PROBABILITY

Simple random Systematic random Stratified random Cluster Quota

Quota sampling is a non-probability sampling method that relies on the non-

Stratified sampling is a method of sampling where the random selection of a predetermined number or proportion of units.
Systematic random sampling is also known as a probability population is divided into subgroups (strata) based on a
A simple random sample is a randomly selected subset of a To study large populations, particularly those that are widely
sampling method in which researchers assign a desired specific characteristic, and then a sample is drawn from each *There are two types of quota sampling:
Definition population. In this sampling method, each member of the
sample size of the population, and assign a regular interval stratum. This ensures that the sample is representative of the
geographically dispersed. Researchers usually use pre-existing
- Controlled: There are restrictions on the choice of samples. Researchers’
population has an exactly equal chance of being selected. units such as schools or cities as their clusters.
number to decide who in the target population will be sampled. entire population, even if certain subgroups are smaller or selection is limited.
have different characteristics. - Uncontrolled: There are no restrictions on sample choices. Researchers are free
to choose subjects at will.

Step 1. Select the Stratifying Variable:

Choose a characteristic that is relevant to the research
question and that divides the population into distinct groups.
Step 1: Define your population: This is the group from which Examples of stratifying variables include age, gender, income,
you are sampling. education level, or geographic location. - Step 1: Divide the population into strata (subgroups):
> Step 1: Define population
Step 2: Settle on a sample size: How many subjects do you Begin by dividing your sample population into exclusive subgroups so individuals
want/need to sample from the population to get a reflective can only qualify for one category.
> Step 2: Divide samples into clusters
idea of it? Step 2. Divide the Sampling Frame into Strata:
Step 1: Define the population Criteria:
Step 3: Assign every member of the population a number: If Create mutually exclusive and collectively exhaustive strata - Step 2: Determine a quota for each stratum:
Step 2: Decide sample size - As diverse as possible
the group you’re looking at consists of, say, 10,000 people, based on the chosen stratifying variable. Next, you estimate the proportions of each stratum in the population.
Step 3: Select sample randomly - Similar distribution/ characteristics similar to population
Steps - Lottery method
start lining them up and giving them numbers. Ensure that each individual belongs to exactly one stratum.
- Together cover the entire population
*For example, if there are 2,000 university students enrolled in the two programs,
Step 4: Decide the sampling interval: This can be achieved by made up of 800 (40%) education students and 1,200 (60%) economics students,
- Random number method - No overlap between clusters
dividing the population size by the desired sample size. Step 3. Draw a Systematic or Random Sample from Each your sample should comprise 40% education students and 60% economics
Step 4: Collect data
Step 5: Choose a starting point: This can be done by selecting Stratum: students.
> Step 3: Randomly select clusters to use as sample
a random number. - Systematic sampling: Choose a starting point at random and
Step 6: Identify members of your sample: If you have a starting then select every kth element from the stratum. - Step 3: Continue recruiting until the quota for each stratum is met
> Step 4: Collect data from the sample
point of 15 and a sample interval of 100, the first member of - Random sampling: Use a random number generator to select
the sample would be 115, and so forth. individuals from each stratum.
The size of the sample from each stratum should be
proportional to the size of the stratum in the population.

- Ensures all relevant subgroups of the population are 1. Low cost, time and administrations
represented in the sample, even if they are small. 2. Quota sampling does not require a sampling frame or strict random sampling
- It’s simple to use > Cluster sampling is time- and cost-efficient, especially for
1. Because it is entirely random, it minimizes the potential for - By dividing the population into strata, the researcher can techniques, which makes this method quicker and easier than other methods.
- It reduces bias samples that are widely geographically spread and would be
researchers biasing the results, even if unintentionally allocate sample sizes to each stratum based on their 3. Among non-probability sampling methods, quota sampling is the most likely
- It’s low risk. Systematic sampling carries a low-risk factor difficult to properly sample otherwise.
2. SRS is the simplest method for obtaining an unbiased variability, leading to a more precise estimate. to accurately represent the entire population, especially when you use proportional
Advatages sample. While the researchers need a list of the entire
because there’s a low chance that the data can be
- Reduce sampling error compared to simple random
> Because cluster sampling uses randomization, if the
quotas. This helps avoid over or underrepresentation, and creates a sample that is
contaminated. This is because of the even distribution of population is clustered properly, your study will have high
population, they don’t need other information about that sampling. more likely to match the population being studied.
members to form samples. external validity because your sample will reflect the
population, its subpopulations, and its features. - Can be more cost-effective than simple random sampling, as 4. The use of a quota sample allows for easier comparison between subgroups.
- Quick and Cost-Effective characteristics of the larger population.
it may require less effort to reach specific subgroups. Since you have broken your quota into strata, analysis of each strata is built into
the model.

