Case Study Research Understanding the Particular in Depth
An empirical inquiry that investigates a contemporary phenomenon in depth within its real-life context
Case study research is among the most widely practised and most widely misunderstood research designs in the social, educational, health, and management sciences. Researchers invoke the term "case study" to describe everything from single-participant narrative accounts to comparative cross-national policy analyses — yet these uses bear little resemblance to what the methodology actually requires. Doing case study research correctly demands a precise understanding of what constitutes a case, what questions the design can and cannot answer, and what rigorous procedures look like from problem formulation to the final report.
Yin 2018 (6th ed.)
Stake 1995 / 2005
Merriam & Tisdell 2016
Miles, Huberman & Saldaña 2020
Tight 2022
Creswell & Guetterman 2024
7Sections
3–4 hrsEst. time
3Major traditions
4Core design types
Progress0 of 7 sections
0%
Section 01
Definition & Scope of Case Study Research
Core TheoryReading · 20 min
Case study research is an empirical inquiry that investigates a contemporary phenomenon — the "case" — in depth and within its real-life context, particularly when the boundaries between that phenomenon and its context are not clearly evident. It is not a data collection technique, a sampling strategy, or a synonym for qualitative research. It is a complete research design with its own logic of inquiry, its own evidentiary standards, and its own analytical procedures.
Formal Definition
A case study is an empirical inquiry that investigates a contemporary phenomenon (the "case") in depth and within its real-life context, especially when the boundaries between phenomenon and context may not be clearly evident. Case study inquiry copes with the technically distinctive situation in which there will be many more variables of interest than data points, relies on multiple sources of evidence with data needing to converge through triangulation, and benefits from prior development of theoretical propositions to guide data collection and analysis (Yin, 2018, pp. 15–16).
What immediately separates this definition from popular usage is its insistence on context. A survey removes respondents from context to measure variables; an experiment controls context to isolate causal mechanisms. Case study research does neither. It deliberately keeps the phenomenon embedded in its context precisely because the context is analytically important — you cannot understand what is happening without understanding where and under what conditions it is happening.
What Exactly Is a "Case"?
Before designing a case study, the researcher must answer one foundational question: What is the case? This is not as straightforward as it appears. Robert Stake (1995, p. 2) described a case as "a specific, complex, functioning thing" — a bounded system. Sharan Merriam and Elizabeth Tisdell (2016, p. 37) define it as "a thing, a single entity, a unit around which there are boundaries." The boundaries are what transform an interesting topic into a researchable case.
A case can be virtually any bounded unit of analysis: a single person, a family, an organisation, a community, a programme, a policy, an event, a decision, a process, or a time period. What matters is not what kind of entity it is but whether it has identifiable, defensible boundaries that the researcher can articulate and justify.
Running Practical Example — Used Throughout This Lesson
To make every concept in this lesson concrete and immediately applicable, all examples will draw from a single illustrative study: How a regional government hospital in the Philippines implemented an electronic health records (EHR) system during the COVID-19 pandemic — and the organisational change processes it navigated. This is a single-case embedded design. The bounded case is the hospital. The embedded units of analysis are three departments: the Emergency Room, the Out-Patient Department, and the Administrative Services Division. Multiple evidence sources include interviews with clinical and administrative staff, implementation documents, system audit logs, and direct observation of ward workflows. This example is used instructionally; all concepts are illustrated against this case throughout Sections 1–7.
The Central Question Case Study Answers
Case study research is appropriate when the researcher is asking "how" or "why" questions about a contemporary phenomenon over which the researcher has no control (Yin, 2018, p. 11). These are explanatory and descriptive questions, not questions that can be answered by counting, measuring, or randomly sampling. The following question types are native to case study design:
How did this organisation navigate this crisis?
Why did this policy succeed in one context but fail in another?
How did these teachers adapt their practice when the curriculum changed?
Why did this community health intervention produce outcomes different from those in the published literature?
Notice what these questions share: they ask about process, not about frequency or distribution. They assume that context matters. And they cannot be answered without sustained, in-depth engagement with the specific case in its specific setting.
Case Study Distinguished from Other Qualitative Designs
Design
Central Aim
Core Question
Primary Unit
Distinguishing Feature
Case Study
Understand a bounded phenomenon in context
How / why did this happen?
The case (bounded system)
Context is integral, not controlled
Grounded Theory
Generate theory of a social process
What theory explains this pattern?
Social process / interaction
Concurrent data collection and analysis
Phenomenology
Describe the essence of lived experience
What is the essence of this experience?
Lived experience of participants
Bracketing of researcher preconceptions
Ethnography
Describe a culture-sharing group
What does this culture look like?
Cultural group / system
Prolonged field engagement (months/years)
Narrative Inquiry
Understand an individual's storied experience
How does this person story their life?
Individual narrative
Restorying of participant accounts
The Most Common Misuse of "Case Study" in Published Research
A survey of 200 published papers claiming case study methodology found that a substantial proportion were actually purposive qualitative studies, interview-based descriptive studies, or single-participant narrative accounts with no attention to bounding the case, developing a protocol, triangulating evidence, or conducting systematic analysis. Calling a study a "case study" without meeting its methodological requirements does not make it one — and misrepresenting the design exposes the research to legitimate methodological critique from peer reviewers and thesis examiners (Tight, 2022; Yin, 2018, pp. 16–17; Merriam & Tisdell, 2016, pp. 36–38).
📚
Section 1: Definition & ScopeMark complete when you can define a case, identify the core question type, and distinguish case study from other designs
Section 02
Historical Development & Intellectual Traditions
HistoricalReading · 25 min
Case study research does not originate from any single discipline or founding text. Its development is better understood as the gradual convergence of methodological practices from sociology, political science, clinical psychology, business management, and education — each field contributing distinct procedural emphases that eventually crystallised into recognisable traditions associated with three principal methodologists.
1941–2021
Robert K. Yin
COSMOS Corporation · Post-positivist Tradition
A social scientist trained in the natural sciences at MIT, Yin brought systematic, quasi-experimental logic to qualitative inquiry. He conceived case study as a rigorous research design with explicit procedures for ensuring validity and reliability — treating it as comparable in scientific standing to experiments and surveys. His insistence on protocol development, chain of evidence, construct validity, and analytic generalization made his framework the dominant reference in management, public administration, health, and educational research worldwide. The sixth edition of his foundational text (2018) remains the most cited methodological work in case study research globally.
