- Purposive sampling is a non-probability technique in which the researcher deliberately selects participants who fit the research question.
- It is also known as judgmental sampling, selective sampling, or subjective sampling.
- It trades statistical generalizability for depth and direct access to information-rich cases.
- It is an umbrella term covering 8 distinct methods, from maximum variation sampling to expert sampling.
What is purposive sampling?
Purposive sampling is a non-probability sampling technique in which a researcher selects participants on purpose, based on characteristics that matter to the study. Selection is driven by the research objective rather than by chance. Every participant is included because they can supply information that no one else can.
The sample exists to serve a purpose, so relevance beats representativeness at every step. That choice comes with a trade-off you accept openly: you gain depth and access to the right people, and you give up the ability to generalize your findings onto a whole population.
Purposive sampling also works without a complete sampling frame, which is a large part of its appeal. Many groups worth studying have never been listed anywhere, and waiting for a full list would end the project before it starts.
Examples
Consider a researcher studying pedagogical practices used by teachers of refugee students. Random sampling would be difficult and many teachers may confuse “refugee” and “migrant” students. Instead, the researcher selects teachers whom they personally verify are currently teaching refugee students and choose teachers from different K-12 grades in order to gain different perspectives.
Here is a second example. A researcher studying remote work policies needs to get insights from managers who have actually written and implemented one. So the researcher screens for these managers and chooses participants from different industries, different ages, different qualifications, and with teams of different sizes, in order to get richer insights.
When should I use purposive sampling?
Use purposive sampling when at least 1 of the following applies.
- Your research question targets a specific, definable subgroup.
- No complete sampling frame exists, or the population is rare or hidden.
- You want depth of understanding rather than an estimate of how common something is.
- Time, budget, or access rules out probability sampling.
- The study is exploratory, or it is a pilot ahead of a larger quantitative phase.
- Specialist knowledge is the whole point, as in expert panels and Delphi studies.
Avoid purposive sampling in the following situations.
- You need prevalence or incidence estimates for a population.
- You are running an opinion poll or an election forecast.
- Your analysis requires confidence intervals, significance tests, or a margin of error.
- Reviewers or regulators expect population-level inference.
Whichever variant you choose, write down your inclusion and exclusion criteria before recruitment starts. This prevents your peer reviewers or dissertation examiners from criticizing that you simply picked participants who suited your argument.
Good criteria are specific enough that someone else could apply them. Don’t use vague terms like “experienced teachers” but put numbers to them where possible, like “teachers with minimum 10 years experience in teaching middle school”.
Purposive sampling methods, with examples
Purposive sampling is an umbrella term rather than a single procedure. Researchers choose a specific variant depending on what they want the sample to do. The 8 most widely used variants are described below.
1. Maximum variation sampling
- What it is: You deliberately select cases that differ as widely as possible on key characteristics. Patterns that survive across very different participants are likely to be robust. It is also called heterogeneous sampling.
- Example: Interviewing first-year students from different backgrounds (urban, rural, international, first-generation learners) about the transition to university.
2. Homogeneous sampling
- What it is: Every participant shares a defining trait, which allows deep exploration of 1 subgroup.
- Example: A focus group made up only of single fathers with children under 5.
3. Typical case sampling
- What it is: You select cases that are “average” or normal for the population, in order to describe what is usual to an audience unfamiliar with the setting.
- Example: Profiling 1 mid-sized district school to show the standard rollout of a new curriculum.
4. Extreme or deviant case sampling
- What it is: You select unusual cases at either end of a spectrum, because the phenomenon under study is often most visible there.
- Example: Studying the 5 hospitals with the lowest infection rates alongside the 5 with the highest.
5. Critical case sampling
- What it is: You select a single decisive case, chosen so that a finding there extends logically to other settings. The reasoning runs like: if it happens here, it will happen anywhere.
- Example: To test whether a low-cost teaching intervention works, you choose the most under-resourced school in the district.
6. Expert sampling
- What it is: Participants are chosen for demonstrable expertise. This is common in Delphi studies and in fields where published evidence is still thin.
- Example: Convening 12 non-profit directors to understand the challenges in grant application writing for mid-sized environmental non-profits.
7. Total population sampling
- What it is: The whole population is studied, because it is small, clearly defined, and fully identifiable.
- Example: All 23 licensed practitioners of a rare surgical technique in 1 country.
8. Confirming and disconfirming case sampling
- What it is: You seek out cases that support an emerging theory and cases that contradict it, then check how well the theory holds. This is typically used late in analysis.
- Example: After finding that flexible hours reduce turnover, you recruit participants from 3 teams that have flexible hours and high turnover anyway.
Choosing the right purposive sampling method
The table below matches each method to the research goal it serves best.
| Method | Best suited to |
| Maximum variation | Finding patterns that hold across diverse cases |
| Homogeneous | Deep exploration of a single subgroup |
| Typical case | Illustrating what is normal or routine |
| Extreme or deviant case | Understanding outliers, failures, and standout successes |
| Critical case | Building a logical argument from 1 decisive case |
| Expert | Forecasting or judgment where evidence is limited |
| Total population | Small, fully identifiable populations |
| Confirming and disconfirming | Stress-testing an emerging theory |
What is the difference between purposive and random sampling?
