· By Thesisroom
Purposive Sampling: Types, Steps and How to Justify It in a Thesis
Purposive sampling explained for a thesis: the main types and what each is for, how to set inclusion criteria, seven steps to choose participants, and a worked methodology paragraph.
This guide has a printable version: open the printable purposive sampling plan
Purposive sampling is a non-random way of choosing participants because they know a lot about, or have experienced, what you are studying, so each is an information-rich case (Palinkas et al., 2015). You select people for what they can tell you about your research question, not to represent a population statistically.
The method is also called purposeful sampling. In a thesis, the choice is only half the work: your sampling section has to show who qualified, why, how you found them, and what the sample lets you claim.
What is purposive sampling?
It is a deliberate selection of cases that match the aims of your study. Campbell and colleagues (2020) give the reason: it matches the sample to the aims and objectives of the research. That match supports the rigor of the study and the trustworthiness of its data.
Palinkas and colleagues describe it as identifying and selecting people who are especially knowledgeable about or experienced with the phenomenon of interest. They add three practical qualities from the methods literature: being available, being willing to take part, and being able to talk about the experience in an articulate, reflective way.
The contrast is with probability (random) sampling. Random selection aims to make findings generalizable to a population by reducing selection bias. Purposive selection aims for depth: a smaller set of cases chosen because they can answer the question well.
Purposive sampling vs convenience sampling
A convenience sample is whoever is easiest to reach. Palinkas and colleagues list it in their table of strategies and note that, although common, it is “neither purposeful nor strategic”.
The line is not always clean. Campbell and colleagues point out that all three of their purposive studies had a convenience element, because consent is voluntary and researchers can only sample from people who agree. In one of the studies, they write that the very specific criteria for taking part are what made the design purposive.
In Ahmad and Wilkins’s (2024) interviews with thirteen professors and senior research experts, only two accepted occasional convenience sampling; the rest criticized it as damaging to trustworthiness.
What are the types of purposive sampling?
The types differ in what they are for. The table below follows Table 1 in Palinkas et al. (2015), with thesis examples we invented to show each one.
| Type | What it is for | Invented thesis example |
|---|---|---|
| Maximum variation | Find patterns that hold across very different cases | Nurses from urban, rural and remote clinics, to see what they share despite the differences |
| Homogeneous | Describe one subgroup in depth, simplify analysis, suit group interviews | First-generation doctoral students in one engineering faculty, for a focus group |
| Typical case | Show what is normal or average to readers unfamiliar with the setting | A mid-sized secondary school chosen because it resembles most schools in the region |
| Extreme or deviant case | Learn from unusual successes or failures | The two clinics with the best and worst records on a new hand-hygiene protocol |
| Intensity | Like extreme case, but rich examples rather than the most extreme ones | Teachers who changed their assessment practice markedly, though not the outliers |
| Critical case | Allow logical generalization: if it holds in this case, it likely holds in others | A hospital that tried hardest to adopt a guideline and still dropped it |
| Criterion | Select every case that meets a set criterion of importance | All program leads who ran the intervention in its first year |
| Snowball | Reach cases through people who know other suitable people | Asking interviewed union organizers to name other organizers |
| Confirming and disconfirming | Test emerging patterns by seeking cases that fit or contradict them | Late in data collection, seeking a manager who disagrees with the dominant theme |
Palinkas and colleagues group the strategies by what they do to the range of cases. Maximum variation, extreme case, intensity and purposeful random sampling widen the range of variation. Homogeneous, typical case, criterion and snowball sampling narrow it and focus on similarities. Their Table 1, though, lists extreme or deviant case under the similarity group.
Stratified, cell and quota designs
When a study has several groups, Robinson (2014) describes three ways to make sure each group appears. A stratified sample sets fixed numbers for separate categories, such as gender by age band. A cell sample allows categories to overlap, like a Venn diagram.
A quota sample sets only a minimum for each category, which is easier to recruit. Robinson stresses that every stratification category needs a clear theoretical reason to expect the groups to differ.
Is snowball sampling a type of purposive sampling?
The sources disagree, so say which view you follow. Palinkas and colleagues list snowball as a purposeful strategy that emphasizes similarity. Robinson treats it as a recruitment route, a way of sourcing participants by referral once your strategy is set.
Either way, a referred person still has to meet your criteria. Ahmad and Wilkins report that their experts suggested snowball referral to solve access problems, but stressed sticking to the predetermined selection criteria to avoid bias.
How is purposive sampling different from theoretical sampling?
Theoretical sampling happens during data collection and analysis: you choose the next cases based on what your provisional analysis shows (Robinson, 2014). It comes from grounded theory. Robinson counts it among the purposive strategies, one that runs during analysis rather than before it. Ahmad and Wilkins found the two are often confused, because grounded theory studies usually start with purposive criteria and then switch to theoretical sampling.