1. For very large populations, it is difficult to have enough

elements to randomize, leading to biases in the way samples
are chosen. 1 . Since quota sampling doesn’t use random selection and the researcher
- The researcher must have prior knowledge of the population
2. When samples are chosen randomly, it can be very time- decides who is included in the sample, it can lead to research biases like selection
and be able to identify relevant strata. > Internal validity is less strong than with simple random
consuming and costly to collect data. For example, research bias.
- Can be more complex than simple random sampling, as it sampling, particularly as you use more stages of clustering.
on people may require random selection from many - There is a possibility of systematic error, which is an error 2. It is not always possible to divide the population into mutually exclusive
requires additional steps and considerations. > If your clusters are not a good mini-representation of the
individuals across different geographical locations. that is always biased towards one side of the actual value, groups. Specifically, people may belong to more than one group.
Disadvantages 3. In studies that focus on comparisons or concentrate on a either lower or higher than that value.
- Can introduce bias into the results if the strata are not population as a whole, then it is more difficult to rely upon your
3. As only specific characteristics of the population are taken into account when
defined appropriately sample to provide valid results, and is very likely to be biased.
small group within a large sample, random sampling can - May be duplicated, resulting in unrepresentative sample you stratify your sample into subgroups, inaccuracy is very possible. For example,
- It may be difficult or impossible to identify suitable strata for > Cluster sampling is much more complex to plan than other
reduce the accuracy of data regarding the specific small group a study with subgroups of gender identity and income may not accurately
a given research question. forms of sampling.
of interest. For instance, comparing consumption levels represent other traits like age, ethnicity, or location in the final sample. This can
between people over 90 and those under 30, while the also lead to information bias.
population has many different age groups, can dilute the data,
resulting in a lack of accuracy.
 NON-PROBABILITY
Purposive
Extreme case purposive Heterogenous purposive Homogenous purposive Critical case purposive Typical case purposive
Extreme case sampling involves selecting a sample of Homogeneous sampling is that type of purposive sampling A type of purposive sampling technique that is particularly Typical case sampling is a type of purposive sampling that’s
- Heterogeneous sampling is a type of purposive sampling
individuals or units that are considered extreme or unusual in which regroups the sample into similar characteristics or useful in exploratory qualitative research, research with limited useful when a researcher is looking to investigate a
used in qualitative research to capture a wide range of
the characteristics or attributes the researcher is interested in traits and then takes them as an independent sample within resources, as well as research where a single case (or small phenomenon or trend as it compares to what is considered
perspectives or characteristics within a population.
Definition studying. This type of sampling is used to understand unusual
- The goal is to include diverse participants who differ
the main sample. (e.g., a group of people that are similar in number of cases) can be decisive in explaining the typical or average for members of a population.
or exceptional experiences or characteristics within the terms of age, gender, background, occupation, etc.). phenomenon of interest. Eg: Selecting individuals who represent the typical target
significantly on key variables or characteristics relevant to the
population and are often viewed as outliers in a wider Homogeneous sampling reduces variation, simplifies analysis, market for a popular product to understand common
research question
population. facilitates group interviewing. experiences and preferences.

