The case study is the most misunderstood design in doctoral research — dismissed as 'just one company' by critics and abused as 'whatever I could access' by weak studies. Done properly, it is the method of choice for how and why questions about contemporary phenomena you cannot control — and it has a rigour toolkit as demanding as any survey.
When a case study is the right design
- Your question is how or why something happens, not how much or how many.
- The phenomenon can't be separated from its context — a merger, a digital transformation, a policy rollout.
- You need multiple evidence sources (interviews, documents, observation, numbers) around one bounded unit.
Define the case — the step everyone skips
A case is a bounded system: an organisation, a programme, a project, an event — with explicit boundaries in time and scope. 'Digital payment adoption in India' is a topic; 'UPI adoption in three Pune cooperative banks, 2020–2024' is a case design. If you cannot state what is inside and outside the case, you don't yet have one.
The design matrix: four types
- Single holistic — one case, one unit of analysis. Justify the case as critical, extreme/unique, typical, revelatory or longitudinal.
- Single embedded — one case with sub-units (e.g. one company, four departments) — adds internal comparison.
- Multiple holistic — several cases compared whole-to-whole using replication logic.
- Multiple embedded — several cases each with sub-units; the richest and heaviest design.
Yin treats case study as quasi-experimental: propositions up front, replication logic across cases, a case-study protocol and database. Stake treats it as interpretive: the intrinsic value of the case, progressive focusing, the researcher as instrument. Committees accept both — what they flag is unknowingly mixing the vocabularies. Cite the tradition you actually follow.
Rigour: the four tests (Yin's framework)
- Construct validity — multiple evidence sources, a chain of evidence, key informants reviewing the draft case report.
- Internal validity (explanatory cases) — pattern matching and explanation building, plus rival-explanation testing.
- External validity — analytic generalisation to theory, not statistical generalisation to populations; multiple cases strengthen it via replication.
- [Reliability](/blog/reliability-and-validity-explained) — a case-study protocol and a case-study database an auditor could retrace.
How many cases for multiple-case designs?
Follow replication logic, not sampling logic: 2–3 cases predicted to show similar results (literal replication), optionally plus cases predicted to differ for theoretical reasons (theoretical replication). Four to six information-rich cases is the practical doctoral ceiling — depth per case falls fast beyond that.
The viva question you will definitely get
'How can you generalise from one case?' The answer: case studies generalise to theory, the way experiments do — a single well-chosen case can extend, refine or refute a theoretical proposition (analytic generalisation). Rehearse this answer; it is asked in nearly every case-study viva. More viva preparation in our viva guide.
Case selection, protocol design and cross-case analysis are exactly the decisions our qualitative research mentoring works through with scholars — before access negotiations lock you into the wrong case.
Frequently asked
Is a case study qualitative or quantitative?+
Neither inherently — it's defined by its focus on a bounded case, not by data type. Most doctoral case studies are qualitative-dominant, but strong ones often embed numbers (performance data, survey snapshots) among their evidence sources.
Can a single case study be a PhD?+
Yes — if the case is justified as critical, unique, typical, revelatory or longitudinal, and the analysis achieves theoretical depth. Single-case PhDs succeed on the quality of theorising, not the number of cases.
What is the difference between a case study and ethnography?+
Ethnography commits to prolonged immersion in a culture and thick description of shared meaning; a case study bounds a system and mobilises multiple evidence types around it, often in weeks not years. A case study may use ethnographic techniques without being an ethnography.
phdguide's mentors are senior academics, former supervisors, statisticians and publication specialists with 25+ years' average experience guiding MBA, MPhil and PhD scholars from topic to viva.
Ethical, compliant guidance: We provide academic support, mentoring, analysis, editing and structuring — not authorship. Your work stays compliant with university policies.