0Quick Answer

1What Is Purposive Sampling?

Purposive sampling, also called judgmental or selective sampling, chooses cases deliberately because they possess characteristics the research question requires. A study of how hospitals recover from cyber attacks selects hospitals that have been attacked. A study of exceptional teaching selects teachers identified as exceptional. Nothing is random, and that is the point: random selection would waste most of the sample on cases with nothing to say about the question.

This makes it a non probability design, but for a fundamentally different reason from convenience sampling. Convenience sampling has no rationale beyond access. Purposive sampling has an explicit analytical rationale, and that rationale is the method. When a reviewer evaluates a purposive sample they are not asking whether it is representative, because it was never meant to be. They are asking three things: was the strategy appropriate to the aim, were the selection criteria stated and applied consistently, and is there enough information in the sample to support the claims being made.

The logic is transferability rather than generalisability. A probability sample lets you infer to a population. A purposive sample lets a reader judge whether your findings might apply to their own setting, which requires you to describe your cases in enough detail for that judgement to be possible. That is why a thin methods section damages a purposive study far more than a small n does.

Figure 1.1 Purposive sampling spends the whole sample on cases that can answer the question.

2Setup: Strategy, Criteria and Cases

Information power scorecard

Malterud's five dimensions. Each answer scores 0 for a factor that increases the sample size needed, 1 for neutral and 2 for a factor that reduces it. A higher total means fewer cases are required.

Variation dimensions: one card per dimension

List each dimension that matters to your question, the levels you need covered, and which level each selected case falls into. The tool then shows you exactly which cells are still empty. Dimension names are editable.

3Results

4Interpretation of Results in Detail

How to read each number

The recommended strategy. The tool maps your stated aim onto one of fifteen recognised purposive strategies. This matters more than any other output, because the commonest failure in purposive sampling is a mismatch: selecting typical cases when the question is about failure, or maximally varied cases when the question needs depth in one setting. If the recommendation surprises you, that is worth pausing over before recruiting anyone.

The coverage percentage. For each dimension the tool compares the levels you said you needed against the levels your cases actually occupy. Coverage is the share of required levels with at least one case. This is the single most checkable claim in a maximum variation study: if you write that your sample spans urban, peri-urban and rural settings, a reader can verify it in one glance at the matrix, and so can a reviewer.

The empty cells list. Every level with no case is named explicitly. An empty cell is not automatically a flaw, because some combinations may not exist in reality, but an unexplained empty cell is. The rule for writing up is simple: either fill it or explain it.

Cases per level. Coverage tells you whether a level has any case; this tells you how many. A dimension where one level holds eight cases and another holds one is technically fully covered but practically lopsided, and any comparison you draw across those levels will rest on very unequal evidence.

The information power score. This follows Malterud's model, which replaced the older habit of quoting arbitrary sample sizes. Five factors decide how many cases you need: a narrow aim needs fewer than a broad one, a dense sample of highly specific participants needs fewer than a sparse one, strong existing theory needs fewer than an exploratory study, high quality dialogue needs fewer than thin data, and a case oriented analysis needs fewer than a cross case comparison. A high score means your study carries a lot of information per case, so a smaller sample is defensible.

The suggested sample size range. The tool combines your qualitative design with your information power score to give a band, not a number. Treat it as the range you must justify within, not a target to hit. The published bands are conventions distilled from methodological reviews, and every one of them is routinely and legitimately departed from with a stated reason.

The saturation point. If you entered new codes per case, the tool finds the first case after which your chosen window of consecutive cases each added no more than your threshold of new codes. That is the case at which saturation was reached under your own stated criterion, which is exactly what a reviewer wants to see: a criterion set in advance and then applied, rather than a bare assertion that saturation occurred.

The saturation curve shape. A healthy curve falls steeply then flattens. A curve that is still descending at your last case means you stopped early, whatever your total n was. A curve that is flat from the beginning usually means your coding scheme is too coarse to detect new content, which is a different problem and is not solved by more interviews.

Cumulative codes. The running total shows how much of your final code set each case contributed. If the last three cases contributed two percent of the codes between them, that is a quantitative statement about saturation you can put in a sentence, and it is far more persuasive than the word saturation on its own.

What none of this can tell you. Coverage of the dimensions you chose says nothing about the dimension you failed to think of, and saturation of the codes your framework produces says nothing about the codes a different framework would have produced. Both metrics measure the completeness of your own design against itself. They are necessary, they are checkable, and they are not sufficient.

5How to Write Your Results in Research

Use the templates below. Each one names the strategy, the criteria and the size justification, which is what reviewers look for first in a purposive design.