1. Define the Research Objectives: Clearly articulate the

purpose of your study and what you aim to learn. This will
guide the selection criteria for typical cases.
2. Identify the Population of Interest: Determine the broader
group or population that is relevant to your research topic. This
will help you understand the context within which "typical"
cases are to be selected.
3. Determine What Constitutes a "Typical" Case: Develop
criteria for what makes a case "typical" or representative of the
phenomenon you are studying. This might involve identifying
Define the Research Question: Clearly identify the specific common characteristics, behaviors, or experiences that are
phenomenon or subgroup you want to study (Xác định rõ đối prevalent in the population.
tượng nghiên cứu) 4. Select the Typical Cases: Based on your criteria, identify
Identify Key Characteristics: Determine the shared and choose cases that fit the definition of "typical." This can
Step 1: Define the research's problem Identify a standard selection sample.
characteristics that define the group you're interested in. involve reviewing records, consulting experts, or using other
Start by deciding your research problem: a specific issue, Research the research problem, identify criteria for selecting
1. Step 1: Define the Research Objective These could be demographic factors (e.g., age, gender, methods to locate suitable cases.
challenge, or gap in knowledge you aim to address in your important cases (important cases). These criteria may be
Clearly define the research question or objective. Understand ethnicity), behaviors, experiences, or other relevant attributes. 5. Recruit Participants: Contact and recruit the selected
research cases that are prominent, cruel, or have a great impact on the
what you want to explore and why capturing a broad range of Create Inclusion Criteria: Develop specific criteria that typical cases to participate in your study. Ensure that you
entire study.
perspectives or experiences is important for your study. participants must meet to be included in the sample. These provide clear information about the study’s purpose and what
Step 2: Determine your population Search and identify typical cases.
2. Step 2: Determine your population criteria should align with the key characteristics you've participation will involve.
clearly defining the population from which your sample will be Based on the identified criteria, you begin to search for
You should begin by clearly defining the population from which identified. 6. Collect Data: Gather data from the selected typical cases
taken, since this is where you will draw your conclusions from. suitable cases, possibly through literature research, secondary
Steps your sample will be taken, since this is where you will draw Recruit Participants: Use various methods to recruit
data, or information contact with experts in the field.
using your chosen research methods (e.g., interviews, surveys,
your conclusions from. individuals who meet the inclusion criteria. This might involve observations). Focus on collecting detailed and relevant
Step 3: Define the characteristics of your sample Collect data. (interviews, surveys, observations, or document
3. Step 3: Define the characteristics of your sample advertising, networking, or reaching out to relevant information that will help you address your research
Set out to identify members of the population who are likely to research) Analyze and interpret data.
Set out to identify members of the population who are likely to organizations or communities. questions.
possess certain characteristics or experiences When to use
possess certain characteristics or experiences Verify Eligibility: Once potential participants are identified, 7. Analyze Data: Analyze the data to identify patterns, themes,
When you want to focus on typical cases or those that can
4. Step 4: Collect Data verify that they meet all the inclusion criteria. This can be done or insights that are representative of the typical cases. This
Step 4: Selecting extreme or unusual units provide important information.
through interviews, surveys, or other screening methods. involves examining how well the cases reflect the common
When a large representative sample is not necessary, but
Obtain Consent: Ensure that all participants provide informed characteristics of the phenomenon under study.
Step 5: Analyze and interpret results notable, exemplary cases are needed
consent to participate in the study. This means they should 8. Interpret Findings: Interpret the results in the context of the
understand the purpose of the research, the potential risks and typical cases. Discuss how these findings contribute to
benefits, and their right to withdraw at any time. understanding the phenomenon and consider the implications
for the broader population.
9. Report Results: Present your findings, including a
description of how the typical cases were selected and how
they represent the broader context. Address any limitations
related to the non-random sampling approach and the
potential impact on generalizability.
10. Reflect on the Sampling Process: Evaluate the
effectiveness of the typical case purposive sampling approach
in achieving your research objectives. Consider whether the
selection criteria were appropriate and if any adjustments are
needed for future research.
Focused Insights: By selecting cases that are representative or
typical of a phenomenon, researchers can gain in-depth
insights into the common features and patterns relevant to
their study.

Relevance: It ensures that the sample is highly relevant to the

Select only individuals who are relevant to research purpose: research questions, as the cases are chosen for their direct
1. Comprehensive Understanding: By including a diverse range Deep Understanding: It allows for a deep exploration of a
Research focuses on the most important cases, thereby relevance to the topic of study.
1. Quality responses: the researcher is able to collect a lot of of participants, heterogeneous sampling allows researchers to specific phenomenon or group, as all participants share
gaining a better understanding of the phenomenon or problem
data from committed respondents that are experts in their explore the full scope of variations within a population, leading similar experiences and perspectives.
being studied. Efficiency: This method can be more efficient than random
fields to a richer understanding of the topic Efficiency: By focusing on a specific group, researchers can
Less costly, more convenient: By focusing on a small number sampling because it targets individuals or cases that are most
Advatages 2. Minimize the margin of error: only the best-fit candidates 2. Rich and Detailed Data: The inclusion of diverse participants collect more detailed and relevant data, potentially leading to
of typical cases, time and cost can be saved compared to likely to provide the information needed, potentially reducing
make the cut often results in richer and more detailed data. This richness more efficient research.
studying a large sample. the time and resources required for data collection.
3. Time and cost efficiency: the pool of candidates is smaller, can lead to deeper insights, especially in qualitative research Feasibility: It can be more feasible than heterogeneous
Flexibility: allows for tailoring sampling criteria based on the
only pay for the engagement you need where understanding the complexity of human experiences is sampling, especially when studying rare or hard-to-reach
specific objectives of the study. Depth of Understanding: Allows for a deep understanding of
key. populations.
the typical characteristics of a phenomenon, as it focuses on
cases that exemplify the most common or average
experiences.