When does purposive sampling fit a thesis?
It fits when your question asks how, why or what something is like for the people involved. It suits interview studies, case studies and the qualitative strand of a mixed methods thesis.
It does not fit a question that asks how common something is in a population, or that tests a hypothesis statistically. Those need probability sampling, because the claim depends on the sample representing the population.
What your sample lets you claim
Robinson calls the boundary set by your criteria the sample universe: the full set of people who could legitimately be in the study. Your findings apply, logically, to that universe and not beyond it. A study that claims more than its sample universe supports loses credibility.
The trade-off is between similarity and variation. A homogeneous sample keeps findings close to a defined setting, which suits methods such as interpretative phenomenological analysis. A heterogeneous sample supports wider claims, but the diversity can make shared themes harder to find (Robinson, 2014).
In a mixed methods thesis
Your qualitative sample can be drawn from your survey sample. Palinkas and colleagues note that criterion sampling can pick cases from standardized questionnaires for in-depth follow-up, and Robinson gives an example of using questionnaire scores to choose interviewees.
How do you choose participants for a purposive sample?
Work from the question to the people, then from the people to the recruitment plan. These seven steps draw on Robinson’s four-point approach (sample universe, size, strategy, sourcing) and the framework in Ahmad and Wilkins.
- Start from what the research question needs to hear about. Palinkas and colleagues frame it as two questions: why recruit these people for this study, and what qualifies them to address its aims? If your research questions are still loose, fix them first, because the criteria come from them. Thesisroom’s Scope Check reads a research proposal and asks about the places where what it asks for and what it plans do not obviously meet.
- Write your inclusion and exclusion criteria. Inclusion criteria name what a person must have; exclusion criteria name what disqualifies them (Robinson, 2014). Give each criterion a reason tied to the question. Ahmad and Wilkins found inclusion and exclusion criteria were the most repeated codes in their experts’ interviews.
- Decide whether you need similarity or variation, then pick the strategy. For depth in one group, look at homogeneous, typical case or criterion sampling. For patterns across differences, look at maximum variation or extreme case sampling. Use the table above to match the type to the job.
- Set a provisional size range, not a fixed number. Robinson suggests giving an approximate range with a minimum and a maximum at the design stage. How to justify the numbers is a separate argument, covered in the proposal methodology guide and the guide to defending your sample size.
- Plan how you will find people. Name your channels: advertising, a gatekeeper who controls access to an organization, or referral. Robinson notes that people who volunteer may differ from those who do not (self-selection bias), so plan how you will note it.
- Analyze while you recruit, and adjust within your criteria. The range of variation is often unknown at the start, so Palinkas and colleagues note that an iterative approach of sampling and re-sampling is usually recommended. Ahmad and Wilkins’s experts likewise urged analyzing data alongside collection, since issues can emerge that require changing the sampling approach.
- Keep a record of every sampling decision. Note who was invited, who was excluded and why, who was referred, and any change to your plan. This record becomes the paragraph your examiners read.
How do you write purposive sampling in the methodology chapter?
Name the strategy, then justify it from the question. Palinkas and colleagues’ first recommendation is to describe the sampling strategy clearly and give its rationale, noting that many mixed methods studies in health services research do not.
What to report
Robinson argues that transparency comes from being explicit about all four points. Include these in your sampling section:
- The strategy: its name, the source you follow, and why it suits the question better than the obvious alternative.
- The sample universe: your inclusion and exclusion criteria, each with its reason.
- The size: the range you planned, what you reached, and the logic behind it.
- Sourcing: where and how you recruited, including gatekeepers and referrals.
- Changes: anything you adjusted during data collection, and why.
- Reach: what the sample lets you claim, and the bias that voluntary participation may add.
Campbell and colleagues tie these details to trustworthiness. Clear inclusion and exclusion criteria, with demographic information, help readers judge transferability and support dependability; stating why you chose the criteria supports confirmability.
A real example of criteria
In one PhD study described by Campbell and colleagues, the participants were child and family health nurses. Each had to be currently employed in that role with at least 2 years of recent experience, within the last 5 years. Nurses without that experience were excluded, so the views gathered came from experienced practitioners.
Worked example paragraph (invented)
The paragraph below is invented to show the parts. It is written in the past tense for a thesis; in a proposal, write it in the future tense, inside the methodology section described in this section-by-section guide to the PhD proposal format.