Key texts
Case Study Research and Applications(6th ed., 2018)
Qualitative Research from Start to Finish(2nd ed., 2016)
Applications of Case Study Research(3rd ed., 2012)
1927–2011
Robert E. Stake
University of Illinois · Constructivist Tradition
An educational evaluator and constructivist, Stake positioned case study as inherently interpretive — a naturalistic, particularistic inquiry into the subjective meanings that make a case what it is. Where Yin sought generalisable theoretical propositions, Stake sought particularistic understanding: the rich, contextual, experiential understanding that readers could apply to their own situations through what he called naturalistic generalisation. His typology — intrinsic, instrumental, and collective case studies — remains the standard classification for purpose-driven case selection. Stake's influence is strongest in education, programme evaluation, and social work.
Key texts
The Art of Case Study Research(1995)
Multiple Case Study Analysis(2005)
Qualitative Research: Studying How Things Work(2010)
1940–
Sharan B. Merriam
University of Georgia · Qualitative Tradition
An adult education scholar, Merriam synthesised case study within the broader framework of qualitative research, positioning it as particularistic (focused on a particular phenomenon), descriptive (rich, thick account), and heuristic (illuminates understanding). Her most important contribution is the clarification that the bounded system is the defining characteristic of case study — not the qualitative nature of the data, not the interpretive orientation of the researcher, but the presence of identifiable and defensible case boundaries. Merriam's framework is widely adopted in education, nursing, social work, and applied professional research.
Key texts
Qualitative Research: A Guide to Design and Implementation(4th ed., with Tisdell, 2016)
Case Study Research in Education(1988)
Learning in Adulthood(with Bierema, 2014)
Roots in the Chicago School of Sociology
Before Yin, Stake, or Merriam formalised the methodology, case study practice was already established in early twentieth-century American sociology. The Chicago School — associated with Robert Park, Ernest Burgess, and their colleagues at the University of Chicago — produced landmark case studies of urban life, immigration, and social deviance between 1920 and 1950. Works such as The Polish Peasant in Europe and America (Thomas & Znaniecki, 1918–1920) and The Taxi-Dance Hall (Cressey, 1932) demonstrated that in-depth engagement with a specific social situation, using life histories, documents, and observation, could generate insight that survey methods could not. These works established the intellectual legitimacy of deep contextual inquiry long before it had a formal methodological name.
The Crisis of Legitimacy and Yin's Response
By the 1960s and 1970s, case study research faced a crisis of methodological legitimacy. The dominance of positivist social science — with its emphasis on hypothesis testing, random sampling, and statistical generalisation — positioned case study as anecdotal, ungeneralisable, and pre-scientific. Yin's 1984 publication of Case Study Research: Design and Methods was a direct methodological response to this critique. He argued that case study research could be conducted with the same rigour as experimental research by adopting analogous procedures: systematic protocol development, construct and internal validity checks, and the logic of replication rather than statistical sampling. His framework rehabilitated case study as a legitimate scientific design — not despite its qualitative data but because of its systematic logic.
The Positivist–Interpretivist Tension
The most substantive methodological debate in contemporary case study research is between Yin's post-positivist tradition — which treats case study as a design capable of producing theoretical knowledge subject to validity and reliability assessment — and Stake's constructivist tradition, which treats the case as an object of understanding in itself, resisting the reduction of context to generalisable propositions. Merriam occupies a middle position, accepting qualitative case study as rigorous research while rejecting Yin's quasi-experimental language. Researchers selecting case study methodology must position themselves in relation to this debate, as it determines their quality criteria, their claims about generalisability, and their analytical procedures (Tight, 2022; Ridder, 2021; Creswell & Guetterman, 2024).
"The case study is not a methodological choice but a choice of what is to be studied. By whatever methods, we choose to study the case. We can study it analytically or holistically, entirely by repeated measures or hermeneutically, organically or culturally, and by mixed methods — but we concentrate, at least for the time being, on the case."
Stake, 1995, p. xi
📚
Section 2: Historical DevelopmentMark complete when you can place Yin, Stake, and Merriam within distinct intellectual traditions and explain the positivist–interpretivist tension
Section 03
Case Study Designs & Typologies
Core TheoryReading · 35 min
Understanding the typology of case study designs is not an exercise in classification for its own sake. The design you select determines what claims you can make, what data you need to collect, what analytical procedures you must apply, and how your research contributes to knowledge. Three classification systems are in active use — Yin's structural typology, Stake's purpose-based typology, and Merriam's qualitative framework — and they are not mutually exclusive. A single study can be described simultaneously as a single-case embedded design (Yin), an instrumental case study (Stake), and a qualitative interpretive case study (Merriam).
Yin's 2×2 Structural Typology
Yin (2018, pp. 45–69) organises case study designs along two dimensions: the number of cases (single or multiple) and the number of units of analysis (holistic or embedded). Their combination produces four design types, each with distinct implications for data collection, analysis, and scope of findings.
Yin's 2×2 Case Study Design Matrix
Single Case
Multiple Cases
Holistic (1 unit)
Type 1: Single-Case Holistic
One case; examined as a whole, undivided unit. No sub-units. Used when the case is unique, critical, revelatory, longitudinal, or extreme. Example: a sole hospital implementing a national pilot EHR programme — the entire hospital is the unit.
Type 3: Multiple-Case Holistic
Several cases; each examined as a whole. Follows replication logic, not sampling logic. Example: three different hospitals in three provinces, each treated as one undivided unit, compared for how they implemented EHR.
Embedded (2+ units)
Type 2: Single-Case Embedded
One case; examined through multiple sub-units within it. Example: one hospital (the case) examined through three departments — ER, OPD, and Admin — as distinct embedded units. This is the running example used throughout this lesson.
Type 4: Multiple-Case Embedded
Several cases; each examined through multiple sub-units. The most complex and resource-intensive design. Example: three hospitals, each examined through three departments — nine embedded units across three cases.
Three Traditions of Case Study — Select a tradition
Yin (Post-positivist)
Stake (Constructivist)
Merriam (Qualitative)
Selecting a Design
Ontology: Post-positivist. A real social world exists independently of the observer, though access to it is always partial. Patterns and causal mechanisms are discoverable through rigorous inquiry, even in single cases.
Purpose classification: Yin (2018, pp. 6–9) classifies case studies by their research purpose: (1) Exploratory — used when questions and propositions are not well-defined; generates hypotheses for later testing; (2) Descriptive — provides a comprehensive, contextually rich description of the case; (3) Explanatory — seeks causal explanations; tests whether a theoretical proposition can account for observed outcomes.
Single-case rationale: Yin identifies five rationales for single-case design: the case is (1) critical — it tests a clearly stated theory; (2) unusual or extreme — its rarity makes it worth documenting; (3) common — it captures circumstances of everyday life; (4) revelatory — it provides access to a previously inaccessible phenomenon; (5) longitudinal — it studies the same case at two or more points in time (Yin, 2018, pp. 47–52).