Practicality: Often practical when dealing with specialized or

hard-to-reach populations, where a representative sample is
difficult or impossible to obtain.
Limited Generalizability: Since the sample is not randomly
selected, the findings are not easily generalizable to the
broader population. The results are specific to the cases
chosen and may not reflect the diversity of experiences or
No guarantee that chosen sample are true representative of perspectives within the population.
the population -> Limited generalizability: Because the
sample was not randomly selected and is not representative of Selection Bias: There is a risk of selection bias, as the
1. The diversity of data can make analysis more complex and the entire population, the results cannot be generalized with researcher’s judgment about what constitutes a "typical" case
1. prone to researcher's bias: a purposive sample is created Limited Generalizability:The findings may not be
time-consuming, as researchers need to account for the wide certainty. can influence the outcome. This bias can affect the validity of
based on the judgement of the researcher representative of the entire population, as the sample is
range of variables or perspectives present in the sample Limited comparison: By focusing on special cases, the results.
2. lmited generalizability: As the sample is not selected restricted to a specific subgroup.
2. Potential for Bias: Since participants are purposefully comparisons with other cases or applying results to broader
randomly, the study’s results may not be generalizable to the Bias: The choice of characteristics to define the homogeneous
Disadvantages larger population -> dẫn đến 1 số nhóm có thể bị exclude selected, there is a risk of researcher bias influencing which
group can introduce bias into the research.
populations can be difficult. Overemphasis on Common Features: Focusing only on typical
individuals are included, potentially skewing the results. Highly applicable results: Identifying truly critical cases cases might overlook important variations or extreme cases
3. ethical consideration: The researcher must ensure that the Difficulty in Recruitment: Finding participants who meet all the
3. Challenges in Identifying Patterns: Because of the variability requires extensive knowledge and can be complex -> the that could provide valuable insights into the phenomenon
study is conducted ethically and that the privacy of the inclusion criteria can be challenging, especially for rare or
in the data, it can be challenging to identify clear patterns or researcher needs experience and high analytical ability to being studied.
participants are protected hard-to-reach populations.
trends, which may lead to less conclusive findings. process and interpret the data accurately.
Researcher bias: Sampling based on subjective judgment can Subjectivity: The process of selecting cases based on what is
lead to bias, affecting the objectivity of the study. considered "typical" involves a degree of subjectivity, which
can introduce biases into the research process.

Limited Scope: May not capture the full range of diversity

within the population, potentially leading to incomplete or
skewed understanding of the research topic.
 NON-PROBABILITY
Purposive Volunteer Haphazard
Theoretical Snowball Self-selection Convenience

Convenience sampling is a non-probability sampling method

> Defined as the process of collecting, coding and analyzing
where units are selected for inclusion in the sample because
data in a simultaneous manner in order to generate a theory. Snowball is a non-probability sampling technique where A self-selected sample is simply one in which the respondents
they are the easiest for the researcher to access. -> là cách
Definition This sampling method is closely associated with grounded existing study participants recruit future participants from put themselves forward for participation in a survey or similar
chọn mẫu mà người nghiên cứu chỉ cần chọn một đặc điểm
theory methodology. Aimed at generating and developing among their acquaintances. form of research
phù hợp với nghiên cứu và thực hiện phỏng vấn những người
theoretical data
có thể tiếp xúc được

