Skip to main content
v2026.11,986 entries · CC-BY 4.0

Case Study Examples: Worked Examples of Every Case Study Type

Illustrative worked examples of intrinsic, instrumental, collective, explanatory, exploratory, descriptive, and critical-instance case studies, each with a research question, case-selection rationale, data sources, and what the design can and cannot conclude.

Written and maintained by CASRAI Editorial Board

Published · Last updated

On this page

Research case study examples show how a bounded case, such as a hospital, programme or event, is investigated in depth with several data sources to answer a “how” or “why” question. Below are worked examples of each design type, four published health-services case studies compared by design, and the staged approach, limits and safeguards that published guidance recommends.

When people search for “case study examples,” they’re usually looking for one of two very different things: business case studies (customer success stories used in marketing) or research case studies (a formal qualitative design used to investigate a bounded, real-world phenomenon in depth). This guide covers the second sense. It assumes you already understand roughly what a case study is; if you need the full methodological grounding first — what qualifies as a case study, when to choose the method, and how single- and multiple-case designs work — see CASRAI’s Case Study Research Method guide. If you’ve already designed your study and need to write it up, see How to Structure a Case Study Paper in Social Science Research.

What follows are worked, illustrative examples of each major case study type, showing how the research question, case-selection rationale, data sources, and conclusions differ across types.

A note on the examples below. Every scenario in this guide is an illustrative composite written by CASRAI to demonstrate design principles — none describes an actual published study, institution, or person. Where a real methodologist’s framework is referenced (Yin, Stake, Eisenhardt), that attribution is real; the scenarios illustrating it are not.

Case Study Examples by Type

A case study example is a worked illustration of an in-depth investigation of one bounded case, such as a hospital, program or event, using several data sources to answer a “how” or “why” question. The seven examples below cover the main research designs: intrinsic, instrumental, collective, explanatory, exploratory, descriptive and critical instance. Each shows the research question, why the case was chosen, the data collected and what the design can conclude.

Case study designs are usually classified along two independent dimensions: Yin’s typology, which groups cases by the kind of question they answer (exploratory, descriptive, explanatory), and Stake’s typology, which groups cases by why the case was chosen (intrinsic, instrumental, collective). A single study is typically described using one label from each dimension — for example, an “explanatory instrumental” case study. The examples below are organized by type, with the research question, case-selection rationale, data sources, and what the design could and could not conclude made explicit for each.

Intrinsic Case Study

Illustrative scenario: A researcher studies a single rural hospital that has sustained an unusually low staff turnover rate for over a decade, not because the hospital is thought to represent hospitals generally, but because the researcher’s funding body specifically wants to understand this one institution’s internal culture and history.

  • Research question: How has this specific hospital sustained low turnover over time?
  • Case selection rationale: The case was chosen for its own sake — the client organization, not the researcher, defined the boundaries of interest.
  • Data sources: Staff interviews across tenure lengths, internal HR records, meeting minutes, and direct observation of shift handovers.
  • What the design can and cannot conclude: It can produce a rich, internally valid account of this hospital’s specific practices and history. It cannot claim these practices would reduce turnover at a different hospital — that would require an instrumental or collective design built for comparison.

Instrumental Case Study

Illustrative scenario: A researcher studying how mid-sized nonprofits adopt outcome-measurement systems selects one nonprofit that recently completed such an adoption, using it as a lens onto the broader question rather than as an end in itself.

  • Research question: What organizational conditions enable outcome-measurement adoption in resource-constrained nonprofits?
  • Case selection rationale: The organization was chosen because it is a reasonably typical instance of the broader population the researcher wants to say something about, not because of any unique quality of the organization itself.
  • Data sources: Interviews with program staff and leadership, adoption-process documents, and board meeting records covering the adoption period.
  • What the design can and cannot conclude: It can generate a theoretically grounded account of adoption conditions that plausibly extends to similar organizations (analytic generalization). It cannot establish how common those conditions are across the population (statistical generalization) — that would require a survey of many nonprofits.