Multiple-case rationale: Multiple cases follow replication logic, not sampling logic. Each case is deliberately selected either because it is predicted to produce the same results as the previous case (literal replication) or because it is predicted to produce contrasting results for predictable theoretical reasons (theoretical replication). This is a critical distinction: adding more cases in case study research is not equivalent to increasing a sample size for statistical power.
Quality criteria: Construct validity, internal validity (for explanatory studies), external validity, and reliability — each with specific procedural tactics for demonstration (see Section 7).
Ontology: Constructivist/relativist. Reality is constructed through social interaction and is always multiple, local, and specific. Case study does not discover pre-existing patterns; it renders interpretive accounts of how participants make sense of their bounded situation.
Purpose classification — three types:
Intrinsic case study: The case itself is of primary interest. The researcher does not select the case to illustrate a broader issue or to represent other cases — the case is studied for what is intrinsically interesting about it. Its particularity and complexity are the point. Example: A researcher fascinated by how one specific hospital's nursing team maintained patient dignity during the chaos of the pandemic — with no intention of generalising to other hospitals.
Instrumental case study: The case is studied primarily to understand something beyond itself. The case serves as the vehicle for understanding a broader phenomenon, issue, or theoretical question. The case is of secondary interest; it facilitates understanding of something else. Example: Studying one hospital's EHR implementation to understand the broader process of organisational resistance to digital transformation in healthcare.
Collective (multiple) case study: Several cases are studied jointly to inquire into a phenomenon, population, or general condition. Each individual case is studied instrumentally, and the collective produces a more robust understanding. Example: Studying EHR implementation across six hospitals in different regions to understand how institutional context shapes adoption.
Naturalistic generalisation: Stake explicitly rejects statistical and even analytic generalisation as the goal of case study. Instead, readers draw naturalistic generalisations — recognising patterns from the case that apply to their own experience and situations. The researcher's task is to provide sufficiently rich description to enable this reader-side generalisation (Stake, 1995, pp. 85–86).
Ontology: Interpretivist. Knowledge is socially constructed; meaning is context-dependent. The researcher is not a neutral instrument but an active interpreter whose perspective shapes the inquiry.
Defining characteristic: Merriam and Tisdell (2016, p. 37) insist that the single most defining characteristic of case study research is the bounded system. If you cannot draw a boundary around your unit of analysis and justify why the boundary sits where it does, you do not have a case study. You may have qualitative research — but not case study research.
Three key characteristics of case study (Merriam):
Particularistic: The case study focuses on a particular situation, event, programme, or phenomenon. The focus is on the specific rather than the general.
Descriptive: The end product is a rich, thick description of the phenomenon under study. It includes as many variables as possible and portrays their interaction over time.
Heuristic: The case study illuminates the reader's understanding of the phenomenon. It can bring about the discovery of new meaning, extend the reader's experience, or confirm what is known.
Data collection: Merriam advocates using whatever data sources — interviews, observation, documents, artefacts — provide the best evidence for understanding the bounded system. The choice of methods is pragmatic, not prescribed.
Generalisability: Merriam endorses transferability over generalisability — the degree to which findings from one case can be transferred to another situation with similar characteristics. Transferability is a reader judgment, not a statistical property of the study (Merriam & Tisdell, 2016, pp. 254–258).
The primary selection criterion is philosophical alignment: Researchers who believe that rigorous, valid, and reliable knowledge can be produced through systematic case study procedures and applied to inform theory should adopt Yin's framework. Researchers who believe that understanding is inherently local, contextual, and constructed should adopt Stake's or Merriam's framework.
Secondary criteria — by research purpose:
Choose Yin's framework when: you want to test or develop theoretical propositions; your study is in management, health, public policy, or applied social science; your institution values explicit validity and reliability criteria; you are conducting explanatory or comparative case research.
Choose Stake's framework when: you are drawn to the intrinsic value of the particular case; your study is in education, programme evaluation, or social work; you prioritise thick, naturalistic description over causal explanation; your primary audience will apply findings through their own judgment rather than through replication.
Choose Merriam's framework when: you are working within qualitative research broadly and want case study to fit within that tradition; your study is in adult education, professional practice research, or applied health sciences; you want flexibility in data collection methods within a rigorous bounded-system framework.
Mixing frameworks without justification produces methodological incoherence. A common error in doctoral theses is adopting Yin's 2×2 design typology while using Stake's naturalistic generalisation as the quality criterion — or claiming Merriam's particularistic focus while asserting Yin's analytic generalisation as the study's contribution. These combinations are not automatically wrong, but they require explicit philosophical justification (Ridder, 2021; Tight, 2022, p. 2053).
Feature
Yin (Post-positivist)
Stake (Constructivist)
Merriam (Qualitative)
Ontology
Post-positivist realism
Constructivist / relativist
Interpretivist
Purpose typology
Exploratory / Descriptive / Explanatory
Intrinsic / Instrumental / Collective
Particularistic / Descriptive / Heuristic
Defining concept
Research design with explicit validity criteria
Bounded phenomenon of intrinsic interest
Bounded system with thick description
Generalisation type
Analytic (theoretical) generalisation
Naturalistic generalisation
Transferability
Role of theory
Guides design and analysis; propositions tested
Emerges from or informs interpretation
Sensitising framework, not a test
Quality standard
Validity and reliability (4 criteria)
Trustworthiness; resonance; reader utility
Credibility, transferability, dependability
Dominant disciplines
Management, health, policy, public admin
Education, evaluation, social work
Adult education, professional practice
📚
Section 3: Designs & TypologiesMark complete when you can describe Yin's 2×2 matrix, Stake's three types, and the selection criteria for each tradition
Section 04
Core Concepts in Case Study Research
Core TheoryReading · 30 min
Case study research employs a set of technical concepts with precise meanings that differ substantially from their everyday usage. Misunderstanding these concepts is the most direct route to design error — and design errors in case study research are typically irreversible once data collection begins. Every concept below carries a methodological consequence: getting it wrong changes what you study, what you can claim, and how your work will be evaluated.
Bounded System
Stake, 1995; Merriam & Tisdell, 2016, p. 37
The defining characteristic of a case study. A bounded system is a unit of analysis with identifiable, defensible limits — in time, place, activity, or membership. The boundary determines what is inside the case and what is context surrounding it. Without explicit bounding, a study lacks a case and becomes general qualitative research. In the running example: the boundary is one government hospital during the period of EHR implementation (January 2021–December 2022). Staff who left before implementation began are context, not data. The EHR vendor's organisation is context, not a second case.