Rules that make a methods paragraph pass review

1. Name the specific strategy, not just "purposive". Write "maximum variation sampling" or "critical case sampling"; the generic label tells a reader nothing.
2. Give the rationale in one sentence. Why these cases and not others is the entire method, so it belongs in the first two lines.
3. List inclusion and exclusion criteria explicitly. A bulleted list is better than prose, and it makes the selection auditable.
4. Show the variation table. Dimensions down the side, levels across the top, counts in the cells. It converts a claim into evidence.
5. Explain every empty cell. Either the combination does not exist, or you could not access it, or you chose not to pursue it. Say which.
6. Justify n by information power, not by convention. Cite the five factors and say which of them apply to your study.
7. Demonstrate saturation rather than asserting it. State the criterion you set in advance and report the new codes per case that met it.
8. Describe your cases richly enough for transferability. The reader must be able to judge whether your setting resembles theirs.
9. Say who identified the cases. If a gatekeeper nominated participants, that is a selection mechanism and must be reported.
10. Do not use the language of representativeness. Purposive samples are information rich, not representative, and mixing the vocabularies invites criticism.

Common wording mistakes and the fix

6Formulas Used

7How to Use This Tool

1. Type your study name and, in one line, the rationale for choosing these cases rather than others.
2. Select your research aim from the dropdown; the tool maps it to a named purposive strategy.
3. Choose your qualitative design so the sample size guidance uses the right conventional band.
4. Answer the five information power questions honestly. A narrow aim and dense sample genuinely do justify fewer cases.
5. Add one card per variation dimension, for example setting, seniority, or years of experience.
6. In each card's levels box, list every level you need covered, separated by commas.
7. In the textarea, list which level each selected case falls into, one entry per case, in the same case order across all dimensions.
8. Enter the new codes produced by each case if you are tracking saturation, and set your window and threshold in advance.
9. Click Assess Strategy and Coverage, then read the strategy card and the empty cells list first.
10. Copy the methods paragraph into your manuscript and download the sampling plan for your supervisor or ethics committee.

8Detailed Reference Tables

Table 8.1 The fifteen purposive sampling strategies

Table 8.2 Information power: the five dimensions

Table 8.3 Conventional sample size bands by qualitative design

Table 8.4 Interpreting coverage and saturation

9Example Results (8 Worked Cards)

10How to Collect Raw Data in the Field

10a. The 18 point field protocol for purposive sampling

Plan

1. Write the selection rationale before you recruit anyone. One sentence saying why these cases and not others. If you cannot write it, you do not yet have a purposive design.
2. Name the specific strategy from the fifteen recognised variants, and check it against your aim rather than against what is easiest to recruit.
3. Write inclusion and exclusion criteria as a numbered list, not as prose, so a second researcher could apply them to the same candidate and reach the same decision.
4. List the variation dimensions and their levels that your findings will claim to span, and build the empty matrix now so you can watch it fill.
5. Set your saturation criterion in advance: how many consecutive cases adding how few new codes will count as saturation. Deciding this afterwards is what reviewers object to.

Figure 10.1 The matrix built empty at the design stage is what turns a coverage claim into evidence.

Kit

6. Carry the criteria list and the matrix to every recruitment conversation, so eligibility is checked against the written rule rather than judged on the spot.
7. Carry a screening form that records each candidate against every criterion, including candidates you turn down.
8. Carry a recorder with spare batteries and a backup device. In a purposive design each interview is a large share of the total evidence, so a lost recording is disproportionately costly.
9. Carry two pencils and a waterproof cover for the field notebook.

Select

10. Screen every candidate against the written criteria and record the outcome, including the reason for exclusion. That record is the audit trail.
11. Fill the sparsest cells first. The natural pull is toward whoever is easiest to reach, which quietly converts a purposive design into a convenience one.
12. Record who nominated each case. If a gatekeeper suggested participants, that is a selection mechanism and must appear in the methods section.

Figure 10.2 The screening form turns inclusion criteria from a stated intention into a documented procedure.

Collect

13. Record the level of every dimension for each case at the point of consent, not from memory afterwards, because the matrix depends on it.
14. Code each transcript before recruiting the next case where the design allows, so the new codes count is real and saturation can guide recruitment.
15. Log the count of genuinely new codes per case in a running table, separating new codes from repeat codes.

Figure 10.3 The saturation log with a criterion set in advance is evidence; the word saturation on its own is not.

Figure 10.4 Only genuinely new content counts toward saturation; merges reduce the running total rather than adding to it.

Check

16. After every third case, review the matrix and redirect recruitment toward the emptiest cells while there is still time.
17. Before stopping, confirm your saturation criterion has actually been met, and confirm no required level is still empty and unexplained.
18. Write the case description table the same week, because transferability depends on readers being able to picture your cases, and that detail fades fast.

