Probability and Non-Probability

Sampling
Aneesa Fatima
23BPY006
 Definition and Principles of Sampling
• The process of selecting a subset of individuals from a population to estimate characteristics of the

whole population.

• Importance of Sampling: Saves time, reduces cost, and increases accuracy in research.

• Types of Sampling:

1. Probability Sampling

2. Non-Probability Sampling

Principles of Sampling:

• Randomization: Ensures each unit has an equal and independent chance of selection.

• Representation: Sample must be representative of the entire population.

• Elimination of Bias: Reduces subjective selection and increases reliability.

Probability Sampling
• In this type of sampling, all the subjects in the population get
an equal chance of being selected as a sample or as a
representative of the population

• It is also known as Random Sampling

• This type of sampling is usually used when the results are

used to generalize in the population

• Mostly used in quantitative research

Types of Probability Sampling
1. Simple Random Sampling (SRS)
2. Stratified Random Sampling

Proportionat Disproportionat
e Stratified e Stratified
Sampling Sampling

3. Cluster Sampling
Simple Random Sampling SRS
• Every member of the population has an equal chance of being selected.

• There are a few different ways to do this, the researcher could number
each member of the population, to keep it simple, say a population of 90
members.

• He or she could then place numbered cards 1 through 90 in a hat or bowl

or mixer and select as many cards as needed to complete the sample, or
they could use a calculator or computer to generate random numbers, or
they could use a random number table.
Stratified Random Sampling
• A stratified sample is where a researcher will divide the population into

subgroups to have members from each segment of the population, and a random

sample is derived from each subgroup.

• For example, let’s say you wanted to know how much money people saved on a

yearly basis. You could have subgroups of people in their 20’s, in their 30’s, in

their 40’s and in their 50’s. You would then take a random sample for each of

these groups.

• This type of sampling is based on the logic that, if the population can be

segmented based on a single feature or characteristic, it is much easier to get an

estimate and increase accuracy in how well the sample represents the population

• It is important to note that the factor which we use to categorize the population
Stratum – Singular
into different stratum is related to the variable being studied, is clearly Strata - Plural
identifiable and homogenous with respect to the characteristic being stratified

(for example if we use age group as the segmenting factor, then everyone in the

stratum should be of the same age group)

 2 types of stratified probability
sampling
• Proportionate Stratified • Disproportionate Stratified

Sampling Sampling

• The sample size which is

• The sample size which is
selected from the stratum is
selected from the stratum
NOT proportionate to the size
is proportionate to the size
of the stratum in the
of the stratum in the
population
population
• Useful when smaller stratum
• Ensures fair representation
needs to be emphasized
Stratum – Singular from each stratum as it more such as studies where
Strata - Plural considers the size of each minority groups are subjects
stratum
 Cluster Sampling
• A cluster sample is obtained by dividing the population into sections
or clusters, then randomly selecting one or more of the clusters and
using all its members as the members of the sample.

• This is often used when the population is large or there is a large

geographic area. For instance, let’s say you wanted to survey small
business owners in a very populated city, it would be costly and time
consuming to survey every single small business owner, so you could
create a cluster sample using zip codes, and maybe survey 2 or 3 of
the 13 different possible zip codes. Cluster samples can be efficient
and cost effective; however, there are times when the cluster does
not represent the population.

• The main difference between cluster sampling and stratified sampling

is that subgroups in the stratified sample have similar characteristics
and the subgroups or clusters in the cluster sample are intended to
vary in characteristics.
Applications of Probability Sampling

1. Social Science Research: Used in national and international surveys to understand societal trends,

attitudes, and behaviours. Examples: Polls, household income surveys.

2. Business & Market Research: Helps companies understand consumer preferences and purchasing

behaviour. Examples: Product testing, customer satisfaction surveys.

3. Medical Studies: Used in clinical trials and epidemiological research to ensure representative patient

selection. Examples: Randomized controlled trials (RCTs) for new medications.

4. Government Surveys: Essential for policy-making, economic planning, and resource allocation.

Examples: National census, labour force surveys.

5. Educational Research: Used to assess student performance, educational trends, and curriculum

effectiveness. Examples: Standardized testing analysis, student satisfaction surveys.