Collective (Multiple-Case) Case Study

Illustrative scenario: A researcher examines four university technology-transfer offices that each restructured their licensing process within the same two-year period, comparing what changed and why across all four.

  • Research question: What common mechanisms explain why some technology-transfer restructurings increased licensing volume and others did not?
  • Case selection rationale: Four offices were selected to represent variation in institution size and prior licensing volume, following replication logic — each case is treated as a separate test of the same emerging theoretical proposition, not as a sample drawn to represent a population.
  • Data sources: Licensing-volume records before and after restructuring, interviews with office directors, and internal process documentation from each of the four offices.
  • What the design can and cannot conclude: Consistent findings across the four cases strengthen confidence that the mechanism is real (literal replication) or clarify the conditions under which it operates differently (theoretical replication). The design cannot produce a population-level estimate of how many technology-transfer offices the mechanism affects.

Explanatory Case Study

Illustrative scenario: A researcher investigates why a specific municipal open-data initiative succeeded in achieving sustained public use when several comparable initiatives in similar cities stalled after launch.

  • Research question: Why did this initiative achieve sustained public engagement where comparable initiatives did not?
  • Case selection rationale: The case was chosen because it is a positive deviant — an outcome that departs from what similar cases produced, making it useful for tracing a causal chain that a survey of many cities couldn’t isolate.
  • Data sources: Interviews with initiative staff and civic users, platform usage analytics, city council records, and comparison documentation from two of the stalled initiatives.
  • What the design can and cannot conclude: It can trace a plausible causal chain specific to this case, explaining the presumed causal links between specific design choices and sustained use. It cannot establish that the same choices would cause the same outcome in a city with a different starting context — the claim is causal-within-case, not causal-across-population.

Exploratory Case Study

Illustrative scenario: A researcher conducts a preliminary study of one research team’s use of a newly released electronic lab notebook system, before designing a larger, hypothesis-driven study of adoption across many labs.

  • Research question: What are the observable dimensions of how a lab actually incorporates a new electronic lab notebook into daily practice?
  • Case selection rationale: The case was an early-adopting lab, selected for accessibility and willingness to participate, precisely because the goal was to generate hypotheses and identify relevant variables, not to test an established theory.
  • Data sources: Field observation of lab members during a four-week period, informal interviews, and screenshots of notebook entries showing usage patterns.
  • What the design can and cannot conclude: It can surface candidate variables and hypotheses — for example, that entry frequency depends on whether the notebook is accessible from shared instruments — worth testing in a larger study. It cannot itself confirm those hypotheses; exploratory case studies are explicitly a precursor to further research, not a substitute for it.

Descriptive Case Study

Illustrative scenario: A researcher documents, in detail, how one interdisciplinary research center structures authorship decisions across its projects, without attempting to explain why the structure emerged or test any causal claim.

  • Research question: What does this center’s authorship decision-making process actually look like, in its real-world context?
  • Case selection rationale: The center was selected because detailed documentation of authorship practice in interdisciplinary settings was largely absent from the literature, making a thorough description valuable on its own.
  • Data sources: Policy documents, interviews with project leads, and a review of authorship-order decisions across the center’s recent publications.
  • What the design can and cannot conclude: It can produce a detailed, contextualized account of practice that other researchers can build on. It does not, by design, explain why the practice takes this form or whether it produces better outcomes than alternatives — those are explanatory questions outside its scope.

Critical Instance Case Study

Illustrative scenario: A researcher selects the one clinical trial site, out of a multi-site trial, that had a documented protocol deviation, specifically to test whether an existing theory about deviation causes holds up against the strongest available test.