Figure 10.5 Four conditions that decide whether a purposive sample is finished, none of which is a target n.

10b. Datasheet column specification

Exclusion rule: excluded candidates never appear in the analysis but must appear on the screening sheet, because the exclusions are what prove the criteria were applied. Code rule: count only content not seen before; a familiar theme from a new speaker is a repeat, not a new code.

10c. Filled worked datasheet

What this sheet shows:

• Rows 3 and 6 are excluded candidates that still appear, because the exclusions are what demonstrate the criteria were applied consistently.
• The Setting and Career stage columns fill the coverage matrix: all three settings and all three career stages are represented by case 10.
• The New codes column falls 18, 12, 9, 6, 4, 2, 1, 0, so three consecutive cases at or below one new code are reached by case 10.
• The last three included cases contributed 3 of 52 codes, under 6 percent, which is a quantitative saturation statement.
• Two participants were nominated by a gatekeeper, which is recorded so it can be reported as a selection mechanism.

10d. Blank print ready datasheet

Download the blank sheet as a .csv file, open it in Excel, Google Sheets or LibreOffice, then print it. The criterion tick columns and the exclusion reason column are what make a purposive sample auditable, so they are pre-labelled here even though most templates leave them out.

11Which Sampling Method Should You Use?

Decision tree. Answer these in order and stop at the first yes.

1. Do you need to estimate a population quantity with a margin of error? Use a probability design; purposive sampling cannot do this.
2. Can you say why each case was chosen, in analytical terms? If not, you are doing convenience sampling and should call it that.
3. Do you want patterns that hold across very different contexts? Use maximum variation sampling.
4. Do you want depth within one tightly defined group? Use homogeneous sampling.
5. Are you testing whether something holds where it most should? Use critical case sampling, where one case can be enough.
6. Are you building theory and letting the data direct the next case? Use theoretical sampling.
7. Do you need fixed numbers in predefined subgroups for a comparison? Use stratified purposive or quota sampling.
8. Is the group hidden with no frame at all? Use snowball or, if you need estimates, respondent driven sampling.

12Troubleshooting and Common Sampling Errors

13Assumptions, Bias and Limitations

• No probability basis. Selection probabilities are unknown and deliberately unequal, so no margin of error, confidence interval or population estimate is available.
• Researcher judgement is the instrument. The quality of the sample rests entirely on the quality of the selection reasoning, which is why the rationale must be written down and defended.
• Dimension blindness. Coverage is measured only against the dimensions you thought of. The variable you failed to consider is invisible to every metric here.
• Coding scheme dependence. Saturation is saturation of your codes under your framework. A different analyst with a different framework would saturate at a different point.
• Confirmation risk. Selecting cases you expect to be informative can shade into selecting cases you expect to agree with you. Disconfirming case sampling is the standard defence.
• Gatekeeper effects. When a third party nominates cases, their preferences enter the design silently unless documented.
• Transferability, not generalisability. Findings travel only as far as a reader's judgement that their setting resembles yours, which depends on how richly you described your cases.
• Information power is a judgement, not a measurement. The score is a structured way of arguing for your n; it carries no sampling theory.

14Conclusion

What purposive sampling gives you

Purposive sampling spends the entire sample on cases that can answer the question. Where a probability design would scatter effort across a population most of whom have nothing relevant to say, purposive sampling concentrates it on the people, sites or events that carry the information. For a study of rare events, exceptional performance, expert judgement or lived experience it is not a weaker alternative to random sampling but a categorically different and more appropriate tool. It also scales down honestly: a single critical case, properly argued, can settle a question that a thousand random cases would not.

What it costs you

You give up inference to a population and you take on the burden of justification. Every case must be defensible, the criteria must be written and applied consistently, and the reasoning must be visible to a reader who was not there. The design also depends completely on the dimensions you thought to specify: coverage of your own matrix says nothing about the factor you never considered, and saturation of your own codes says nothing about the codes a different framework would have produced. Both metrics measure your design against itself.

What to check before you publish

Confirm five things: the specific strategy is named rather than the generic word purposive, the selection rationale appears in the first two lines of the methods, inclusion and exclusion criteria are listed explicitly, every empty cell in the variation matrix is either filled or explained, and the sample size is justified by information power with a saturation criterion that was set in advance and then demonstrated. Add rich case descriptions so a reader can judge transferability to their own setting, and report any gatekeeper involvement in nomination.

What to do next

If a cell is still empty, decide now whether to fill it, merge the level into a neighbouring one, or narrow the claim your paper makes about coverage. If saturation was not reached, either continue recruiting or reframe the findings as preliminary; there is no third option that survives review. And if the strategy this tool recommends differs from the one you used, that mismatch is worth resolving in writing before you submit, because a reviewer who spots it will ask the same question you would rather have answered yourself.