Participants were selected by purposive sampling, because the study asks how early-career teachers adapt lessons for mixed-ability classes in rural secondary schools. Inclusion criteria were teaching in a rural state secondary school, being in the first three years of full-time teaching, and teaching at least one mixed-ability class. Teachers on short cover contracts were excluded, as they do not plan lessons across a full term. Within these criteria, maximum variation sampling was used across subject and school size, so that patterns shared across these differences could be identified. Head teachers in two districts acted as gatekeepers and circulated the invitation. Three further participants were referred by colleagues and checked against the same criteria before being invited. Recruitment and analysis ran in parallel within a planned range of twelve to eighteen interviews. Because participants volunteered, teachers most willing to discuss their practice may be over-represented, and this is discussed in the limitations.
What each sentence does:
- Sentence 1: names the strategy and ties it to the research question.
- Sentences 2 and 3: set the sample universe, with a reason for the exclusion.
- Sentence 4: names the type of purposive sampling and its purpose.
- Sentences 5 and 6: describe sourcing, and show that referrals met the same criteria.
- Sentence 7: gives the planned range and shows the iterative approach.
- Sentence 8: names the self-selection bias and points to where it is handled.
What mistakes weaken a purposive sample?
Each of these five comes from naming the strategy without doing the work it requires.
- A convenience sample labeled purposive: Campbell and colleagues warn that novice nurse researchers may default to a convenience sample for the wrong reasons. If access drove your sample, Robinson’s advice is to define the sample universe as demographically and geographically local, and limit your claims to it.
- No stated rationale: Palinkas and colleagues re-examined 28 implementation studies and found only 5 made a specific reference to purposeful sampling. Twenty did not describe their sampling strategy, though a rationale could be inferred from who was recruited.
- Purposive and theoretical sampling mixed up: if your criteria changed because of your analysis, say that you moved to theoretical sampling and when.
- Claims beyond the sample universe: a sample of experienced nurses in one service cannot speak for all nurses. Name the reach in your limitations section.
- Referrals that skipped the criteria: a snowball chain can drift from your sample universe unless you check each referral against the criteria.
The same choices come up again in the full chapter. The guide on how to write a methodology chapter shows how to present the sample as a justified choice among the others in your design.
Printable purposive sampling plan
Fill in this plan before you recruit, and keep it with your ethics documents.
| Criterion | Include or exclude | Why it matters for the question | How you will check it |
|---|---|---|---|
| Planning item | Your answer |
|---|---|
| Research question this sample must answer | |
| Strategy (name and source) | |
| Similarity or variation, and why | |
| Sample universe in one sentence | |
| Provisional size range (minimum and maximum) | |
| Recruitment channels and gatekeepers | |
| Referral rule (how referrals are checked) | |
| Likely self-selection bias | |
| When you will review the sample during analysis |
- Every criterion has a reason tied to the research question
- The strategy is named and cited in the methodology
- Referred participants are checked against the same criteria
- Each change to the sample is logged with its reason
- Claims in the findings stay inside the sample universe
Frequently asked questions
What is purposive sampling in simple terms?
It means choosing participants on purpose, because they have the knowledge or experience your research question needs. You do not pick them at random, and you do not aim to represent a whole population. Each person is chosen to be an information-rich case.Is purposive sampling qualitative or quantitative?
It is mainly used in qualitative research, where the aim is depth. In mixed methods studies, a purposive qualitative sample can be drawn from a larger survey sample. Studies that estimate how common something is, or test hypotheses statistically, need probability sampling instead.What is the difference between purposive and convenience sampling?
A convenience sample is whoever is easiest to reach. A purposive sample is chosen against stated criteria tied to the research question. Recruitment may still depend on who volunteers, but in a purposive sample every participant must meet the criteria, and the criteria have reasons.Is snowball sampling a type of purposive sampling?
Sources differ. Palinkas and colleagues (2015) list snowball sampling as a purposeful strategy. Robinson (2014) treats it as a way of recruiting participants by referral. In either case, check each referred person against your inclusion and exclusion criteria.Can you generalize findings from a purposive sample?
Not statistically. Your findings apply to the sample universe your criteria define, the group of people who could legitimately have been in the study. A clearly described sample lets readers judge whether the findings transfer to their own setting.Is purposive sampling the same as theoretical sampling?
Not quite. Purposive sampling sets criteria before data collection. Theoretical sampling, from grounded theory, chooses the next cases based on what the ongoing analysis shows. A grounded theory study often starts with purposive sampling and then moves to theoretical sampling.Sources
- Palinkas et al. (2015), Purposeful sampling for qualitative data collection and analysis in mixed method implementation research
- Campbell et al. (2020), Purposive sampling: complex or simple? Research case examples
- Robinson (2014), Sampling in interview-based qualitative research: A theoretical and practical guide (University of Greenwich repository)
- Ahmad and Wilkins (2024), Purposive sampling in qualitative research: a framework for the entire journey (University of Bath repository)
How this guide was made: Thesisroom drafted this guide with the help of AI writing tools. Every fact, figure and source in this post was checked against its original source before publishing.