  • Research question: Does the prevailing theoretical explanation for protocol deviations hold in this specific, well-documented instance?
  • Case selection rationale: The site was chosen because it represents a critical test — if the theory doesn’t explain this well-documented case, that’s strong evidence against the theory; if it does, that’s meaningful support, precisely because the case was not cherry-picked to be favorable.
  • Data sources: The site’s monitoring reports, deviation logs, and interviews with site staff involved in the deviation.
  • What the design can and cannot conclude: A single, well-chosen critical case can confirm, disconfirm, or extend a theoretical proposition with more logical force than an arbitrarily chosen case would. It cannot establish how frequently the mechanism it identifies occurs across other sites.

A Real Published Case Study: Roseto, Pennsylvania

The scenarios above are composites, so it helps to see how a real case played out. One of the best-known is Roseto, Pennsylvania. In 1961 the town’s physician, Benjamin Falcone, told Stewart Wolf, then head of medicine at the University of Oklahoma, that the community had a low rate of heart attacks among its Italian American residents, which sparked a series of investigations. A 1964 paper in JAMA by Stout, Morrow, Brandt and Wolf reported an unusually low incidence of death from myocardial infarction in the town and attributed it to strong social bonds within a tight-knit community. A 1992 follow-up by Egolf, Lasker, Wolf and Potvin compared Roseto with the neighbouring town of Bangor over 50 years and found that Roseto’s heart disease rates rose to match its neighbours’ as residents adopted more individualistic lifestyles. Later studies questioned the social-bonds explanation, noting methodological flaws in the original and that Roseto’s heart-attack death rate was comparable to that of Framingham, Massachusetts, and a 2024 article attributed any difference to moderate wine consumption (see the overview of the Roseto effect).

Roseto works as an example because it shows the full arc of a case study: an unusual case that prompted a causal explanation, a longitudinal follow-up that tested it, and published critique of what the original design could not support.

How to Conduct a Case Study: Four Steps

Practical guides to case study research, such as Dovetail’s overview of case study design, break the process into four steps. Select the case by identifying the problem and choosing the case that best illuminates it. Build a theoretical framework that links the study to existing literature and theory. Collect data, typically qualitative sources such as interviews and documents and, where useful, quantitative data. Analyse the case and structure the write-up so that findings connect back to theory. For the design decisions behind each step, see CASRAI’s case study research method guide, and for the qualitative techniques used in the data collection step, see qualitative research methods.

Sample Case Study Structure and Research Questions

A sample case study write-up follows the same logic as the worked examples above: an introduction that states the research question and why the case matters, a methods section explaining case selection and data sources, findings organised around the question or theory, and a discussion that states the limits of generalization. Each worked example in this guide supplies the first three elements in compact form. For ready-made questions to adapt to your own case, see CASRAI’s research question examples, and for full paper layouts see the case study paper structure guide.

Single-Case vs. Multiple-Case Design and Replication Logic

The intrinsic example above is necessarily single-case: the whole point is that the case itself, not a comparison, is the object of study. The collective example is necessarily multiple-case, and it illustrates replication logic — the principle, developed by Robert K. Yin, that governs how many cases a multiple-case design needs. Replication logic treats each additional case the way an experimenter treats each additional experimental run: added not to build statistical power the way a larger survey sample would, but to test whether the same theoretical proposition holds (literal replication) or to deliberately vary conditions and see how the outcome changes (theoretical replication). A multiple-case study with three cases that each confirm the same mechanism is making a different, and in some ways stronger, argument than a single case making the same claim — but it is still not a sample, and reviewers who ask “why only four cases and not forty” are usually applying sampling logic to a design that doesn’t use it.

How Case Selection Is Justified

None of the scenarios above chose a case at random, and that’s deliberate — case study rigor depends heavily on being able to explain why this case, and not another, was chosen. The justifications that recur across the examples map onto a small set of standard rationales:

  • Typical case — chosen because it represents the ordinary, modal instance of the phenomenon (the instrumental nonprofit example above).
  • Deviant or extreme case — chosen because its outcome departs sharply from what similar cases produced, making the mechanism easier to see (the explanatory open-data example above).
  • Critical case — chosen because it provides the strongest possible test of an existing theoretical proposition (the critical instance example above).
  • Maximum-variation case selection — used in multiple-case designs, where cases are deliberately chosen to differ from each other on relevant dimensions (the collective technology-transfer example, which selected offices varying in size and prior licensing volume), so that any pattern holding across all of them is less likely to be an artifact of one shared characteristic.