15Test Yourself

16Frequently Asked Questions

17Cite This Tool

18Related Tools

• Convenience Sampling Calculator and Bias Checker for scoring bias in a non probability sample.
• Simple Random Sampling Calculator for a seeded probability draw.
• Systematic Sampling Calculator for interval k with a random start.
• Stratified Random Sampling Calculator for proportional, equal and Neyman allocation.
• Proportionate Stratified Sampling Calculator with a self weighting check.

19Glossary of Terms

20References

1. Patton, M. Q. (2015). Qualitative Research and Evaluation Methods (4th ed.). SAGE. Publisher page
2. Palinkas, L. A., et al. (2015). Purposeful sampling for qualitative data collection and analysis in mixed method implementation research. Administration and Policy in Mental Health, 42(5), 533-544. https://doi.org/10.1007/s10488-013-0528-y
3. Malterud, K., Siersma, V. D., & Guassora, A. D. (2016). Sample size in qualitative interview studies: guided by information power. Qualitative Health Research, 26(13), 1753-1760. https://doi.org/10.1177/1049732315617444
4. Guest, G., Bunce, A., & Johnson, L. (2006). How many interviews are enough? An experiment with data saturation and variability. Field Methods, 18(1), 59-82. https://doi.org/10.1177/1525822X05279903
5. Hennink, M., & Kaiser, B. N. (2022). Sample sizes for saturation in qualitative research: a systematic review of empirical tests. Social Science and Medicine, 292, 114523. https://doi.org/10.1016/j.socscimed.2021.114523
6. Saunders, B., et al. (2018). Saturation in qualitative research: exploring its conceptualization and operationalization. Quality and Quantity, 52(4), 1893-1907. https://doi.org/10.1007/s11135-017-0574-8
7. Guest, G., Namey, E., & Chen, M. (2020). A simple method to assess and report thematic saturation in qualitative research. PLOS ONE, 15(5), e0232076. https://doi.org/10.1371/journal.pone.0232076
8. Etikan, I., Musa, S. A., & Alkassim, R. S. (2016). Comparison of convenience sampling and purposive sampling. American Journal of Theoretical and Applied Statistics, 5(1), 1-4. https://doi.org/10.11648/j.ajtas.20160501.11
9. Suri, H. (2011). Purposeful sampling in qualitative research synthesis. Qualitative Research Journal, 11(2), 63-75. https://doi.org/10.3316/QRJ1102063
10. Coyne, I. T. (1997). Sampling in qualitative research: purposeful and theoretical sampling. Journal of Advanced Nursing, 26(3), 623-630. https://doi.org/10.1046/j.1365-2648.1997.t01-25-00999.x
11. Glaser, B. G., & Strauss, A. L. (1967). The Discovery of Grounded Theory. Aldine. https://doi.org/10.4324/9780203793206
12. Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic Inquiry. SAGE. https://doi.org/10.1016/0147-1767(85)90062-8
13. Flyvbjerg, B. (2006). Five misunderstandings about case study research. Qualitative Inquiry, 12(2), 219-245. https://doi.org/10.1177/1077800405284363
14. Sandelowski, M. (1995). Sample size in qualitative research. Research in Nursing and Health, 18(2), 179-183. https://doi.org/10.1002/nur.4770180211
15. Vasileiou, K., et al. (2018). Characterising and justifying sample size sufficiency in interview based studies. BMC Medical Research Methodology, 18, 148. https://doi.org/10.1186/s12874-018-0594-7
16. Creswell, J. W., & Poth, C. N. (2018). Qualitative Inquiry and Research Design (4th ed.). SAGE. Publisher page
17. Miles, M. B., Huberman, A. M., & Saldana, J. (2020). Qualitative Data Analysis: A Methods Sourcebook (4th ed.). SAGE. Publisher page
18. Braun, V., & Clarke, V. (2021). To saturate or not to saturate? Questioning data saturation as a useful concept for thematic analysis. Qualitative Research in Sport, Exercise and Health, 13(2), 201-216. https://doi.org/10.1080/2159676X.2019.1704846
19. Baker, R., et al. (2013). Summary report of the AAPOR task force on non probability sampling. Journal of Survey Statistics and Methodology, 1(2), 90-143. https://doi.org/10.1093/jssam/smt008
20. Tong, A., Sainsbury, P., & Craig, J. (2007). Consolidated criteria for reporting qualitative research (COREQ). International Journal for Quality in Health Care, 19(6), 349-357. https://doi.org/10.1093/intqhc/mzm042