The specific entity that the research question is about — the "case" itself. Defining the unit of analysis is the single most important decision in case study design. A common and serious error is to collect data from individuals (e.g., interviews with 20 staff) but then claim the unit of analysis is the organisation. If the unit of analysis is the hospital, then the hospital — not individual staff members — must be the subject of analytical conclusions. Individual interviews are data sources, not units of analysis. In the running example: the primary unit is the hospital as an organisation; the embedded units are three departments.
Yin, 2018, pp. 30–33; Ridder, 2021, pp. 45–52
Analytic Generalisation
Yin, 2018, pp. 21–22
The form of generalisation appropriate to case study research. Case study does not generalise from a sample to a population (statistical generalisation). It generalises from empirical findings to theoretical propositions — testing whether the theory holds in this case, and thereby extending or refining the theory's scope. The logic parallels that of an experiment: one experiment does not represent a population, but it can test a proposition. Multiple cases are like multiple experiments, each contributing to the theory's robustness. This distinction is frequently misunderstood and must be stated explicitly in the research design.
Yin, 2018, pp. 21–22; Tight, 2022; Creswell & Guetterman, 2024
Replication Logic
Yin, 2018, pp. 54–60
The rationale for selecting multiple cases. Cases are not sampled the way survey respondents are sampled. Each case is selected because it is predicted to produce either the same results as prior cases (literal replication) or contrasting results for theoretically explicable reasons (theoretical replication). A multiple-case study with three hospitals is not stronger because it has more participants — it is stronger because each hospital provides independent confirmation or contrast of the theoretical proposition being tested. If a researcher cannot explain why each case was selected in replication terms, the multiple-case design is unjustified.
Yin, 2018, pp. 54–60; Ridder, 2021
Triangulation
Yin, 2018, pp. 118–122; Merriam & Tisdell, 2016
The use of multiple sources of evidence to converge on the same finding, thereby strengthening construct validity. Case study triangulation is primarily across sources (interviews, documents, observation), not across methods. When three independent sources — an interview with a nurse manager, an implementation committee minute, and a direct observation of a ward briefing — all point to the same conclusion (e.g., that staff were given inadequate training time), the conclusion is far more defensible than if it rested on interviews alone. Triangulation does not mean conducting three separate studies; it means using multiple lenses on the same phenomenon within the same case.
Yin, 2018, pp. 118–122; Miles, Huberman & Saldaña, 2020, pp. 299–301
Chain of Evidence
Yin, 2018, pp. 127–128
The principle that a reader of the case study should be able to trace the derivation of any conclusion back through the analysis, to the data, to the protocol, to the original research questions — in either direction — without ambiguity at any step. A broken chain of evidence means a reader cannot verify how a conclusion was reached: they must accept it on the researcher's authority rather than on evidential grounds. Maintaining the chain of evidence requires rigorous documentation of every data collection and analytical decision — a case study database, a protocol, and transparent write-up practices are the primary mechanisms.
Yin, 2018, pp. 127–128
Pattern Matching
Yin, 2018, pp. 166–172
The most analytically powerful strategy in explanatory case study research. The researcher develops predictions about what patterns should be observed if a theoretical proposition is true, then examines the empirical data to determine whether the observed pattern matches the predicted one. If the patterns match, internal validity is strengthened — the proposition can account for what happened. If they do not match, the proposition must be revised. Example: Theory predicts that hospitals with stronger pre-existing IT infrastructure will implement EHR faster. The researcher observes actual implementation timelines across departments and checks whether the pattern matches the prediction.
Yin, 2018, pp. 166–172; Miles, Huberman & Saldaña, 2020
Rival Explanations
Yin, 2018, pp. 172–177
An essential validity strategy in explanatory case studies. The researcher explicitly identifies alternative explanations for the observed patterns — explanations that do not rely on the theoretical proposition being tested — and systematically examines whether the evidence can rule them out. Failure to consider rival explanations is the most common source of internal validity threats in case study research. Example: If EHR implementation was faster in Department A, was it because of IT infrastructure (the proposed explanation) or because Department A happened to have a particularly motivated department head (a rival explanation)? Both must be examined with evidence.
Yin, 2018, pp. 172–177
The Unit of Analysis Error: The Most Consequential Design Mistake in Case Study Research
In doctoral case study research, the most frequently observed and most damaging design error is the conflation of data sources with units of analysis. A researcher conducts 30 interviews with staff members at one hospital and then concludes: "This case study of 30 participants found that..." — but 30 participants is not the unit of analysis. The hospital is. The participants are sources of evidence about the hospital-level unit. The analytical conclusions must be about the hospital (the case), supported by evidence from the 30 interviews, not about the 30 individuals as if they were 30 separate cases. This error undermines the entire logic of the case study design and is rarely caught before thesis submission (Yin, 2018, pp. 30–33; Ridder, 2021, pp. 45–52; Tight, 2022).
📚
Section 4: Core ConceptsMark complete when you can define all eight concepts and explain the difference between a unit of analysis and a data source
Section 05
Data Collection & Sources of Evidence
Applied MethodReading · 40 min
The distinguishing feature of data collection in case study research is not any single method but the deliberate use of multiple sources of evidence whose independent contributions converge on the same analytical conclusions. This convergence — triangulation — is what gives case study findings their credibility. A case study that relies on interviews alone is not following the methodology's core evidentiary principle; it is qualitative research with a case study label.
Yin's Six Sources of Evidence
Yin (2018, pp. 110–132) identifies six major sources of evidence in case study research, each with distinct strengths and limitations. A well-designed case study does not use all six; it uses whichever combination provides the most convergent evidence for the specific research questions. The researcher must justify the selection.
Six Sources of Evidence — Strengths, Limitations & Case Study Application
Observation of ward morning briefings where EHR issues were raised; observation of staff navigating the system during patient admissions
Reveals actual (not reported) behaviours around the new system
The Case Study Database
Yin (2018, pp. 123–126) requires that case study researchers maintain a case study database — a formal, organised collection of all evidence gathered during the study, kept separately from the final report. The database contains raw field notes, interview transcripts, documents, photographs, quantitative data, and any other evidence collected. Its purpose is to enable other researchers to inspect the evidence independently and to maintain the chain of evidence between raw data and conclusions.
The database is not the same as the case study report. The report is a selective, analytical synthesis; the database is the raw evidential foundation. A case study without a database cannot be reliably audited — a significant problem in assessments of methodological rigour. At doctoral level, the database (or a representative portion of it) is typically submitted as an appendix or made available for examiner inspection.