6. Environmental Studies: Helps in estimating pollution levels, biodiversity, and climate change effects.

Examples: Random sampling of air quality in different regions.

Definition and Principles of Non-
Probability Sampling
• Definition: Sampling method where selection is based on non-random
criteria; not every unit has an equal chance of selection.

• Principles:

1. Subjective Selection: Researcher discretion influences the sample

selection.

2. Ease & Accessibility: Often used when probability sampling is not feasible.

3. Exploratory Research: Useful for preliminary studies or when population

characteristics are unknown.
Types of Non-Probability
Sampling
1. Convenience Sampling/ Accidental Sampling:

• Selection based on ease of access and availability.

• This means to simply interview or select a person

immediately after meeting them and confirming that they
possess the characteristics needed to be a sample in our
research

• Commonly used when a sampling frame is unavailable.

• Quick and cost-effective but prone to selection bias.

• Used in market research and pilot studies.

• Example: Surveying people at a shopping mall.

2. Judgemental/ Purposive Sampling

• Researcher selects subjects based on their knowledge

and expertise

• This type of sampling is more common in qualitative

research

• The researcher will only approach people who they

believe will help them achieve objectives of their study
and those who will be willing to share information with
Quota Sampling
• The main consideration in quota sampling is the ease of access of the researcher to these
participants

• These participants are selected from a location which is convenient for the researcher

• For example, if the researcher sees any participant who is of interest to them based on factors
such as gender or age, the researcher will approach them and ask them to be a part of the
research until they meet their quota

• Advantages: Easiest way to sample participant, cost-effective, does not require any other
information like sampling frame or location of participant

• Disadvantages: The samples may not exactly be a representative of the general population as
they may have their own unique characteristics so the results cannot be generalized
Snowball Sampling
• Starts with a few participants who refer
others, expanding the sample.

• Useful for studying hidden or hard-to-

reach populations.

• Often used in sociological and criminal

research.

• Example: Studying drug users by

asking initial respondents to refer
peers.
Expert Sampling

• In this type of sampling, the respondents themselves must be experts

in the field of study which is of interest to the researcher

• The difference between judgemental and expert sampling is that in the

former, the researcher quantifies the ability of the respondents to give

appropriate information while in the later, the respondents themselves have

to be experts

• This type of sampling is used in both types of research but more in

qualitative research
Applications of Non-Probability Sampling
Qualitative Research: Used in exploratory research to gain insights into attitudes and motivations. Helps understand complex

human behaviour. Common in psychological and sociological studies. Example: Conducting in-depth interviews to understand

customer preferences.

Exploratory Research: Used when little is known about a subject and hypotheses need to be developed. Helps researchers

generate new ideas. Provides flexibility in study design. Example: Investigating emerging trends in social media usage.

Pilot Studies: Used to test research methods before conducting full-scale studies. Helps refine data collection techniques.

Identifies potential problems before large investments. Example: Conducting a small-scale survey before launching a nationwide

study.

Hard-to-Reach Populations: Used to study groups that are difficult to identify or access. Effective in researching marginalized

communities. Often employed in social work and public health research. Example: Studying the experiences of undocumented

immigrants.

Business and Market Research: Helps businesses gather consumer insights quickly. Useful for preliminary testing of new products.

Cost-effective and adaptable to dynamic markets. Example: Conducting focus groups to assess potential customer reactions.

Policy Analysis and Decision Making: Used by policymakers to collect expert opinions. Helps in crafting policies based on real-

world insights. Common in government and think tanks. Example: Consulting industry experts before forming trade regulations.
 Probability Sampling Non-Probability Sampling

Selection Method Random Non-Random

Application Used wherever a Used when quick cost-

representative sample is effective samples are required
required

Risk of Bias Low High (selection bias is

possible)

Generalization High (results can be Limited generalization

generalized to a population) possible

Data types Suitable for quantitative Qualitative research

research

Features Expensive & time consuming Faster, cheaper & easier to

implement

Level of complexity Higher (as it requires sampling Simple (no sampling frame
frame) required)

Examples Drawing 10 names randomly Surveying only your friends