How to Write a Case Study

First: which case study have you been set?

Most of the confusion around case studies comes from the fact that the phrase covers three separate genres:

• A case study analysis. Common in business, management and law. You're given a case — a company, a decision, a dispute — and asked to diagnose the problem and recommend action. The deliverable is a recommendation.
• A research case study. A qualitative or mixed-methods study that investigates one bounded instance in its real-world context. Common as a dissertation chapter. The deliverable is a finding.
• A clinical or professional case report. Medicine, nursing, psychology, social work. One patient or client, written up so others can learn from the encounter. The deliverable is a documented lesson.

Read the brief for the verb. "Recommend", "advise" and "evaluate the options" point to the first. "Investigate", "explore" and "what factors explain" point to the second. If your brief says "report" rather than "essay", our guide to essay vs report structure covers the difference in more detail.

Choose the case — and be able to say why

When you get to pick your own case, the choice is part of the argument. Robert Stake's The Art of Case Study Research (1995) separates three purposes:

• Intrinsic — the case itself is what interests you.
• Instrumental — the case is a way to understand a wider issue; the case is secondary.
• Collective — several instrumental cases studied together.

Stake's own point is that the difference lies in your purpose, not in the case: the same hospital can be an intrinsic or an instrumental case depending on what you want from it. Say which you're doing in the methodology.

Robert Yin's framework crosses two further decisions — single or multiple case, holistic or embedded — giving four designs: single-holistic, single-embedded, multiple-holistic and multiple-embedded. "Embedded" means you examine more than one unit of analysis inside the case: one school (the case) but three year groups (the units).

Structure 1: the case study analysis

1. Executive summary. The recommendation, up front. Written last.
2. Background. Only the facts a reader needs to follow the analysis — not a retelling of the case.
3. Problem statement. One or two sentences naming the actual decision to be made.
4. Analysis. Evidence applied through a named framework, with the case's own figures.
5. Alternatives. Two or three real options, each with its trade-off.
6. Recommendation and implementation. One option, chosen on stated criteria, with steps, owners and timing.

The commonest failure here is a long, faithful summary of the case followed by a short opinion. Markers are reading for the ratio: the analysis and recommendation sections should be the bulk of the word count.

Structure 2: the research case study

1. Introduction. The problem, the research question, why a case design suits it.
2. Literature review. The theory this case will test, extend or complicate.
3. Methodology. Case selection and justification, the boundaries of the case, the unit of analysis, your design in Yin's or Stake's terms, data sources, analysis approach, ethics approval.
4. Findings. What the evidence shows, organised by theme rather than by source.
5. Discussion. What it means against the literature, plus limitations.
6. Conclusion. The answer to the question, and what it changes.

The discussion is where most of the marks sit, and it follows the same moves as any empirical paper — we break them down in how to write a discussion section.

Structure 3: the clinical case report

Don't improvise this one. The CARE guidelines (2013) give a 13-item checklist for clinical case reports and are endorsed by the EQUATOR Network; many journals require them. The items run from title and key words through patient information, clinical findings, timeline, diagnostic assessment, therapeutic intervention, follow-up and outcomes, discussion, and patient perspective and informed consent. The timeline and the patient perspective are the two most often left out — and the two that make a case report reusable by anyone else.

Gather evidence from more than one source

Yin identifies six sources of case study evidence: documents, archival records, interviews, direct observation, participant observation and physical artefacts. They are complementary, and the point of using several is triangulation — when an interview claim, a minute in a document and an observed practice all agree, your finding is far harder to dismiss. When they disagree, say so: a contradiction between what an organisation documents and what it does is often the most interesting result in the study.

Be careful what you claim at the end

A single case cannot tell you how common something is. But that is not the same as telling you nothing — Bent Flyvbjerg's Five Misunderstandings About Case-Study Research (Qualitative Inquiry, 2006, 12(2), 219–245) takes on exactly the belief that you cannot generalise from one case and argues that a discipline without well-executed cases lacks exemplars altogether.

The practical version, in your conclusion:

• Don't write: "This shows that most firms in the sector under-invest in training."
• Do write: "This case is difficult to reconcile with the assumption that under-investment is driven by cost alone; here the constraint was scheduling."

The second sentence is a claim about theory, and one case is legitimate evidence for it.

Four mistakes that cost marks

• Description instead of analysis. If a paragraph could be replaced by the source document, cut it.
• No stated unit of analysis. The study then quietly becomes about everything.
• Recommendations with no evidence trail. Every recommendation should point back to a specific finding.
• Uncited case material. Company figures, policy documents and interview quotes all need attribution — including the case document itself.

Frequently asked questions

Keep reading

• Essay vs Report: What's the Difference?
• How to Write the Discussion Section of a Research Paper
• How to Write a Lab Report: Structure, Sections & Examples
• Free APA Citation Generator