The Case Study Protocol
The case study protocol is a document prepared before data collection begins. It is more than an interview guide — it contains the entire logic and procedure for the data collection phase. Yin (2018, pp. 84–95) specifies four components that every protocol must contain:
Overview of the project: Purpose, research questions, theoretical propositions, relevant readings
Field procedures: Credentials and access to sites, sources of information, procedural reminders
Case study questions: The specific questions the investigator must keep in mind while collecting data — these are the investigator's questions, not the interview questions given to participants. They reflect the analytical logic of the study.
Guide for the case study report: Outline and format for the final report; suggested evidence sources for each section
Illustrative Example: Triangulation Across Evidence Sources (Running Example — Hospital EHR Implementation)
Evidence Source & Content
Analytical Contribution to Finding
Interview — Nurse Manager, ER (Feb 2022): "We had two days of training. Two days. For a system we'd be using every single shift. Nobody was ready."
Perception of inadequate training time; possible source of adoption resistance in ER
Document — Implementation Committee Minutes (Nov 2021): "Training schedule revised from 5 days to 2 days due to hospital operations constraints. All departments notified."
Corroborates training reduction; places responsibility at organisational-decision level, not IT team level
Archival Record — IT Help Desk Logs (Jan–Mar 2022): ER submitted 47 tickets in January vs. 11 in OPD and 8 in Admin. ER tickets declined to 14 by March.
Behavioural indicator: ER had highest initial system difficulties; all departments converged over time, suggesting learning curve not structural failure
Direct Observation — ER morning briefing (Mar 2022): Ward clerk demonstrated workaround: printing paper backup forms and entering data retrospectively after shift.
Reveals informal adaptive behaviour not reported in any interview — evidence of persistent adoption gap despite declining help-desk tickets
This example illustrates triangulation in practice. Each source contributes something the others cannot: the interview captures subjective experience; the document places the decision in organisational context; the archival data provides objective behavioural metrics; the observation reveals informal practices invisible to both interviewees and documents. No single source tells the full story. Their convergence — and the tensions between them (the IT logs suggest improvement while observation reveals continuing workarounds) — produces the analytical richness that case study methodology promises.
How Many Interviews Is Enough in a Case Study?Applied guidance▾
This question reflects a fundamental misunderstanding of case study logic. There is no prescribed number of interviews in case study research because interviews are not the unit of analysis — the case is. The number of interviews is determined by (1) the informational needs of the research questions, (2) the need to hear from all relevant stakeholder perspectives within the bounded case, and (3) the principle of information sufficiency — collecting evidence until no new analytical insights are emerging from additional interviews.
In practice, doctoral case studies typically involve 15–30 interviews per case for single-case designs, depending on the complexity of the organisation or phenomenon. Multiple-case studies may involve 8–15 interviews per case. More important than the number is the purposive selection logic: each interviewee should be selected because they have specific knowledge relevant to the research questions — not because they volunteered or were conveniently available (Yin, 2018, pp. 110–118; Merriam & Tisdell, 2016, pp. 96–102; Miles, Huberman & Saldaña, 2020, pp. 57–60).
Can Case Study Research Include Quantitative Data?Methodological clarification▾
Yes — and Yin has insisted on this since the first edition of his text. Case study research is a research design, not a data type. The design can accommodate qualitative data (interview transcripts, field notes, documents), quantitative data (operational records, survey sub-studies, administrative statistics), or both. In the running example, the EHR system log-in data and help-desk ticket counts are quantitative; the interview transcripts and observation notes are qualitative. Both are legitimate case study evidence. The integration of quantitative evidence into case study analysis is not "mixing methods" in the sense of conducting a separate quantitative study — it is using quantitative data as one additional source of evidence within the case study's triangulation framework (Yin, 2018, pp. 9–10; Creswell & Guetterman, 2024, pp. 418–421).
Thick Description: What It Is and What It Is NotCore concept▾
The term "thick description," borrowed from the anthropologist Clifford Geertz (1973), is widely used in case study methodology but frequently misunderstood. Thick description is not merely a detailed description. It is a description that includes enough contextual information — cultural meaning, social relationships, situational dynamics, historical background — for a reader to judge the applicability of the findings to their own situation. It is the vehicle for transferability in Merriam's framework and for naturalistic generalisation in Stake's. A description of what happened in the hospital's ER during EHR implementation is thin if it lists events. It becomes thick when it explains the staffing ratios, the pre-existing IT literacy levels, the management culture, the COVID-19 pressure, and how all of these contextual factors shaped what happened and why. Thick description is what makes the case legible to someone who was not there (Merriam & Tisdell, 2016, pp. 254–255; Tight, 2022).
📚
Section 5: Data Collection & EvidenceMark complete when you can describe the six evidence sources, explain triangulation in practice, and articulate what a case study protocol must contain
Section 06
The Case Study Research Process
Applied MethodReading · 35 min
Case study research proceeds through a recognisable sequence of design decisions and procedural stages, though not as rigidly as experimental research. The sequence below synthesises the procedural guidance of Yin (2018), Merriam and Tisdell (2016), and Miles, Huberman and Saldaña (2020) into a practical framework for doctoral researchers. Each stage has specific deliverables; completing each stage without the required deliverables means entering the next stage with compromised analytical foundations.
01
Define the Research Problem, Research Questions, and Theoretical Propositions
The research problem must identify a gap in understanding — something about a real-world phenomenon that existing literature has not adequately explained, particularly in a specific context. Case study research questions are "how" or "why" questions about contemporary phenomena. Alongside the research questions, the researcher develops theoretical propositions — prior theoretical statements about what the answer might be, derived from the existing literature. These propositions are not hypotheses to be statistically tested; they are guides for data collection that tell the researcher what evidence to look for and what patterns to expect. Without propositions, a case study has no analytical direction and becomes an open-ended description that cannot contribute to knowledge. In the running example: Proposition 1 — Hospitals with stronger existing IT governance will demonstrate faster EHR adoption rates. Proposition 2 — Insufficient training time is the primary source of initial adoption resistance.
Yin, 2018, pp. 33–42; Ridder, 2021, pp. 71–82
02
Select the Case(s) Purposively and Justify the Selection
Cases are never randomly selected. They are purposively selected because they possess the specific characteristics necessary to address the research questions and test the theoretical propositions. For single cases, the researcher justifies selection using one of Yin's five rationales (critical, unusual, common, revelatory, or longitudinal). For multiple cases, selection follows replication logic: each additional case is selected because it will either confirm (literal replication) or usefully contrast (theoretical replication) the propositions. The selection justification must be written explicitly in the methodology chapter — "convenience" or "access" are not theoretically defensible selection rationales in doctoral research, though they may be acknowledged as practical constraints alongside the theoretical justification. In the running example: The hospital was selected as a revelatory case — EHR implementation during a pandemic provided access to an implementation process under conditions of maximum organisational stress, which existing literature had not documented in a Philippine healthcare context.
