Case Study Methodology: pick the method, then defend it

Method Match: three questions that name your design

Methodology sections go wrong before the first sentence: the writer picks a method because it sounds impressive, and every later block argues with that choice. The match tool reverses the order. Question one asks what kind of data answers your question — meanings and processes (qualitative), numbers on a fixed instrument (quantitative), or both in sequence (mixed). Question two asks how deep you can honestly go — one case studied well (single), or several cases compared (multiple). Question three asks what the study is for — opening an unexplored situation (explore), documenting how it works (describe), or arguing why it happened (explain).

The output is deliberately name-dropping: an explanatory single-case study or a multiple-case design with literal replication is vocabulary your marker expects, and each recommendation arrives with the reason it fits plus two or three alternatives — because most methodology sections defend a choice, not just state one. Your answers save to this browser (key csw-csm-v1) alongside the builder's blocks.

Two readings make the recommendation honest rather than a horoscope. The goal question decides the design's job: an exploratory study earns its keep by mapping terrain, a descriptive one by documenting a working system, an explanatory one by defeating at least one rival account of cause. The depth question decides the replication logic: a single case must justify itself as revelatory, unusual or a critical test, while multiple cases must say why each one repeats or contrasts the finding. When your goal is to argue a claim in one sentence first, the thesis statement generator drafts that sentence — the design then has a promise to keep.

The methodology skeleton: five blocks, one paragraph

Whatever the match names, the methodology section itself has the same five-part spine. Draft it block by block, and the builder assembles the blocks into a paragraph with the gaps visible — so you can see the shape of the section before the prose exists:

• Research design — the named design and the case boundary: what is inside the case, what is context, and why this design can answer the question.
• Data collection — the sources and the plan: who is interviewed, what is observed, which documents and records, over what period.
• Analysis — the path from material to finding: thematic coding, pattern matching, cross-case synthesis, explanation building.
• Validity — the trustworthiness moves: triangulation across sources, member checking, a documented chain of evidence.
• Limitations — the limits admitted in advance, because a methodology that claims no weakness invites the marker to supply them.

The frame chips under each block insert standard, honest framings — adapt them to your case rather than stacking all of them. Four hundred characters per block is a draft budget: enough for the move, not for the final prose.

What the builder refuses to do: it never writes your content. The chips show the shape of an honest block — a named design, a sourced plan, a stated path — not the substance. And the assembled paragraph is scaffolding: paste it into your proposal and expand each sentence until it names your actual case. When the wider document is next, the how-to-write-a-case-study skeleton takes over section by section, and the research paper abstract formula compresses the finished design into its six sentences.

The three method families, without the folklore

Method folklore says qualitative case studies are the real ones and everything else is a survey wearing a costume — or that numbers make a case study scientific. Both claims confuse the genre with one method family. A case study is defined by its unit (a bounded system in context), not by its data type; each family just changes what counts as evidence.

Qualitative — the default for how and why

Qualitative case studies collect meanings, decisions and processes: interviews with the people who lived the event, observation of the setting, documents written inside it. The family fits questions that begin with how or why and cannot be answered by a frequency count — how a team absorbed a restructuring, why one branch outperformed its twins. Its discipline is depth: fewer sources, read closely, with the analysis path stated before the material is touched. Most student case studies live here — the dedicated chapter builder for the qualitative case study methodology is this page's sibling.

Quantitative — legitimate when indicators exist

A quantitative case study measures one bounded system against fixed indicators: attendance counts, defect rates, agreement scores, archival performance records. The family fits questions of extent inside the case — how much, how often, how far — and its discipline is instrument honesty: the indicators must exist before the study starts, and the case's context limits what the numbers can claim. What it cannot do is manufacture statistical generalization; one organization is not a population. If your question keeps drifting toward variables you can pre-state, check whether the honest home for a pre-stated hypothesis test is a different genre — the case study vs experimental research page walks that line.

Mixed — a sequence, not a blend

Mixed-methods case studies run both families in a stated order, and the order is the design. Exploratory sequential (qualitative first) maps the terrain, then measures what surfaced. Explanatory sequential (quantitative first) establishes the pattern in the numbers, then interviews to learn why the pattern holds. Convergent runs both in parallel and merges them at the interpretation. The discipline is the handoff: each strand must name what it hands the other, or the study becomes two unconnected projects sharing a word count.

The three match questions, graded

The match tool compresses the three decisions markers actually grade. Here is each one with the wording that earns credit.

Question 1 — the data: what would count as an answer?

Write the answer your question implies, then look at its form. "Because the staff distrusted the rollout" is a meaning — qualitative. "Because defect rates fell 18 percent" is a measurement — quantitative. "Both: the number fell, and the interviews explain the fall" is a sequence — mixed. The mistake to avoid is choosing the method first and bending the question toward it; the research paper methodology recipe calls this rerun-proofing: state the question, then the evidence it needs.

Question 2 — the depth: one case or several?

