Category: Teaching

There is no absolute right or wrong in using convenience sampling and quota sampling. However, in rigorous educational research, they are often discouraged because they introduce systematic weaknesses that can undermine the credibility and defensibility of findings.

Here is the precise rationale:

---

1. Threat to Representativeness (External Validity)

• Convenience sampling selects participants based on ease of access (e.g., your own students).
  • In action research, convenience sampling is usually the default because a teacher is studying the environment where he/she teaches already in. Thus, it is appropriate because action research is participatory. A teacher is the “insider” researcher, and the participants are the students or colleagues directly involved in the process one wants to improve.
  • However, a teacher cum researcher who is using action research can move from selecting participants through convenience sampling to purposive sampling. So, it is more suitable because it is often seen as more “rigorous” in action research because it ensures the data comes from the people most affected by the issue.
  • Example: A lecturer wants to explore his/her supervision in terms of its effectiveness and enhancing students’ learning experience. So, even though students who are under his/her supervision is convenience sample, yet the lecturer cum researcher can use criterion Sampling when he/she focuses only on specific criteria such as his/her research samples are students who do not have any background in education who are doing PhD in Education. In this regard, criterion sampling as a type of purposive sampling is chosen because the lecturer cum supervisor cum researcher selects participants because they meet a specific predetermined criterion and thus, it shows a more deliberate research design. It moves the study from “I just used who was there” to “I strategically chose these participants because they have the specific experience needed to solve the research problem.“

• Quota sampling ensures proportions (e.g., gender, age) but still relies on non-random selection. While the primary strength of quota sampling is its ability to mirror the population’s known distribution of certain characteristics (or controlled characteristics), ensuring these proportions (e.g., gender, age, location) are reflected in the sample, yet it may not be representative of the population for other uncontrolled characteristics (e.g., income, specific attitudes). Since selection within quotas is non-random, certain segments of a quota group might be systematically under- or over-represented.

The issue:
Neither method (convenience and quota sampling) gives every member of the population an equal chance of selection. This leads to sampling bias, meaning the sample may not reflect the actual population.

Consequence: Findings cannot be confidently generalised to the wider educational context.

Note: For action research, findings are not meant for generalisation and thus, using convenience sampling or purposive sampling suits with its nature. Action research, by its very nature, is a localized and context-specific form of inquiry aimed at solving immediate, practical problems and implementing improvements within a particular setting (such as a classroom, school, or organization). Consequently, the findings generated from an action research study are typically not intended or suitable for broad generalization to other populations or settings. Therefore, employing non-probability sampling techniques, such as convenience sampling or purposive sampling, is entirely consistent with the core principles and aims of action research.

---

2. High Risk of Systematic Bias

These methods are particularly vulnerable to:

• Selection bias (researcher chooses who is “available” or “fits”)
• Volunteer bias (participants who agree may differ systematically)
• Context bias (e.g., one class, one school culture)

Note: In education, this is critical because student performance, motivation, or behaviour can vary widely across contexts (schools, regions, SES).

---

3. Weak Alignment with Inferential Statistics

In quantitative studies, especially those aiming for hypothesis testing, prediction and generalisation, sampling design must support statistical assumptions. Thus, with convenience or quota sampling:

• You cannot justify probability-based inference
• Statistical conclusions become methodologically fragile

---

4. Limited Transferability (Qualitative Context)

Even in qualitative research, convenience sampling often produces shallow or homogeneous data and it may not capture information-rich cases. This weakens, the depth of analysis, conceptual development and credibility of interpretations

---

5. Methodological Inconsistency

Sampling must align with research intent:

• If your goal is theory generation -> you need purposeful/theoretical sampling
• If your goal is generalisation -> you need probability sampling

Using convenience or quota sampling often signals a mismatch between research design and sampling logic.

What is the exception of using convenience sampling?

1.Preliminary Research

Used to generate preliminary insights or hypotheses.

Example: A psychology lecturer in Malaysia surveying her own students to test whether a new questionnaire is understandable before wider use. This is in the stage of refining research instrument (developmental stage of a research instrument), prior to pilot study. This exploration phase, not the actual data collection and can be done in several cycles of research instrument refinement.

2. Pilot Studies / Feasibility Testing

Helps refine research tools, methods, or logistics before committing resources to a full-scale study.

Example: A research student is studying about digital usage among postgraduate university students and plan to conduct the data collection using online mode. The research student wants to test whether an online survey platform works smoothly by first distributing it to postgraduate classmates or fellow research colleagues.

3. Resource-Constrained Situations

When time, budget, or access limitations prevent random sampling.