Yin, 2018, pp. 47–60; Merriam & Tisdell, 2016, pp. 96–102
03
Develop the Case Study Protocol
Before entering the field, the researcher prepares the case study protocol. This is a critical validity mechanism: a documented, pre-specified procedure for data collection that can be inspected by examiners to confirm that data collection was systematic, consistent, and guided by the research questions. The protocol contains: (1) an overview of the project background, purpose, and propositions; (2) field procedures including site access, contact information, and ethical approvals; (3) the investigator's case study questions — the analytical questions the researcher is mentally working through during every data collection activity; and (4) the report guide, specifying how findings from each evidence source will be organised. For multiple-case studies, the protocol must be followed consistently across all cases to enable legitimate cross-case comparison. Deviations from protocol during fieldwork must be documented and justified.
Yin, 2018, pp. 84–95
04
Collect Evidence from Multiple Sources Simultaneously
Data collection in case study research is multi-source from the outset. The researcher does not complete all interviews, then collect all documents, then conduct all observations — evidence sources are activated in response to the emerging analytical picture, guided by the protocol's investigator questions. As evidence accumulates, the researcher begins conducting within-case analysis: identifying emerging patterns, noting discrepancies between sources, and identifying gaps that require additional evidence. Fieldwork notes must document not only what was observed or said but also the researcher's analytical reflections — these are the case study equivalent of memo-writing in grounded theory. The case study database is maintained throughout this stage: every piece of evidence — a scanned document, a transcribed interview, a field observation note, a downloaded archival record — is catalogued with date, source, and retrieval information.
Yin, 2018, pp. 100–132; Miles, Huberman & Saldaña, 2020, pp. 57–88
05
Conduct Within-Case Analysis
Analysis begins with a thorough examination of each case individually. The researcher compiles all evidence about the case, organises it thematically or chronologically, identifies patterns within the evidence, and begins matching observed patterns against the theoretical propositions. The primary analytical strategies — pattern matching, explanation building, and time-series analysis — are applied at this stage. For embedded case studies, within-case analysis proceeds at both levels: the embedded units are analysed first (e.g., what happened in each of the three departments), and then these sub-analyses are integrated to produce conclusions at the level of the primary case (the hospital as a whole). Evidence that contradicts the theoretical propositions must be given equal analytical attention — disconfirming evidence is not a problem to be minimised but an analytical resource that forces theoretical refinement.
Yin, 2018, pp. 160–206; Miles, Huberman & Saldaña, 2020, pp. 290–328
When the design involves multiple cases, cross-case analysis examines what the cases share and how they differ, and interprets those convergences and divergences in terms of the theoretical propositions. Cross-case synthesis is not about averaging findings across cases; it is about identifying which propositions held across cases (literal replication confirmed) and where and why cases diverged (theoretical replication pattern). Tables and matrices that display each case's evidence for each proposition side-by-side are standard tools at this stage (Miles, Huberman & Saldaña, 2020, pp. 290–328). The goal is to elevate the analysis from case-specific description to theoretical insight that reaches beyond any individual case.
Yin, 2018, pp. 160–165; Miles, Huberman & Saldaña, 2020, pp. 290–328
07
Write the Case Study Report
The case study report is a structured analytical account, not a narrative story. Yin (2018, pp. 210–240) identifies six report structures: linear-analytic (most common in doctoral work), comparative, chronological, theory-building, suspense, and unsequenced. For doctoral research, the linear-analytic structure — research problem, methods, findings organised by research questions, discussion in relation to theory, conclusions — is typically required by institutional convention. The report must present the findings with sufficient evidence for the reader to judge the conclusions independently. Thin description — stating conclusions without showing the evidence trail — is the most common quality failure in case study writing. The report is situated in the existing literature at the discussion stage, not as an introduction: the theory serves as the lens through which the findings are interpreted, not the framework that predetermines what is found.
Yin, 2018, pp. 210–240; Merriam & Tisdell, 2016, pp. 228–252
On the Difference Between Case Study Analysis and Case Study Description
The most frequently cited quality failure in published case study research is that studies describe what happened in the case without analyzing what it means theoretically. A case study report that recounts the sequence of events in the hospital's EHR implementation — who was trained when, what problems arose, when adoption improved — is a case study chronicle, not a case study analysis. Analysis requires connecting those events to theoretical propositions, identifying which patterns the evidence confirms or challenges, explaining the mechanisms that produced the observed outcomes, and making explicit claims about what the case contributes to understanding beyond itself. Description without analysis cannot support analytic generalisation and does not constitute research in the methodological sense (Yin, 2018, pp. 161–162; Tight, 2022, pp. 2054–2056; Ridder, 2021).
📚
Section 6: The Research ProcessMark complete when you can outline and justify all seven stages for a doctoral case study, including the difference between within-case analysis and cross-case synthesis
Section 07
Rigour, Quality Criteria, and Limitations
Core TheoryReading · 25 min
How case study rigour is evaluated depends entirely on the tradition within which the research is conducted. Yin's post-positivist framework applies an adapted version of the validity and reliability standards used in experimental research. Merriam's qualitative framework applies Lincoln and Guba's (1985) trustworthiness criteria. Stake's constructivist framework applies criteria oriented toward the authenticity and utility of the interpretive account. A researcher must select quality criteria consistent with the epistemological position of their chosen tradition — applying Yin's validity language to a Stake-designed study, or claiming Lincoln and Guba's trustworthiness for a study designed according to Yin's protocol, constitutes philosophical incoherence.
Quality Criteria Across Case Study Traditions — Select a framework
Yin's Four Tests
Lincoln & Guba / Merriam
Stake's Criteria
Tactics per Criterion
Yin (2018, pp. 41–45) proposes four tests for evaluating the quality of case study research, adapted from the social science standard of validity and reliability:
Construct validity: Has the researcher identified correct operational measures for the concepts being studied? Are the conclusions about the concepts actually supported by the evidence collected? Tactics: use multiple sources of evidence; establish chain of evidence; have key informants review draft case study report.
Internal validity (explanatory studies only): Have the researcher's causal conclusions been established rather than merely assumed? Has the alternative explanation been ruled out? Tactics: pattern matching; explanation building; addressing rival explanations; using logic models.
External validity: Can the domain to which findings can be generalised be specified? For single-case studies, which theoretical proposition does the case test? For multiple-case studies, does the replication logic hold across cases? Note: the relevant form of generalisability is analytic, not statistical. Tactics: use theory in single-case designs; use replication logic in multiple-case designs.
Reliability: Could another researcher, following the same procedures, reach the same findings? Tactics: use a case study protocol; maintain a case study database; document all methodological decisions.