Depth is a budget question disguised as a courage question. One case lets you open the black box — decisions, meetings, the texture of practice — but binds every claim to that boundary. Multiple cases buy contrast, at the price of depth per case and a replication logic you must state. The honest test: could you name, before collecting data, what would make a second case repeat your first (literal replication) or contrast with it for a stated reason (theoretical replication)? If not, one well-bounded case is the stronger study.

Question 3 — the goal: explore, describe, or explain?

The goal decides the design's name and its burden of proof. Explore promises terrain: the finding is a map of what is there, and its quality is coverage. Describe promises a working picture: the finding is the system shown operating in context, and its quality is accuracy. Explain promises an argument: the finding defeats the plausible rival account, and its quality is the survival of that contest. Sections fail when the goal stated in the design and the analysis actually run disagree — an exploratory design that smuggles in causal claims is the classic version.

The named designs, and what each one promises

The standard vocabulary is worth learning because markers use it. The classic typology sorts designs on two axes — one case or several, and holistic or embedded units — then lets the goal name the variant. Four rows cover most student studies:

Read the risk column before the fit column: most graded methodologies lose their marks on the risk they ignored, not the design they chose. The lab report method guide makes the same point from the experiment side — a methods section is graded on whether someone could rerun the study, and every genre's method section inherits that test.

Data collection: the four source families

A methodology section names sources, not "data" in the abstract. Four families cover the field, and a defensible case study usually draws on at least two — which is already the beginnings of triangulation.

Interviews are the workhorse. Semi-structured is the default format: a question route, asked in the participant's order of importance, with follow-ups allowed. The discipline is role diversity — the people who sponsor a change and the people who operate it rarely describe the same event.

Direct observation catches what interviews sanitize: the actual meeting, the actual queue, the workaround nobody mentions. Its cost is footprint — observation changes settings — and its method section must state when, where, and in what role the researcher stood.

Documents and archives are the quiet strength of case studies: minutes, reports, dashboards, correspondence. They predate the study, so they cannot perform for it. Their discipline is provenance: who wrote each document, for whom, and when — because a document is evidence of its occasion as much as of its content.

Physical artifacts and indicators cover what remains: the product, the config file, the attendance log, the archived metrics. In quantitative and mixed designs this family carries the fixed-instrument side; in qualitative designs it anchors claims that would otherwise float on memory.

Triangulation: the defeat of the single-source objection

Triangulation is the methodology section's strongest validity move and its most abused word. Done properly, it is not "I used lots of data"; it is one finding placed where at least two angles could disagree with it — and survived the disagreement. The classic four angles:

• Data-source triangulation — the same finding appears in interviews and in documents; the strongest and cheapest upgrade.
• Method triangulation — the finding survives a second method: the survey agrees with the interview theme.
• Investigator triangulation — a second coder or observer reaches the same reading of the material.
• Theory triangulation — the finding holds under two competing theoretical readings, not just the researcher's favorite.

A workable floor for a graded study: two source types crossed on at least one central finding, named in the validity block, with the crossing shown in the analysis — not asserted in a sentence and abandoned. Everything beyond the floor is margin, and margins are where the word count goes to hide. The trustworthiness vocabulary markers expect (credibility, transferability, dependability, confirmability) is mapped to concrete moves in the sibling guide on qualitative case study methodology.

One build, five blocks: an illustrative example

Here is the invented build the Load example button fills — a mixed single-case study of a grocery chain's click-and-collect rollout, chosen because it shows the five blocks pulling in one direction. None of it is saved; your case replaces every word.

Read the five against each other: the design names the boundary the data respects, the data names the sources the analysis codes, the analysis names the path the validity moves protect. That alignment — each block pointing at the previous one — is what "methodology" means when a marker says the section holds together.

What this page cannot do

The honest limits, because a tool that states them is the only kind worth trusting with a graded methodology:

• It does not choose your topic or negotiate access. The match assumes you have a case and a question; whether the gatekeepers let you in is a conversation this page cannot have.
• It does not verify feasibility. A multiple-case design with twelve interviews sounds rigorous and can be impossible in your six weeks. Budget the design against your real calendar before you commit it to paper.
• The named designs are defaults, not laws. Your course may teach a specific template, and your marker's rubric outranks any typology — including the one this page uses.
• Four hundred characters is a draft budget. The blocks hold the moves; the final section needs your case's specifics expanded around them.
• The example is invented. The grocery chain, the fifteen interviews and the four weeks of records are illustrative teaching material. A real methodology cites real, obtained data only.

Kept in its lane, the page does one thing well: it turns "what methodology should I use?" into a named design and a five-block draft you can defend. The study guides hub holds the rest of the case study shelf, and the free writing tools — word counter, readability checker, citation generators — handle everything after the draft.

Quiz: five questions on the methodology

Five questions on the decisions this page makes — data families, replication logic, triangulation, and honest limits. Every wrong answer reveals the right one, and four options keep the guessing honest.

Frequently asked questions

Did the match name the design you will defend? 👍 Yes 👎 Not yet