Example: A small NGO collecting quick feedback from nearby communities due to limited funding about the effectiveness of their services which the communities receive and use.

4. Classroom or Informal Research

Used in teaching, training, or internal assessments where generalizability is not required.

Example: A statistics class at a university using classmates as a sample to practice survey analysis. It acknowledges the practical reality of being a practitioner-researcher.

What you must do (critical for assessment)

If you use convenience sampling, you must demonstrate methodological awareness:

1. Justify the choice

Explain clearly:

• Why this sample is accessible and relevant
• Why alternative sampling methods were not feasible

Example: “Convenience sampling was employed due to accessibility to a defined cohort within the institution, allowing timely data collection within the study period.”

---

2. Acknowledge limitations

Be explicit about:

• Limited representativeness
• Restricted generalisability
• Potential sampling bias

---

3. Align with research design

Ensure consistency:

• If quantitative research -> avoid strong claims of population generalisation
• If qualitative research -> emphasise contextual depth, not representativeness

In educational research, sampling design adheres to the same foundational logic as general research methodology, but it is contextualised around educational populations such as students, teachers, school leaders, or institutions. It is not a procedural afterthought; rather, it is a methodological decision that must be explicitly aligned with the research problem, research questions, and overall research design.

Sampling design serves several key functions:

• It determines who or what constitutes the source of data (e.g., learners, classrooms, schools).
• It ensures the appropriateness and credibility of the findings.
• It enables the researcher to operationalise the intent of the study whether that intent is to generalise, explain, explore, or generate theory.

Crucially, sampling design must be coherent with the research design. Different research designs imply different logics of sampling:

• In quantitative designs (e.g., experimental, survey), sampling strategies such as probability sampling (e.g., simple random, stratified) are typically employed to enhance representativeness and generalisability.
• In qualitative designs, sampling is usually purposeful and criterion-based, focusing on depth, richness, and relevance of data rather than representativeness.

For example, if the research aim is to develop a theory that explains an educational phenomenon, a Grounded Theory design is appropriate. In this case, theoretical sampling is used because participants are selected iteratively based on their potential to contribute to the emerging theory. Sampling decisions are made continuously throughout the study, guided by the evolving analysis rather than predetermined at the outset.

In short, an effective sampling design in education is not chosen in isolation as it is logically derived from the research purpose and tightly integrated with the research design, ensuring methodological congruence and the production of meaningful, defensible findings.

The process typically involves defining three levels:

• Population: The broad group of interest sharing a characteristic (e.g., all primary and secondary teachers in a country, or all undergraduate students).
• Target Population: The specific subset you can reasonably identify and study (e.g., primary teachers with a specific certification, or undergraduates at three specific universities).
• Sample: The actual individuals from that target population who are selected to participate in your study.

Educational research often relies heavily on non-probability sampling because getting a complete, randomized list of all students or teachers (a sampling frame) is rarely feasible. For example, researchers might use convenience sampling by distributing a questionnaire in a specific class or teacher’s social media group via WhatsApp or Telegram group. Alternatively, they might use purposive extreme case sampling to study specific subgroups, like selecting only Dean’s List students to understand high-achievement study habits.

Sampling design is the overarching framework of your sampling process. While a “sampling scheme” is the specific technique used to select units (like the river sampling – which is one types of convenience sampling), a sampling design includes the entire structure: the number and types of schemes, the sample size, and the relationship between participants.

An effective sampling design involves three main pillars:

1. The Sampling Scheme (The “How”)

This is the method used to select your sample from the population. It generally falls into two categories:

1. Probability Sampling: Every member has a known and equal chance of being selected (e.g., simple random or stratified sampling). This is ideal for statistical generalization.

• Non-Probability Sampling: Selection is based on accessibility or specific criteria (e.g., convenience, river, or purposeful sampling). This is used when a full list of the population is unavailable.

2. Sample Size (The “How Many”)

This determines the statistical power and the depth of your data.

Some Examples of Non-Randomised Sampling (Not Purposive Sampling)

1. Convenience Sampling

The researcher selects participants who are the easiest to reach or most available. It is often used for pilot testing or when resources are limited.

• Example 1 for convenience sampling: A university lecturer surveying students in their own classroom to get quick feedback on a new teaching tool. It is appropriate to use convenience sampling, but it is often better to use purposive sampling.
  • Convenience Sampling: You choose participants because they are easy to reach (e.g., “I am teaching this class, so I will study this class”).
  • Purposive Sampling: You choose participants because they fit the purpose of the study (e.g., “I am studying my class on how AI helps struggling writers but I will specifically select the 5 students in my class who have the lowest writing scores“).
• Example 2 for convenience sampling: A company representative standing in a shopping mall and asking passersby for their opinion on a new product.