Yin's framework is the standard for researchers working in post-positivist traditions and is the most commonly required framework in management, health sciences, public administration, and policy research.
Drawing on Lincoln and Guba (1985) and Merriam and Tisdell (2016, pp. 237–262), qualitative case study research is evaluated on four trustworthiness criteria:
Credibility (parallel to internal validity): Are the findings believable from the perspective of the participants? Tactics: prolonged engagement with the case; triangulation of evidence sources; member checking (returning interpretations to participants for verification); peer debriefing.
Transferability (parallel to external validity): Can the findings be applied to other situations? This is a reader judgment, not a researcher claim. The researcher's responsibility is to provide sufficient thick description to enable readers to judge transferability. Tactics: rich, thick description of the case and its context; purposive sampling that captures maximum variation within the case.
Dependability (parallel to reliability): Could the study be audited? Would the study be consistent if repeated? Tactics: audit trail documenting all methodological decisions; code-recode procedure; peer examination of the audit trail.
Confirmability (parallel to objectivity): Are the findings clearly grounded in the data rather than in the researcher's preferences? Tactics: reflexivity; confirmability audit; triangulation; negative case analysis.
Stake's (1995, 2010) quality criteria are less systematised than Yin's or Merriam's but have been elaborated by subsequent scholars into recognisable evaluative dimensions:
Authenticity: Does the case study accurately represent the multiple, sometimes conflicting perspectives of participants within the case? An authentic case study does not flatten participant diversity into a single analytical account.
Naturalistic generalisation: Does the case study provide sufficient experiential detail for readers to recognise patterns applicable to their own situations? The test is whether readers can draw on the case to inform their own practice or judgment, not whether the researcher has established generalisable propositions.
Triangulation: Has the researcher consulted multiple observers, multiple perspectives, and multiple theoretical frameworks to check that an interpretation is not merely a product of one particular viewpoint?
Researcher self-disclosure: Has the researcher been transparent about their own perspective, prior experience, and relationship to the case — factors that shape interpretation in constructivist inquiry?
Stake's criteria are most commonly applied in educational evaluation, programme studies, and social work research where interpretive richness and reader utility are valued over causal explanation and theoretical generalisation.
The following table maps specific procedural tactics to each quality criterion across traditions. Each tactic should be explicitly described in the methodology chapter of a doctoral thesis.
Criterion
Specific Tactics
Phase
Construct validity
Multiple sources of evidence; chain of evidence; key informant review
Case study protocol; case study database; documented decision trail
Data collection
Credibility
Prolonged engagement; triangulation; member checking; peer debriefing
Throughout
Transferability
Thick description; purposive maximum-variation sampling within case
Reporting
Known Limitations of Case Study Research
The Generalisability Objection: "You Can't Generalise from One Case"Most common critique▾
This is the most frequently levelled critique against case study research — and it is based on a categorical confusion between two types of generalisation. Statistical generalisation proceeds from a representative sample to a population: it requires random or probability sampling, and a sample of one is statistically meaningless. Case study research does not and cannot make statistical generalisations. It makes analytic generalisations: the case is used to test, extend, or refine a theoretical proposition. A single experiment does not represent a population, but it can test a theoretical prediction. The logic is identical. When a researcher asks "Can this hospital's experience with EHR implementation tell us anything about organisational change theory?", they are asking an analytic, not statistical, question. The answer depends on how well the case tests the theoretical proposition, not on how representative the hospital is of all hospitals. Doctoral researchers must be able to articulate this distinction fluently — it is routinely examined at viva (Yin, 2018, pp. 20–22; Tight, 2022, pp. 2052–2053; Ridder, 2021, pp. 139–155).
Researcher Bias and the Integrity of Case SelectionValidity threat▾
Because cases are purposively selected and because the researcher conducts sustained, close engagement with the case site, case study research is particularly vulnerable to confirmation bias — the tendency to collect, attend to, and report evidence that confirms prior expectations while neglecting disconfirming evidence. Yin's tactic of explicitly searching for rival explanations is the primary safeguard; if the researcher does not actively look for evidence that contradicts the theoretical propositions, they will typically not find it even when it exists. The second major safeguard is the case study protocol: propositions and data collection procedures specified before fieldwork begins are harder to retrospectively adjust to fit unexpected findings. A third safeguard is the maintenance of a complete case study database: an examiner who can inspect all collected evidence, not just the excerpts that appear in the report, can assess whether the analysis was selective (Yin, 2018, pp. 44–45; Miles, Huberman & Saldaña, 2020, pp. 309–312).
Time and Resource DemandsPractical limitation▾
Case study research is substantially more resource-intensive than single-method qualitative designs. Negotiating access to a case site — particularly an organisation such as a hospital, school, or government agency — requires sustained relationship management, ethical approvals from multiple bodies, and time investments that cannot be fully planned in advance. Collecting evidence from multiple sources simultaneously requires the researcher to be in the field for extended periods. Maintaining the case study database and protocol documentation throughout the study adds administrative burden beyond what descriptive qualitative designs require. Doctoral researchers should budget realistically: a single-case embedded study of an organisation typically requires 12–18 months of active fieldwork and analysis, including access negotiations, data collection from multiple sources, within-case analysis, and write-up. Multiple-case designs require proportionally more time per case, though each subsequent case becomes more efficient once the protocol is established (Yin, 2018; Merriam & Tisdell, 2016, pp. 52–54).
The "Exceptional Case" Problem: When the Case Is Too Unusual to Generalise FromDesign limitation▾
The very features that make a case interesting — its uniqueness, its extreme characteristics, its revelatory nature — can simultaneously limit the reach of its analytical generalisations. A case selected because it is unusual is, by definition, not representative of typical cases. This is not a fatal problem if the research question is genuinely about the unusual — but it becomes a problem when researchers overstate the theoretical implications of their findings. The hospital selected for EHR implementation during COVID-19 is an extreme case: it experienced conditions (maximum organisational disruption, staff depletion, resource scarcity) that most hospitals implementing EHR do not face. The theoretical propositions it tests are valid, but the scope conditions of any resulting theoretical claims must explicitly acknowledge that the findings apply most directly to organisations facing implementation under crisis conditions — not to implementation generally (Yin, 2018, pp. 47–52; Ridder, 2021, pp. 139–141).
Primary References for This Lesson
Yin, R. K. (2018). Case study research and applications: Design and methods (6th ed.). Sage.
Stake, R. E. (1995). The art of case study research. Sage.
Stake, R. E. (2005). Multiple case study analysis. Guilford Press.
Merriam, S. B., & Tisdell, E. J. (2016). Qualitative research: A guide to design and implementation (4th ed.). Jossey-Bass.
