capacity and age groups. However, you can’t travel on all
of the trains to collect the data. So you randomly select four trains. These four trains act as clusters for your sampling
experiment. In single-stage sampling, you would interview every person on the selected trains. In double-stage sampling, you
would randomly select people on the given trains to give interviews. In multi-stage sampling, you would shrink your clusters
by first selecting random cars and then selecting individuals in those cars to interview.
Non-Probability Sampling
• Non-probability sampling is a
technique in which the sample
is selected using a non-random
manner based on the
researcher’s subjective
judgment. Therefore, each
element of the population does
not stand an equal chance of
being selected.
Convenience sampling
• Convenience sampling is the most common type of non-probability
sampling. It involves the selection of a sample that can be easily
accessed, located, or contacted by the researcher.
• Though it is an easy and inexpensive method of sampling, this
method of sampling may not provide the correct representation of
the population and therefore presents a challenge when generalizing
the result to the whole population. For example:
• You want to study the working conditions of the employees in malls.
So, you ask the mall employees near your house to fill a survey on the
topic. Though it is a convenient way of collecting data, it does not
represent the employees in all malls.
Voluntary response sampling
• Voluntary response is similar to convenience sampling in that it also relies
on ease of access. However, in this form of sampling, the sample comprises
people who have voluntarily agreed to participate in the study instead of
the researcher selecting the participants.
• While it is an easy and cost-effective method of sampling, the researcher
lacks the control of selecting a sample, resulting in a slight bias. For
example:
• You send out surveys about working conditions to the employees of all the
malls in your city. Many employees fill out the survey, which helps you
collect data related to the issue. However, there is a chance that the data
may be skewed. It may be that the survey is more likely to be answered by
employees who are not satisfied with their working conditions; therefore,
the opinions expressed in the survey might not reflect the opinions of all
employees.
Purposive sampling:
• Also known as judgmental sampling, this method involves selecting
the participants based on the researcher’s knowledge and judgment
of who would be an appropriate participant in the survey. Though this
method can provide in-depth knowledge about a particular issue, this
method is vulnerable to errors that may arise due to the researcher’s
judgment, low reliability, and increased chances of bias. For example,
• You want to know about the working conditions of employees in the
malls in your city. You know who has worked there the longest, and so
you purposely send your survey only to the long term employees.
Snowball sampling:
• Snowball sampling is used when the potential participants are rare to
find. In this method, the research participants to which you have
access help recruit other participants for the study. For example,
• You are looking for female participants who belong to a particular
country X. You have access to only one such participant. She helps
you access the other participants for your research.
Probability vs. non-probability sampling