> Step 1: Make initial decisions are made regarding specific Identify Initial Participants (Seeds): Start by selecting a small
individuals or groups of people who have knowledge about the group of initial participants (seeds) who meet the criteria of
research area. the population you wish to study. These individuals are often
chosen through convenience sampling.
Step 1: Publicise your need for units or cases
> Step 2: Analyse the initial data until theoretical ideas start to - Step 1: Determine your research question and your target
You need to let potential applicants or organisations know
emerge and particular concepts arise. Interview Initial Participants: Conduct interviews or surveys population.
about your study. This will involve some kind of advertising or
with the selected participants, gathering the necessary data
promotion, whether print media, the radio, an online notice
> Step 3: Choose further participants, events or situations on for your study. - Step 2: Develop a questionnaire that helps answer your
board, or some other medium
the basis of theoretical ideas and concepts revealed in the question. Consider including quantitative and qualitative
Steps previous stage. Ask for Referrals: After the interview, request each participant questions for more insights.
Step 2: Check the relevence of units and invite or reject them
to refer others who meet the same selection criteria. The goal
Not all applicants will be relevant to your study. They may have
> Step 4: Continue with steps 2 and 3 above until theoretical is for each participant to introduce additional participants from - Step 3: Implement the survey through your chosen
not all read or understood what the study is about.
saturation is reached. Theoretical saturation “signals the point their social network or community. communication methods such as email, social media, in-
Furthermore, they may not be the type of applicants you are
in grounded theory at which theorizing the events under person, mail and/or telephone.
looking for
investigation is considered to have come to a sufficiently Contact Referred Individuals: Reach out to the referred
comprehensive end”. individuals, inviting them to participate in the study. Interview
or survey these new participants just as you did with the initial
group.

Reliability: By leveraging existing social relationships, it is easy

to find participants with mutual trust, which can improve
participation rates and data reliability.

Effective in special case: effective in studying hard-to-reach or

> The possibility to strengthen the rigor of the study if the
hidden groups, e.g. minority communities, people involved in 1. High levels of simplicity and ease
study attempts to generate a theory in the research area. 1. reduce the amount of time necessary to search for
illegal activities, or rare groups 2. Less time and cost required: It can be conducted without
> The application of theoretical sampling method can provide appropriate units
any financial costs in some cases. A survey can be sent online
a certain structure to data collection and data analysis 2. Cost effective since the researcher doesn't have to invest
Saving time and human resources: Because participants help which requires no travel.
Advatages processes, thus addressing one of the main disadvantages of
find additional subjects, this method can save time and
money in finding units
3. Suitable for Preliminary Research: This method provides
qualitative methods that relate to lack of structure. 3. The units are most likely to be committed to take part in the
resources, compared with other sampling methods. samples that are readily available which makes it easy for
> This type of sampling usually integrates both, inductive and study, and have a greater willingness to provide meaningful
researchers to test out ideas in pilot studies before launching
deductive characteristics, thus increasing comprehensiveness insights.
Flexibility: This method allows researchers to flexibly adjust more formal studies.
of studies.
the sample size, depending on the number of people referred
and the needs of the study -> flexible to achieve the desired
sample size.

Lack of representativeness: Snowball sampling does not

produce a random sample, so the results may not be
representative of the entire population. The sample may be
limited to a particular social group, leading to bias in the
results.

> Because it is a highly systematic process, application of Difficulty controlling sample size: In some cases, relying on 1. self selection bias: When individuals choose to participate in 1. Highest level of sampling error: Getting responses only
theoretical sampling method may require more resources such participants to refer others may not achieve the desired a study group, they’re more likely to have specific from the participants who are easiest to contact affects the
as time and money compared to many other sampling sample size if referrals stop early or no additional participants characteristics that make them different from the rest of the accuracy of your data and runs the risk that important cases
methods. are recruited. population are not detected, leading to undercoverage bias.
Disadvantages > There are no clear processes or guidance related to the 2. Sample is not representative of population: Since the 2. Sample is not representative of population: Since the
application of theoretical sampling in practice Lack of anonymity: Since participants are referred from sample is not chosen through random selection, it is sample is not chosen through random selection, it is
> Overall, theoretical sampling is the most complicated than networks of acquaintances, participants' anonymity may be impossible that your sample will be fully representative of the impossible that your sample will be fully representative of the
other sampling methods affected, especially if they feel uncomfortable referring their population being studied. population being studied.
acquaintances.

Difficulty controlling data quality: Participants may not be

suitable subjects for the research criteria, or information
collected from social relationships may lack objectivity,
leading to inaccurate or unreliable data.