Whichever rationale applies, it belongs explicitly in the methods section — not left for a reader to infer. See CASRAI’s Sampling Methods reference for how these case-selection logics relate to, and differ from, probability sampling used in quantitative designs.

Triangulation Across Data Sources

Every scenario above draws on more than one data source — interviews plus documents plus observation, at minimum — and that’s not incidental to the examples; it’s a defining feature of rigorous case study design. Triangulation is the practice of checking whether independent sources of evidence converge on the same conclusion, and it functions in case study research the way replication across labs functions in quantitative work: no single source is treated as sufficient on its own. In the descriptive authorship example above, policy documents alone might describe an intended process that interviews reveal isn’t actually followed in practice — triangulation is what surfaces that gap.

Yin, Stake, and Eisenhardt: Three Traditions, Compared

The examples above are written to be compatible with all three major methodological traditions, but the traditions themselves differ in emphasis, and it’s worth knowing which one a given piece of guidance (or a given reviewer) is drawing on:

  • Robert K. Yin treats case study as a formal method with an explicit research design, a case study protocol, and structured evidentiary standards (construct validity, internal validity, external validity, reliability) borrowed and adapted from the language of quantitative design. Yin’s approach is the most procedural of the three and underlies most of the type labels used above (exploratory, descriptive, explanatory; single- vs. multiple-case; replication logic).
  • Robert Stake treats case study less as a fixed method and more as a choice of what to study — a bounded system examined as an integrated whole — and emphasizes naturalistic, interpretivist inquiry over standardized procedure. Stake’s intrinsic/instrumental/collective typology, used throughout the examples above, comes from this tradition, as does its greater tolerance for the researcher’s own interpretive voice in reporting findings.
  • Kathleen Eisenhardt, writing primarily for organizational and management research, developed a structured approach to building theory from multiple cases: a roadmap running from case selection through within-case analysis, cross-case pattern search, and iterative comparison against emerging and existing theory, with theoretical saturation (rather than a fixed case count) determining when to stop adding cases. The collective/multiple-case example above, and its use of replication logic to build rather than just test theory, sits closest to this tradition.

None of the three is “correct” to the exclusion of the others — the choice depends on the field’s conventions and on how procedural versus interpretive the resulting paper needs to be. Stating explicitly which tradition a study follows, in the methods section, heads off reviewer confusion about why the paper doesn’t follow a different tradition’s conventions.

Validity and Generalizability: Writing About Limits Honestly

Every “what the design can and cannot conclude” line in the examples above is doing the same job: stating the limit of the claim before a reviewer has to point it out. The single most common reviewer objection to case study research is that findings from one or a few cases can’t be generalized — and the honest response, following Yin, is not to deny the limitation but to name the correct kind of generalization being claimed. Case studies generalize analytically, extending or refining a theoretical proposition, not statistically, in the sense of estimating a population parameter. See CASRAI’s Generalizability in Research guide for the full distinction and for how to phrase a generalizability statement in a methods or discussion section without overclaiming or underselling the work.

Several of the scenarios above involve a single, identifiable hospital, nonprofit, research center, or clinical trial site — and that identifiability is itself a design consideration, not just a write-up detail. Unlike a survey respondent who can be reported as one anonymized row among hundreds, a case study organization or individual is often recognizable from the description alone, even without being named. That raises two separate obligations: securing informed consent from participants within the case (interviewees, observed staff) on the same basis as any qualitative study — see CASRAI’s informed consent reference — and, separately, negotiating with the case organization itself about how much identifying detail will appear in the published account, since organizational consent and individual participant consent are not the same thing and an IRB or ethics board will typically expect both to be addressed. Where a case cannot be adequately anonymized without destroying its analytic value, that trade-off should be disclosed to participants and reviewers rather than solved silently by either overclaiming anonymity or omitting the disclosure altogether.