2. Quota (Ratio) Sampling

While “ratio” is often used in stratified sampling (one type of probability sampling) in random context, in a non-random context, it usually refers to Quota Sampling. For quota sampling, the researcher ensures the sample reflects specific proportions (ratios) of certain traits in the population (e.g., 50% male, 50% female).

Example: A researcher wants to study students satisfaction of using university library. If they know the university has the ratio of 60% female and 40% male users/students, they recruit exactly 60 female students and 40 male students from the library lobby or library user log until those “quotas” are filled.

3. Snowball Sampling

Used when the target population is “hidden” or difficult to locate. Existing participants recruit or refer others from their social circle who meet the criteria.

Example: A study on the experiences of university students using neuroenhancer drugs. The researcher finds one participant, who then introduces them to others in their “community”.

Some Examples of Non-Randomised Sampling (Purposive Sampling Techniques)

1. Theoretical Sampling

It is a qualitative sampling strategy where you select new participants or data sources based on the interpretative theory that is emerging from your ongoing data analysis. It is a specific type of purposive (or purposeful) sampling. It is the principal sampling method used in the grounded theory approach. Instead of aiming to simply increase the overall sample size, the goal is to collect specific data that helps develop emergent themes, assess their relevance, and refine your concepts.

Example: A researcher studying “burnout” starts by interviewing medical students until reaching theoretical saturation (the point where concepts are dense, relationships are stable, and new data no longer adds explanatory power). After discovering that “lack of support” is a key theme, they specifically seek out medical students who work in high-support environments during their internship to see if the theory holds.

2. Deviant (Extreme) Case Sampling

This method focuses on “outliers” or cases that are unusual, highly successful, or notable failures. It is a specific type of purposive (or purposeful) sampling. Studying these extremes can provide unique insights that “average” cases cannot.

Example: A study on “academic success” that specifically interviews students with the highest possible GPAs and students who have dropped out, rather than those with average grades.

3. Intensity sampling

It is a purposive, non-random sampling strategy used to select cases that manifest the phenomenon of interest intensely, but not extremely. In qualitative research, while “Deviant Case” sampling looks at the outliers (the absolute best or worst), Intensity Sampling focuses on “rich” cases that are excellent examples of the situation of being so unusual that they no longer represent the general population.

Example: A study on the perceived impact of case-based learning on student engagement, researchers would not interview a student who hates school or a student who is a genius. Instead, they would select students who are consistently engaged and vocal in class. They provide “intense” data about how the method works because they are actively experiencing it, but they are still “normal” students.

4. Criterion Sampling

It is a specific type of purposive (or purposeful) sampling. It involves actively selecting participants who meet a predetermined, important criterion because that characteristic makes them “information-rich” regarding your research topic

Example: Selecting only individuals who have used a specific software for more than 10 hours a week for a user experience study.

5. Typical case sampling

It involves selecting participants who represent what is considered “average” or “normal” to illustrate a standard experience. It is a specific type of purposive (or purposeful) sampling.

Example: If a researcher wanted to study how a new curriculum benefits the average student, they would purposefully select only students with average grades.

6. Maximum Variation Sampling

Maximum variation sampling involves intentionally selecting participants or cases that are as different from one another as possible along a specific dimension. So, it is a specific type of purposive (or purposeful) sampling. While it might seem counterintuitive to look for such extreme differences, the primary goal is actually to identify core, shared patterns that remain consistent despite that high level of diversity. By deliberately expanding the range of variation, researchers can capture the true breadth of a phenomenon and confidently identify common themes that cut across different environments.

Example: A study exploring how teachers adapt to a new digital grading system. Instead of selecting a random group of teachers, the researcher purposefully selects a brand new first-year teacher, a mid-career teacher, and a veteran teacher with 30 years of experience. The goal is to capture as much diversity as possible along the dimension of “teaching experience.” If the researcher finds that all three of these vastly different teachers struggle with the exact same software issue, that finding is very strong because it represents a shared pattern that cuts across their differences.

7. Expert Sampling

Expert sampling is a type of purposive sampling where a researcher specifically select participants who possess a high degree of specialized knowledge or expertise in a study area. It is commonly used in the early stages of research to help shape the research questions and the study’s design.

Example: To study problematic mobile social media use among students, a researcher would not use expert sampling to select the undergraduate participants. However, a researcher might use it to select a small group of psychology professors or media researchers to review and validate the survey questionnaire before distributing it for data collection.