Miles, M. B., Huberman, A. M., & Saldaña, J. (2020). Qualitative data analysis: A methods sourcebook (4th ed.). Sage.
Tight, M. (2022). Case study research revisited. Higher Education Research & Development, 41(6), 2048–2060. https://doi.org/10.1080/07294360.2021.2000990
Ridder, H. G. (2021). Case study research: Approaches, methods, contributions to theory. Schäffer-Poeschel.
Creswell, J. W., & Guetterman, T. C. (2024). Educational research: Planning, conducting, and evaluating quantitative and qualitative research (6th ed.). Pearson.
Creswell, J. W., & Poth, C. N. (2018). Qualitative inquiry and research design: Choosing among five approaches (4th ed.). Sage.
Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry. Sage.
📚
Section 7: Rigour & Quality CriteriaMark complete when you can apply quality criteria from the appropriate tradition and respond to the generalisability objection
Quiz
Knowledge Check
Assessment6 Questions · Doctoral Level
Question 1 of 6Definition & Design
A doctoral researcher is studying how a rural municipal government in the Philippines managed the transition from paper-based to digital procurement systems. She plans to interview 25 municipal employees, collect all internal memos and policy circulars related to the transition, observe three procurement committee meetings, and analyse the digital procurement system's transaction logs from 2021 to 2023. She defines the municipality as the unit of analysis and boundaries the study to the procurement department during this period. Which of the following most accurately describes this research design?
A phenomenological study, because the researcher is exploring how employees experienced the transition to digital procurement
A single-case embedded case study using multiple sources of evidence, with the municipality as the case and the procurement department as the embedded unit of analysis
A single-case holistic case study, because the researcher is examining the entire municipal organisation through one bounded inquiry
A mixed-methods study, because the researcher is using both quantitative transaction logs and qualitative interview data
Question 2 of 6Unit of Analysis
A researcher conducting a case study of a national teacher professional development programme interviews 40 teachers across 10 schools, surveys 200 teachers for quantitative data, and analyses programme documents. In the findings chapter, he writes: "This case study of 40 teachers found that professional development had minimal impact on classroom practice." A doctoral examiner objects that the researcher has made a fundamental design error. What is that error, and why does it matter?
The researcher used too many participants — case study should not involve more than 15 interviews
The researcher conflated the data sources (the 40 teachers) with the unit of analysis. If the case is the national programme, conclusions must be made about the programme as a bounded system — not about the 40 individuals as if they were 40 separate cases
The researcher should not have included a survey because case study is exclusively a qualitative design
The researcher needed to conduct a cross-case analysis before drawing conclusions about programme impact
Question 3 of 6Replication Logic
A researcher studying how public universities implement research commercialisation policies has completed a thorough case study of one university (University A) and found strong evidence that institutional leadership commitment was the primary driver of successful policy adoption. She now wants to add two more cases. According to Yin's replication logic, which of the following case selection strategies is most methodologically defensible?
Select two additional universities randomly from the national university registry to increase statistical representativeness
Select one university predicted to produce the same result (strong leadership, successful adoption) for literal replication, and one predicted to produce a contrasting result (weak leadership, poor adoption) for theoretical replication — each selection justified by prior knowledge of institutional conditions
Select two universities from the same region as University A to control for geographic confounds
Select two universities that are most conveniently accessible to reduce data collection costs and time
Question 4 of 6Traditions & Quality Criteria
A doctoral researcher has conducted an intrinsic case study of a single community-based disaster recovery initiative, following Stake's constructivist tradition. She has written rich, contextually detailed descriptions of community members' experiences, included multiple participant perspectives, and provided sufficient contextual detail for readers to judge applicability to their own situations. Her thesis examiner questions whether the findings can be "generalised." The researcher argues that generalisation in the statistical sense is not the appropriate evaluative criterion for this study. Which of the following most accurately represents the methodologically defensible position?
The examiner is correct: all doctoral research must be able to make statistical generalisations to have scholarly value
The researcher is correct: within Stake's constructivist tradition, the appropriate form of knowledge transfer is naturalistic generalisation — readers draw on the case to inform their own situated judgments, and this requires sufficient thick description, not a generalisable sample
Both the examiner and the researcher are partially correct: the researcher should add a survey to provide statistical generalisation while retaining the qualitative case
The researcher should have selected Yin's framework instead of Stake's, because Yin's analytic generalisation provides a more academically defensible basis for doctoral conclusions
Question 5 of 6Triangulation & Evidence
A researcher conducting a case study of a hospital's patient safety improvement programme finds the following: (1) Twelve out of fifteen nurses interviewed report that the new incident-reporting system has improved their willingness to disclose near-misses. (2) The hospital's incident report database shows a 340% increase in reported near-misses since the new system was introduced. (3) Direct observation of three ward safety briefings reveals that senior nurses routinely praise incident reports and explicitly discourage blame language. (4) The clinical governance committee's minutes document a formal commitment to no-blame reporting culture adopted eighteen months before data collection. How should the researcher interpret the relationship among these four evidence sources?
The four sources produce redundant findings and only one — preferably the interview data — needs to be reported in the findings chapter
The four sources triangulate on the same conclusion through independent channels — subjective perception (interviews), objective behaviour (database records), real-time observation (briefings), and formal organisational commitment (minutes) — each adding a distinct, irreplaceable dimension to the finding and collectively strengthening construct validity
The quantitative database record (340% increase) is the most reliable source and should anchor the finding, while the qualitative sources serve only as illustrative examples
The researcher should treat the four sources as a mixed-methods study and analyse each independently before integrating at the conclusion stage
Question 6 of 6Analysis & Generalisation
A doctoral researcher has completed a single-case embedded case study of an e-government implementation in a Philippine local government unit. Her analysis identifies a pattern: implementation progress was significantly slower in departments with weak internal IT governance structures than in departments with dedicated IT coordinators. This pattern matches her theoretical proposition, derived from existing organisational change literature, that internal IT governance capacity is the primary determinant of digital adoption speed. She now wants to state the contribution of her research. Which of the following best represents a methodologically defensible generalisation from this study?
"These findings demonstrate that all Philippine local government units with weak IT governance will adopt e-government slowly, which has implications for national digital policy."
"Because only one organisation was studied, no generalisation of any kind is possible, and the study's contribution is limited to the description of this specific case."
"The empirical pattern observed in this case provides analytic support for the theoretical proposition that internal IT governance capacity shapes digital adoption speed. This extends and refines the existing theory by demonstrating the mechanism's operation in a resource-constrained Philippine local government context — a population not previously examined in this literature."
"This study should be replicated with a random sample of 50 local government units before any theoretical conclusions can be drawn."