Four Published Case Studies and How They Were Designed

Because the scenarios above are composites, it helps to see how published studies describe their own designs. Crowe and colleagues (2011), writing in BMC Medical Research Methodology, illustrate the case study approach with four worked examples from health services research. The data table below this guide compares them by design label, number of cases and data sources; the notes here explain what each example shows.

1. A single intrinsic case: recruiting South Asians into UK asthma research

The first example is described as a single intrinsic case study. Its data were in-depth interviews with UK and US asthma researchers, supplemented by a questionnaire. It shows the intrinsic logic: the case is of interest in itself, and the aim is a detailed account of how the recruitment problem is understood and handled, not a statement about all research recruitment.

2. A multiple-case design: GPs with a special interest in respiratory services

The second example is a multiple-case design covering four primary care organisations (PCOs) in England and Wales. Screening telephone interviews with 30 PCOs came first; the four cases were then studied using interviews, e-mail discussions, local documents, patient diaries, news items and a national workshop. The pattern to copy is the funnel: a wider screen to choose the cases, then deep data collection in a small number of them.

3. A collective, mixed-methods case study: electronic health records

The third example follows electronic health record implementation across five NHS acute and mental health Trusts. Crowe and colleagues describe it as a mixed methods, longitudinal, multi-site, socio-technical collective case study. Data came from interviews, documents, field notes, observation and quantitative data, which is the same combination of qualitative and numeric evidence discussed in the FAQ below.

4. A multi-site collective case study: patient safety learning in education

The fourth example comprises eight case studies, two for each of four professional groups: medicine, nursing, physiotherapy and pharmacy. Evidence included documents, the views of course leads, tutors and students, and observations. Interviews or paired discussions were used where focus groups did not work, a practical reminder that data collection methods are adapted in the field.

Taken together, the four examples span one to eight cases, and every one uses more than one kind of data source. The categories are also not mutually exclusive: Crowe and colleagues note that a case studied for its own sake (intrinsic) can later be used to illuminate a wider issue (instrumental).

Stages in Published Case Study Guidance

Crowe and colleagues set out the approach as a sequence. First, define the case and its boundaries. Second, select the case or cases, which in a collective study means choosing several. Third, collect and analyse the data. Fourth, analyse each case before looking across cases. Fifth, report the findings. The order matters most for multiple-case work: within-case analysis comes before any cross-case comparison, which is also the logic behind the collective technology-transfer scenario above.

For analysis, the same source describes the Framework approach in five stages: familiarisation with the data, identifying a thematic framework, indexing, charting, and mapping and interpretation. It is one structured way to move from large volumes of interview and document data to a defensible account of each case.

Definitions Used by the Methodologists

  • Stake, as quoted by Crowe and colleagues (p. 237), describes a case study as both the process of learning about the case and the product of our learning.
  • Yin, as stated in the same source, defines a case study as an empirical inquiry into a contemporary phenomenon in its real-life context, where the boundaries between the phenomenon and its context are not clearly evident.

Yazan (2015) compares the approaches of Yin, Merriam and Stake in education research; it is a useful starting point for choosing between the traditions summarised earlier in this guide.

Strengths, Pitfalls and Safeguards

Crowe and colleagues list the practical difficulties of the approach: the large volume of data a case study generates, the dependence on gatekeepers for access, the difficulty of defining the boundary of the case, the risk that participants or organisations can be identified, and the temptation to force a theory onto the findings. Matching safeguards in the same guidance are theoretical sampling of cases, respondent validation, triangulation of data sources, and transparency about the researcher’s own role.