In random sampling, every member of the population has a known and equal chance of being chosen; in purposive sampling, that probability is unknown, because the researcher decides who enters the sample. Neither technique is better than the other. They just answer different questions. Random sampling answers how many and how much. Purposive sampling answers how and why.
| Feature | Random sampling | Purposive sampling |
| Basis of selection | Chance (probability sampling) | Researcher judgment against set criteria |
| Sampling frame | Required | Not required |
| Generalizability | Statistical, to the whole population | Analytical, to similar cases or settings |
| Typical sample size | Large, often in the hundreds | Small, often under 50 |
| Typical study design | Quantitative | Qualitative |
| Cost and speed | Higher cost, slower | Lower cost, faster |
| Main weakness | Sampling error | Researcher bias |
One consequence matters more than the rest. Because selection probabilities are unknown, purposive samples do not support inferential statistics. You cannot calculate a margin of error or a confidence interval from them, so findings are reported as transferable to similar contexts rather than as generalizable to a population.
Example
Suppose you want to know what share of a university’s students skip breakfast. That is a random sampling question, because you need a percentage that holds for the whole student body.
Now suppose you want to know why students who skip breakfast do so. That is a purposive sampling question. You would recruit 15 students who already skip breakfast regularly, because nobody else can explain the reasoning behind the habit.
Advantages and limitations of purposive sampling
Advantages
- Efficient: it costs less and takes less time than probability sampling, because you recruit fewer people.
- Information-rich: every participant is chosen for what they can contribute, so interviews yield more usable material.
- No sampling frame needed: it works when no full list of the population exists.
- Suited to rare populations: small or hard-to-reach groups are workable, where random sampling is not.
- Flexible: criteria can be adjusted as findings emerge, which supports iterative qualitative designs.
- Supports theory building: it helps researchers reach saturation and develop concepts in depth.
Limitations to be aware of
- Researcher bias: the same judgment that makes the sample relevant can also skew it.
- No statistical generalizability: results cannot be projected onto a wider population.
- Hard to replicate: another researcher may apply the same criteria and assemble a different sample.
- Open to challenge: reviewers push back quickly when selection criteria are vague or undocumented.
These limitations can be reduced with a few standard practices.
| Limitation | How to reduce it |
| Researcher bias | Set and publish selection criteria before recruiting |
| Hard to replicate | Keep an audit trail of every sampling decision |
| Hidden assumptions | Include a reflexivity statement in the write-up |
| Weak single-source evidence | Triangulate with documents, observation, or survey data |
Purposive sampling vs snowball sampling
These 2 techniques share a lot of ground, which is why they are often confused.
- Both are non-probability techniques.
- Both are used heavily in qualitative research.
- Both work when no sampling frame is available.
The key difference is who controls recruitment. In purposive sampling, the researcher decides who enters the sample. In snowball sampling, existing participants refer the next wave, so recruitment travels along social networks.
| Feature | Purposive sampling | Snowball sampling |
| Who selects | The researcher | Existing participants, through referral |
| Ideal use case | A defined subgroup you can identify | Hidden or stigmatized populations |
| How it grows | Against fixed criteria | Wave by wave, through networks |
| Main risk | Researcher bias | Network homogeneity: People refer others like themselves, so a sample can drift toward 1 tightly connected circle and miss everyone outside it. |
In practice the 2 techniques are often combined. Researchers purposively select a small set of seed participants, then ask those seeds for referrals. This hybrid is what many published studies actually do, and it is worth naming explicitly in your methods section.
Frequently asked questions
Is purposive sampling qualitative or quantitative?
Purposive sampling is predominantly qualitative, but it is not exclusive to qualitative research. The fit with qualitative work is natural, because qualitative studies look for information-rich cases rather than representative ones.
It also is used in quantitative and mixed-methods research, in these situations.
- Pilot studies and instrument pre-testing.
- Expert surveys, including Delphi rounds.
- Reaching hard-to-reach populations in a mixed-methods design.
How many respondents are needed in purposive sampling?
There is no fixed number. Sample size is governed by data saturation, which is the point at which new participants stop producing new themes.
These commonly cited ranges are useful starting anchors.
- Phenomenological studies: roughly 5-25 participants.
- Grounded theory studies: roughly 20-60 participants.
- Case studies: often 1-5 cases, with several participants inside each.
Research on homogeneous groups frequently reaches saturation within the first dozen or so interviews. Heterogeneous samples and broader research questions push the number higher, sometimes well past 30.
Whatever number you land on, explain how you judged that saturation had been reached. Reviewers care more about that reasoning than about the final count.
What is another name for purposive sampling?
Some of the other names for purposive sampling in the literature are
- Judgmental or judgment sampling: the most frequent alternative, standard in textbooks and market research.
- Selective sampling
- Subjective sampling
- Authoritative sampling: used occasionally when a subject expert chooses the sample.
Two terms are not synonyms for it, although they are often mixed up with it: convenience sampling, which selects whoever is easiest to reach, and quota sampling, which fills preset numeric targets for each subgroup.
Comment