The same source notes that instrumental cases are often chosen because they are typical, but that a deviant or atypical case can be more informative. That is the reasoning behind the deviant and critical examples earlier in this guide. Read the validity and ethics sections above together with these safeguards when you write your methods section.

Frequently Asked Questions

What are the main types of case study, and is there one universal example?

No — that’s the point of this guide. The type that best fits a research question (intrinsic, instrumental, collective, explanatory, exploratory, descriptive, or critical instance) determines what kind of example is even relevant, because each type answers a structurally different question. See the worked examples above for each type.

How is a case study example different from a case report?

A case study, in the research-method sense used throughout this guide, is a bounded investigation using multiple triangulated data sources to answer a “how” or “why” question about a phenomenon in its real-world context. A case report, common in clinical fields, is typically a single-patient or single-event write-up documenting an unusual clinical presentation, without the same design apparatus (case-selection rationale, replication logic, triangulated data collection). See CASRAI’s case study research method guide for the fuller distinction.

Can a case study example use quantitative data?

Yes. Case study is a design, not a data type — it commonly combines qualitative sources (interviews, documents, observation) with quantitative ones (usage analytics, financial records, licensing volume, as in the collective example above). What makes it a case study is the bounded, in-depth, triangulated treatment of one or a few cases, not the exclusion of numbers.

How many cases does a multiple-case study need?

There is no fixed number. Crowe and colleagues describe a collective case study simply as one involving several cases, and the published examples above range from four to eight. Under replication logic, the right number is enough cases to show consistent replication or to vary conditions until the theoretical pattern is clear; the justification, not the count, is what a methods section needs to defend.

Do case study examples need IRB approval?

If the case involves identifiable people (interviews, observation of staff, patient-level data) or identifiable organizations whose participation carries risk, yes — the same human-subjects and consent obligations that apply to any qualitative study apply here, plus the added consideration of organizational-level consent discussed above.

Which published examples of case studies exist in health research?

Crowe and colleagues (2011) present four: recruitment of South Asians into UK asthma research (single intrinsic case), GPs with a special interest in respiratory services (four PCOs, multiple-case), electronic health record implementation (five NHS Trusts, mixed methods collective) and patient safety learning in pre-registration education (eight cases). See the section above and the data table below.

What is the difference between an intrinsic and an instrumental case study example?

In an intrinsic case study the case itself is the point of interest, as in the asthma recruitment example. In an instrumental case study the case is used to understand something wider. See the intrinsic and instrumental scenarios earlier in this guide.

Can a case study be both intrinsic and instrumental?

Yes. The categories are not mutually exclusive; Crowe and colleagues note that a case first examined for its own sake can later serve an instrumental purpose.

What are the limitations of case study research?

Large data volumes, reliance on gatekeepers for access, difficulty bounding the case, identifiability of participants and the temptation to force a theory onto the data, as listed in the strengths and pitfalls section. For limits on generalisation see the validity section above.

Sources

Sources.

Data: Design features of four published case studies in health services research

Compiled from the four worked example tables in Crowe et al. (2011), recoding each study's number of cases, design label and data-collection types into comparable columns. Counts are as stated in the source; blank counts are omitted rather than estimated.

Period covered: 2011.

Design features of four published case studies in health services research
Published exampleDesign label as publishedCases or sites (cases)Data sources used
Recruiting South Asians into UK asthma researchSingle intrinsic case study1In-depth interviews; supplementary questionnaire
GPs with a special interest in respiratory services, four PCOsMultiple-case design4Interviews, e-mail discussions, local documents, patient diaries, news items, national workshop
Electronic health record implementation, NHS TrustsMixed methods longitudinal multi-site collective case study5Interviews, documents, field notes, observations, quantitative data
Patient safety learning in pre-registration educationMulti-site mixed method collective case study8Documents, course lead/tutor/student views, observations, policy and management views

Sources

Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · free to try

Ask about Case Study Examples: Worked Examples of Every Case Study Type

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →