---
name: research-design-and-methodology-chapter
description: >
  Writes and justifies the methodology chapter of a dissertation: philosophical
  position at proportionate depth, approach and strategy, population and
  sampling named honestly, instrument adoption or development, reproducible
  procedure, a specified analysis plan, quality criteria in one coherent
  vocabulary, limitations and ethics. Use for "methodology chapter", "chapter 3",
  "justify my research design", "how much epistemology do I need", "convenience
  sample", "validity and reliability section", "trustworthiness criteria",
  "my methodology just describes what I did".
category: 15 Academic University Research
ref: "15.08"
tier: 3
inherits: [K2, K3, K4, K5]
---

# Research Design and Methodology Chapter

## 1. One-line description

A method for writing the methodology chapter as a sustained justification rather than a description, in which every choice is argued against the alternatives it beat, the sample is named honestly, the analysis is specified rather than gestured at, and the quality criteria come from one coherent tradition.

## 2. What this skill is used for

**The research problem it solves.** "Describes without justifying" is the most common criticism written on methodology chapters, and it is not a stylistic point. A chapter that says the study used semi-structured interviews with twelve participants recruited through the researcher's network has told the reader what happened and nothing about whether it should have. The examiner's questions are all the ones the chapter did not answer: why interviews rather than a survey, why twelve, why that recruitment route, what it excludes, and what the findings can therefore claim. Four further failures cluster here. Paradigm discussion is either absent where the discipline requires it or inflated into six pages of philosophy with no consequence for the design. Convenience samples are described in language that implies representativeness. Quality vocabularies are mixed incoherently, so a chapter claims both external validity and transferability without noticing that these come from different traditions and rest on different logics. And limitations appear only in the final chapter, where they read as an afterthought rather than as design honesty.

**Where it sits in the research lifecycle.** After the framework fixes what must be observed, and before ethics submission and fieldwork, since the ethics committee assesses the procedure this chapter specifies. It is revised after fieldwork to report what was actually done, which is almost never identical to what was planned.

**Typical use cases.**
- Writing chapter three of a masters dissertation from a design that exists only as intentions.
- Converting a description into a justification after supervisor feedback.
- Deciding how much philosophical positioning the discipline actually requires.
- Naming and defending a non-probability sample without overclaiming from it.
- Justifying the adoption of an existing instrument, which requires as much argument as building one.
- Specifying an analysis approach precisely enough that a reader could repeat it.
- Choosing between the validity vocabulary and the trustworthiness vocabulary, and using one consistently.

**Who uses it.** Masters candidates writing or revising the methodology chapter; supervisors reviewing a design before ethics submission; students across quantitative, qualitative and mixed traditions, in any discipline where a methodology chapter is examined.

## 3. When to use it

- The dissertation requires a methodology chapter and the design decisions have been made but not argued.
- A supervisor has returned the chapter with "you describe, you do not justify".
- You need to decide how much epistemology belongs in the chapter for your discipline.
- The sample is a convenience sample and you need to say so without destroying the study's standing.
- You are choosing between adopting a published instrument and building your own.
- The analysis section currently says "thematic analysis will be used" and nothing more.
- You need to write the quality criteria section and are unsure which vocabulary applies.
- The chapter must be revised post-fieldwork to report what actually happened.

## 4. When NOT to use it

- **The design has not been chosen yet.** This skill justifies and documents a design; it does not select one from a blank page. Choosing between survey, experiment, interview, observation, case study, secondary analysis and mixed approaches against a research question is **01.04 Research Method Selection**, which this skill sits downstream of. In the other direction: 01.04 stops when the method is chosen and does not write an examined chapter, does not handle paradigm positioning, and does not produce the trustworthiness section.
- **The sampling problem is the whole problem.** Frame definition, probability and non-probability approaches, sample size logic, quota construction and the consequences of each for inference are **01.06 Sampling Strategy**. This skill states the sampling decision, names it honestly and reports its consequences for the claims; it does not teach sampling theory.
- **The instrument itself needs designing.** Question wording, response formats, scale construction, item order and bias detection are **02.01 Survey Questionnaire Design**, **02.03 Interview Question Development** and **02.07 Scale and Measurement Selection**. This skill justifies the instrument decision and reports its provenance and validation evidence; it does not write items.
- **The requirement is a proposal rather than a chapter.** A proposal is shorter, forward-looking, and assessed on feasibility and defensibility rather than on execution. Use **15.04 Research Proposal Writing**, or **01.08 Research Proposal and Scope Development** for applied work. This skill takes over once the study is running or complete.
- **Doctoral-standard methodological defence is required.** Where every choice must be defended against named alternatives, validity threats enumerated and addressed individually, and the philosophical position sustained through the whole thesis, that is **15.16 Doctoral Methodology Justification and Rigour**. This skill is calibrated to masters expectations, where a proportionate justification is sufficient and an exhaustive one is padding.
- **The design has a fatal flaw that writing cannot fix.** A study that cannot reach its population, whose instrument does not measure its constructs, or whose analysis cannot answer its questions has a design problem, not a writing problem. Writing a persuasive chapter around it wastes the effort and delays the discovery. Say so, and go back to the design. `RESEARCHER DECISION REQUIRED` (K5 §2.7).
- **Your institution prohibits AI assistance for this task.** See §12 item 1.

## 5. Required inputs

**Required.**
- **The research questions or hypotheses, final.** Every justification in the chapter is a justification relative to these. A chapter written against draft questions has to be rewritten when they settle.
- **What was actually decided or actually done.** Real numbers, real recruitment routes, real dates, real dropouts. A chapter cannot be written from intentions once fieldwork has happened, and inventing the procedure is fabrication (K4 §2.5).
- **The departmental requirements and the marking rubric for the chapter.** Weighting, required subsections, expected length, and whether the department expects a philosophy section at all.

**Optional, and what each one adds.**
- **The conceptual framework and operationalisation table.** Turns the instrument section from assertion into a demonstrable link between constructs and items. Without it, the chapter cannot show that the instrument measures the framework.
- **Recent accepted dissertations from the same department.** The most reliable available evidence of what this department expects, including how much paradigm discussion is normal here. Worth more than any general guidance.
- **The published source and validation evidence for an adopted instrument.** Lets you report reliability and validity evidence from prior use rather than claiming it, and is required if you are to justify adoption properly.
- **The ethics application or approval.** Ensures the procedure described in the chapter matches the procedure approved, since a discrepancy between the two is a serious problem.
- **A fieldwork log.** Records what deviated from plan, which is the material for the honest procedure section and for the limitations.
- **The analysis software or approach available and the student's competence with it.** A specified analysis nobody can execute is not a plan.

## 6. Questions to ask before starting

1. **How much philosophical positioning does this discipline and this department require?** The range is genuine: some fields require a sustained epistemological argument that shapes every subsequent choice, others treat more than a paragraph as padding, and both positions are defensible within their own conventions. Default if unanswered: one page connecting the position to at least two concrete design decisions, and nothing that does not earn its place by having a consequence.
2. **Is the study quantitative, qualitative or mixed, and if mixed, what is the integration point?** Determines the entire quality-criteria section and the analysis section. Default: whichever the research questions require, stated explicitly, with mixed designs required to specify where and how the strands meet, since "mixed methods" without an integration point is two studies in one document.
3. **What sampling was actually used, in plain words?** Determines what the study can claim. Default: name it accurately, including "convenience" where that is the truth, and derive the claim limits from it rather than apologising for it.
4. **Is the instrument adopted, adapted or built?** Determines what must be justified and what evidence must be reported. Default: assume it must be justified either way, because adoption without justification is as weak as unjustified construction.
5. **Has fieldwork happened yet?** Determines tense and content. A pre-fieldwork chapter states a plan; a post-fieldwork chapter reports execution including deviations. Default: write it forward-looking, and schedule the revision.
6. **What did not work, and what was changed mid-study?** This is the material that makes a chapter credible and it is routinely omitted. Default: ask directly, since students under-report deviations they assume are failures, when reporting them is a strength.
7. **Which quality vocabulary will the chapter use?** Validity and reliability, or credibility, transferability, dependability and confirmability, or a design-specific set. Default: choose one aligned to the design, use it exclusively, and never present both sets as though they were a checklist.

## 7. Step-by-step methodology

**Step 1. Establish the local expectations before writing a word.**
Read the departmental handbook's methodology requirements and the marking rubric, and skim two or three recent accepted dissertations from the same department. You are looking for four things: whether a philosophy section is expected and how long it runs, what the required subsections are and in what order, the expected length of the chapter as a proportion of the whole, and whether the department uses a particular vocabulary for quality criteria. Methodology conventions are local, and a chapter written to a generic template will be marked against a specific one. Where the department's expectations conflict with what a general methods text says, the department wins, and where they are genuinely unclear, ask the supervisor and record the answer. *Correct result: a one-page outline of required subsections with target word counts, derived from local evidence rather than from a template.*

**Step 2. Convert every planned subsection into a justification question.**
This is the mechanical fix for the "describes without justifying" criticism, and it works. Before writing, rewrite each subsection heading as the question the reader will ask: not "Research design" but "why is a cross-sectional survey the right design for these questions, and what did it beat"; not "Sample" but "why these people, why this many, and what does that exclude"; not "Data analysis" but "why this analytical approach, and how exactly will it be executed". Then write to answer the question. The structural rule that follows: **every design statement in the chapter is followed, within two sentences, by the reason and by the alternative considered.** A paragraph that states a choice and moves on has not done its job. *Correct result: a subsection plan in which each heading is paired with its justification question, and no heading whose question is unanswerable.*

**Step 3. Position the study philosophically, proportionately, and only where it pays.**
The honest guidance is that the required depth varies by discipline and the chapter must serve its own. Where a substantial epistemological discussion is expected, it must be an argument rather than a taxonomy: name the position, say what it holds about what can be known and how, and, critically, connect it to at least two decisions the study actually makes. Where the discipline treats this as padding, one paragraph stating the position and its consequence is sufficient and more is a liability. Either way, apply the consequence test to every philosophical sentence: **if the design would be identical under the opposite position, the sentence is decoration.** The reliable test of a good positioning section is that a reader can predict at least two later choices from it. Avoid the common failure of listing several paradigms in the abstract and then announcing an allegiance with no connection to the study, which reads as a textbook summary and is marked as one. *Correct result: a positioning section of a length appropriate to the discipline, in which every claim has a stated design consequence.*

**Step 4. State the design layers in order, each with its reason and its rejected alternative.**
Work down: **approach** (broadly deductive, inductive or abductive, and why the research questions demand it); **strategy** (survey, experiment, case study, ethnography, action research, secondary analysis, and so on, with the two most plausible alternatives named and rejected for stated reasons); **time horizon** (cross-sectional or longitudinal, and where cross-sectional, an explicit statement of what that costs the study's claims about change or causal order); and **methods** (the specific techniques of collection and analysis). Keep the layers consistent: an inductive approach with a fully pre-specified hypothesis test is incoherent and visible. The most valuable sentence in this section is usually the one about the rejected alternative, because it demonstrates that a choice was made rather than defaulted into. *Correct result: four layered subsections, each containing a decision, a reason tied to the research question, and a named alternative with the reason it was set aside.*

**Step 5. Describe population, frame and sample in the order that makes the gaps visible.**
Four distinct things, and collapsing them is where overclaiming begins. The **target population** is who the study is about. The **accessible population** is the part of it you could in principle reach. The **sampling frame** is the actual list or route through which people were reached. The **sample** is who took part. State each, then state the gaps between them, because every gap is a limit on the claim. Then name the sampling method honestly: probability methods by their type, and non-probability methods as what they are, convenience, purposive, quota, snowball or volunteer. **Calling a convenience sample convenient is not a weakness; concealing it is.** The correct treatment is to name it, state why it was the available option given the access and timeline, and state precisely what it means for the findings: that the results describe this sample, that generalisation to the population is not supported by the sampling design, and that any wider claim rests on argument about similarity rather than on statistics. For sample size, give the logic, not just the number: a power calculation with its assumptions for quantitative work, or for qualitative work a stated basis such as the range conventional for this design in this field, the point at which new data stopped producing new categories, or the practical limit imposed by access, each stated as what it is. Claiming saturation requires evidence that you looked for it, so say how it was judged. *Correct result: four stated populations with the gaps named, a sampling method named accurately, and a sample size with a stated logic rather than a bare number.*

**Step 6. Justify the instrument, whether you adopted it, adapted it or built it.**
All three require argument, and students routinely assume adoption is self-justifying. **Adopting** an existing instrument requires: the source, evidence of its validity and reliability in prior use, the population it was developed on and whether that matches yours, whether it measures your constructs as your framework defines them, permission where required, and the reason it beat the alternatives. An instrument validated on a different population in a different language may need re-validation, and saying so is a strength. **Adapting** requires everything above plus a full account of what was changed, why, and what the change does to the prior validity evidence, which it usually weakens. **Building** requires: the derivation of items from the framework's constructs, the item-generation process, expert or supervisor review, piloting with a described sample, what the pilot changed, and any reliability evidence obtained. In every case, include the mapping from construct to instrument component, which is what demonstrates that the instrument measures the framework rather than something adjacent. Put the full instrument in an appendix and reference it. *Correct result: an instrument section that a reader could use to judge whether the instrument measures what the study claims, plus a construct-to-item mapping table.*

**Step 7. Write the procedure so a stranger could repeat it.**
Reproducibility is the test. Include: how participants were approached and by whom, what they were told and when, where and how data were collected, the mode and setting, the duration, recording arrangements, who collected the data, the sequence of activities, and the dates. For interviews, state whether the guide was followed in order or used flexibly. For surveys, state the distribution route, the field period, reminders, and the completion and dropout figures. For observation, state the observer's role and its visibility. Then report deviations from the plan, which every study has: the site that withdrew, the recruitment route added in week three, the two interviews conducted online rather than in person because of travel. Reporting them makes the chapter credible and pre-empts the examiner's question. Concealing them makes the study unrepeatable and, if discovered, raises questions about everything else. *Correct result: a procedure a competent stranger could execute, with dates, numbers and deviations included.*

**Step 8. Specify the analysis rather than naming it.**
"Thematic analysis was used" is a label. A specification states: the analytical approach and its named variant where variants differ materially, the unit of analysis, the sequence of steps in order, who did each step, how codes or categories were developed and whether inductively or deductively or both, how disagreements were handled if more than one person coded, how the framework was used if at all, and what software supported it (as a general category, since the analytical logic must not depend on a product). For quantitative work: the tests to be run against each hypothesis or question, the assumptions each requires and how they will be checked, the treatment of missing data decided in advance, the handling of multiple comparisons, the significance threshold and effect size measures to be reported, and any planned subgroup analysis, distinguished from exploratory analysis conducted after seeing the data. Pre-specifying which analyses are planned and which are exploratory is what allows the results chapter to report both honestly. Link each analysis to the research question it answers, in a table. *Correct result: an analysis section detailed enough that another researcher could run the analysis and get the same output, plus an analysis-to-question mapping.*

**Step 9. Choose one quality vocabulary and use it coherently.**
For quantitative designs, the conventional set is reliability (internal consistency, test-retest, inter-rater as applicable), internal validity (whether the design supports the inferences drawn, and specifically what threatens it here), construct validity (whether the instrument measures the construct, with the evidence), and external validity (to whom the results extend, which for a non-probability sample is limited and must be said). For qualitative designs, the conventional trustworthiness set is credibility (whether the account is a defensible reading of the data, addressed by practices such as prolonged engagement, member checking, peer debriefing or negative case analysis), transferability (whether the reader can judge applicability elsewhere, addressed by thick contextual description rather than by claiming generalisation), dependability (whether the process is documented enough to be followed, addressed by an audit trail) and confirmability (whether the findings derive from the data rather than the researcher's disposition, addressed by reflexivity and by traceable evidence). Two rules. **Do not mix the vocabularies**, since they rest on different assumptions about what a finding is, and a chapter claiming both external validity and transferability signals that neither concept was understood. **Do not list criteria without saying what you actually did about each**, since an unimplemented criterion is a claim, not a quality practice. In mixed designs, apply each vocabulary to the strand it belongs to and say so explicitly. Add reflexivity where the tradition requires it, written as a specific account of how the researcher's position affected access, the interaction and the interpretation, not as a paragraph of general self-description. *Correct result: one vocabulary, every criterion paired with the specific practice used, and any criterion not addressed stated as not addressed.*

**Step 10. Put limitations in this chapter, next to the choices that create them.**
The methodology chapter is where limitations belong first, because that is where they arise and where the reader meets the choice that caused them. Each limitation is stated with three parts: the limitation, the reason the design was chosen anyway, and the specific consequence for the findings. "Small sample" is not a limitation, it is a fact; "the sample of 34 gives limited power to detect small effects, so a null result on the secondary hypothesis cannot be read as evidence of no effect" is a limitation. Where a limitation was mitigated, say how and how far. The final chapter then revisits limitations in the light of what was actually found, which is a different and shorter discussion (see **15.11 Discussion and Contribution Development**). *Correct result: a limitations subsection in which every entry names its design cause and its consequence for interpretation.*

**Step 11. Report the ethical procedure, distinct from the ethics application.**
The chapter reports what was done: the approving body and reference, the informed consent process and how it was documented, voluntariness and the withdrawal arrangement, confidentiality and anonymisation practices, data storage and retention, the handling of any distress or disclosure, incentives if any, and any ethical issue that arose during fieldwork and how it was handled. The last of these is the one that distinguishes a real ethics section from a boilerplate one. Preparing the submission itself is **15.09 Ethics Clearance Application**; this section reports its outcome and execution. *Correct result: an ethics section that reports procedure and any live issues, not a restatement of general principles.*

**Step 12. Run the coherence audit across the whole chapter.**
Read the chapter as a chain and check every link: research questions → philosophical position → approach → strategy → population and sample → instrument and constructs → procedure → analysis → quality criteria → limitations. Look specifically for the standard incoherences: a positivist position with an interpretive analysis, an inductive approach with pre-specified hypotheses, constructs in the framework with no corresponding instrument component, analyses that cannot answer the research questions, a quality section from the other tradition, a sample that cannot support the claims made in the discussion, and a procedure that does not match the ethics approval. Each of these is individually easy to fix now and individually fatal in a viva. *Correct result: a completed chain check, with any break either fixed or explicitly acknowledged as a limitation.*

## 8. Analytical framework

**The justification chain.** Every element of the chapter occupies a position in one chain, and its justification is always relative to the element above it:

    Research question → Philosophical position → Approach →
    Strategy → Time horizon → Population and sampling →
    Instrument → Procedure → Analysis → Quality criteria → Limitations

Applying it: read downward and each element must be *required* by the one above; read upward and each element must be *sufficient* for the one above. Downward failures produce arbitrariness (a strategy the question does not call for). Upward failures produce inadequacy (an analysis that cannot answer the question). The chapter is coherent when both readings hold at every link, and the audit in Step 12 is this chain traversed twice.

**The three-part justification unit.** Every design decision in the chapter is written as the same small structure:

    Decision → Reason tied to the research question →
    Alternative considered and why it was set aside →
    Consequence for what the study can claim

Applying it: the first part alone is description, which is the criticism the chapter exists to avoid. The first two are the minimum acceptable. All four is what a strong chapter does throughout, and the fourth part is where limitations come from, which is why they belong here rather than only at the end.

## 9. Output format

**1. Introduction.** The chapter's structure, and the research questions restated compactly.

**2. Philosophical position.** Proportionate to the discipline, with stated consequences for the design.

**3. Research approach and strategy.** Approach, strategy, time horizon, each with reason and rejected alternative.

**4. Population and sampling.**

| Element | This study | Gap from the level above | Consequence for claims |
|---|---|---|---|
| Target population | | | |
| Accessible population | | | |
| Sampling frame | | | |
| Achieved sample | | | |

Plus the sampling method named accurately, and the sample size with its logic.

**5. Instrument.** Adopted, adapted or built, with the justification appropriate to each, validation evidence, and a construct-to-item mapping table. Full instrument in appendix.

**6. Data collection procedure.** Reproducible detail, with dates, numbers and deviations from plan.

**7. Data analysis.** Specified steps, plus:

| Research question | Data source | Analysis method | Planned or exploratory | Output it produces |
|---|---|---|---|---|

**8. Quality criteria.** One vocabulary, each criterion paired with the specific practice used.

**9. Limitations of the design.** Each with its cause and its consequence for interpretation.

**10. Ethical procedure.** Approval reference, consent, confidentiality, data handling, and any issue that arose.

**When the design is weaker than the student would like, the format must not force overclaiming (K4 §1).** A convenience sample is reported as one, with the claim limits stated, not as "purposive sampling of information-rich cases" unless that is genuinely what was done. An instrument with no validation evidence is reported as unvalidated. A response rate that cannot be calculated because the distribution route was open is reported as incalculable, not estimated. Every one of these is a small honest sentence that costs a fraction of what the same issue costs when an examiner finds it.

## 10. Quality checks

Run before the chapter goes to a supervisor. These sit on top of K4 §8.

1. Does every design statement have a reason within two sentences of it?
2. Is at least one rejected alternative named for each major choice?
3. Does every philosophical claim have a stated consequence for the design, and would the design differ under the opposite position?
4. Are target population, accessible population, frame and sample distinguished, with the gaps stated?
5. Is the sampling method named by its accurate name, including "convenience" where that is what it was?
6. Does the sample size have a stated logic rather than a bare number, and is any saturation claim supported by a description of how it was judged?
7. Is the instrument's provenance stated, with validation evidence where it exists and its absence stated where it does not?
8. Does every construct in the framework map to a component of the instrument, and does every instrument component map to a construct?
9. Could a competent stranger repeat the procedure from what is written?
10. Are deviations from the plan reported?
11. Is the analysis specified step by step, and does every research question have an analysis that can answer it?
12. Are planned analyses distinguished from exploratory ones, in advance?
13. Is exactly one quality vocabulary used, with each criterion paired with an actual practice?
14. Does every limitation name its design cause and its consequence for interpretation?
15. Does the procedure described match the procedure the ethics committee approved?
16. Does the chain hold from research question through to limitations, in both directions?

## 11. Common failure modes

| Failure | How to recognise it | How to prevent it |
|---|---|---|
| **Description without justification** | Every paragraph says what, none says why | Rewrite headings as justification questions (Step 2) |
| **Paradigm padding** | Three pages of philosophy, no design consequence | Apply the consequence test to every sentence (Step 3) |
| **Paradigm absence** | No positioning in a discipline that requires it | Check local expectations first (Step 1) |
| **Sample laundering** | A convenience sample described as purposive or representative | Name it accurately and derive the claim limits (Step 5) |
| **The bare number** | "A sample of 15 was used" with no logic | State the basis, including practical limits, as what it is (Step 5) |
| **Unjustified adoption** | An instrument used because it is well known | Adoption requires as much argument as construction (Step 6) |
| **Construct-instrument mismatch** | Framework constructs with no items behind them | Build the construct-to-item mapping table (Step 6) |
| **The unrepeatable procedure** | No dates, no numbers, no route, no deviations | Write for a stranger repeating it (Step 7) |
| **Analysis by label** | "Thematic analysis will be used", and nothing further | Specify the steps and map to questions (Step 8) |
| **Mixed quality vocabularies** | External validity and transferability claimed together | Choose one tradition and use it exclusively (Step 9) |
| **Unimplemented criteria** | Trustworthiness criteria listed with no practice attached | Pair every criterion with what was actually done (Step 9) |
| **Late limitations** | Limitations appear only in chapter five | Put them where the choice is made (Step 10) |
| **Generic limitations** | "Small sample, more research needed" | Name the cause and the consequence for interpretation (Step 10) |
| **Chain breaks** | An analysis that cannot answer the question it sits under | Run the coherence audit (Step 12) |
| **AI-invented procedure** | A plausible fieldwork account nobody performed | Only what was actually done enters the chapter (§12) |
| **AI-invented validation evidence** | A reliability figure attributed to a prior study | Report only figures read in a retrieved source (§12) |

## 12. AI guardrails

Skill-specific. The universal prohibitions in K4 apply in full and are not repeated. The human in the loop for every K5 marker in this skill is the candidate together with their supervisor.

1. **Academic integrity is a condition of use, not a footnote.** These skills assist a researcher's thinking, structure and rigour. They do not produce work to be submitted as the student's own unaided output. The user must comply with their institution's AI use policy and its declaration requirements, which vary by institution and by assessment. Where an institution prohibits AI assistance for a task, this skill must not be used for it. The skill never writes a passage for submission as though the student wrote it; it interrogates, structures, critiques and teaches. Operationally in this skill: it will supply the subsection structure, the justification questions each section must answer, the population and analysis mapping tables, the coherence audit and a critique of the student's draft. It will not write the chapter's justifications, because a justification the student cannot reproduce in a viva is worse than no justification.

2. **Never describe a procedure that was not carried out** (K4 §2.5). Not a recruitment route, not a pilot, not a member-checking session, not a field period, not a set of dates. Where a section has no content because the activity did not happen, the correct output is the empty structure with a note of what is required, not a plausible account.

3. **Never supply a sample size, a response rate, a completion figure or a reliability coefficient.** These come from the student's own data or from a source they have retrieved. A figure supplied here will be carried into an abstract and defended in a viva by someone who does not know where it came from.

4. **Never report validation evidence for an instrument from recall.** Reliability and validity figures are properties of specific studies on specific populations, and a plausible figure is worse than none. Name what evidence needs to be found and where to look for it.

5. **Never assign a philosophical position to a study on the student's behalf without their reasoning.** Position statements are examined by asking the candidate what they mean. Offer the positions, the questions that distinguish them, and the design consequences of each, and let the student choose and explain.

6. **Never launder a sampling method.** If the description the student gives is a convenience sample, it is a convenience sample in the chapter, whatever it is called elsewhere. Renaming it is misrepresentation of methodology, and it is the kind of thing an examiner tests by asking how participants were found.

7. **Never claim saturation, representativeness or generalisability on the student's behalf.** Each is a substantive empirical claim requiring evidence. Supply the tests each would have to pass.

8. **Never mix quality vocabularies to make a section look fuller.** Adding transferability to a validity section, or reliability to a trustworthiness section, produces a chapter that fails on understanding rather than on execution.

9. **Where the design cannot answer the research question, say so plainly and early.** The helpful response to an unanswerable design is not a better-written chapter. It is naming the mismatch while there is still time to change the design or the question. `RESEARCHER DECISION REQUIRED` (K5 §2.7).

10. **Never present a post hoc analysis as pre-planned.** The distinction between planned and exploratory analysis is a matter of research integrity, and the methodology chapter is where it is fixed. If the analysis was decided after seeing the data, it is exploratory, and it is reported as exploratory.

## 13. Best-practice principles

- **The rejected alternative is the load-bearing sentence.** A chapter that never says what was not done reads as though nothing was chosen.
- **Local convention beats general guidance, every time.** Two recent dissertations from the same department tell you more than any methods textbook about what this chapter has to look like.
- **Philosophy earns its place by consequence.** If the design would be identical under a different epistemology, the epistemology section is decoration in either direction.
- **Name the sample honestly and then bound the claims accordingly.** The penalty for a convenience sample is a narrower claim, which is survivable. The penalty for concealing one is a credibility problem across the whole thesis.
- **Adoption of an instrument needs as much argument as construction, and usually gets less.** The question is not whether the instrument is good, it is whether it is good for this construct, this population and this question.
- **Write the procedure while it is happening.** Post-hoc procedure sections lose exactly the details that make them reproducible, and they lose deviations first.
- **A specified analysis is a promise the results chapter has to keep.** Which is precisely why specifying it in advance produces an honest results chapter.
- **One quality vocabulary, held consistently, signals methodological literacy more clearly than any other single feature of the chapter.**
- **Limitations written at the point of choice read as design honesty; the same limitations written at the end read as excuses.**
- **Deviations reported are strength; deviations concealed are risk.** Every real study deviates, and examiners know it.
- **The chapter should be revisable into the methods section of a paper.** If it cannot be compressed to two pages without losing anything essential, some of it was padding.

## 14. Worked example

Generic fictional scenario, academic, in a healthcare service context.

**INPUT**

A masters candidate in health services management is studying how ward-level nursing staff experience a newly introduced electronic handover system. Three research questions, all experiential. The plan is semi-structured interviews with nurses in one hospital group, recruited through the candidate's professional contacts, because the candidate works in the sector. The draft chapter runs six pages, describes each step, and was returned by the supervisor with "this describes what you will do, it does not justify it".

**PROCESS**

*Step 1.* Two recent dissertations from the department show a two-paragraph philosophy section, a required subsection order, and consistent use of the trustworthiness vocabulary. The chapter is expected to run around 4,000 words. This immediately settles the largest open question, which was whether to write six pages of epistemology.

*Step 2.* Headings are rewritten as questions. "Sampling" becomes "why these nurses, why this many, and who does this exclude". The exercise exposes that three subsections have no answerable question behind them and are cut.

*Step 3.* The positioning section states an interpretive position in two paragraphs and connects it to two decisions: that the study seeks participants' accounts rather than measures of system use, and that the analysis will build categories from the data rather than apply a pre-specified frame. A draft sentence about the nature of social reality is deleted under the consequence test, because the design would be unchanged without it.

*Step 5.* The four population levels are written out. Target: ward nursing staff using the system. Accessible: staff in the two hospitals where the candidate has contacts. Frame: the candidate's professional network plus onward referral. Sample: 14 nurses across four wards.

*The judgement call.* The draft called this "purposive sampling of information-rich participants". It is not. Participants were reached because the candidate knew them or was referred by someone who did. The decision is made to name it as convenience sampling with a snowball element, and to say so plainly, adding a purposive component that was genuine: an active effort to include both day and night shift staff and both experienced and recently qualified nurses, described as maximum variation within a convenience frame. The consequence is then written out: findings describe these participants' experience, offer no basis for claiming typicality across the group, and carry a specific risk that people who agreed to be interviewed by a known colleague may be more positive, or more candid, than the ward population. That risk is named as a limitation with a direction, rather than as a generic caveat, and this single paragraph turns the chapter's weakest point into a demonstration of methodological awareness.

*Step 8.* "Thematic analysis" becomes a specified six-step sequence with the unit of analysis stated, the coding described as inductive at first pass with a second pass mapping to the framework's sensitising concepts, the treatment of the residual described, and a note that the supervisor reviewed a sample of coded transcripts. A table maps each research question to the interview sections and the analytical output that answers it.

*Step 9.* The trustworthiness vocabulary is used exclusively. Credibility is addressed by a described member-checking step with four participants and by an explicit search for disconfirming cases. Transferability is addressed by thick description of the ward context, with a stated refusal to claim generalisation. Dependability is addressed by a retained audit trail of coding decisions. Confirmability is addressed by a reflexivity section that is specific: the candidate is a colleague, which improved access and candour about workload but almost certainly suppressed criticism of named individuals, and that asymmetry is stated as affecting the findings on interpersonal conflict specifically. A draft sentence claiming "external validity through thick description" is removed as a vocabulary mix.

**OUTPUT**

A chapter of roughly 4,200 words in which every choice carries a reason and a rejected alternative, the sample is named accurately and bounded, the analysis is specified to the step, one quality vocabulary is held throughout with each criterion attached to an actual practice, and limitations sit next to the choices that create them. The convenience sample is now a stated feature of the design with known consequences rather than a vulnerability.

`RESEARCHER SIGN-OFF REQUIRED` on the chapter before submission, and `RESEARCHER REVIEW RECOMMENDED` on the reflexivity section, since the judgement about how the candidate's dual role as colleague and researcher affected what participants would say is one only the candidate can make (K5 §2.2, §2.8).

## 15. Advanced usage

**Writing the chapter as the design happens.** Draft each subsection at the moment the decision is made, while the alternatives are still live and the reasons are still remembered. Chapters written this way contain the rejected alternatives naturally, because they were genuinely under consideration; chapters written afterwards have to reconstruct them, and it shows.

**The pre-registration discipline, informally applied.** Even where formal pre-registration is not conventional in the discipline, writing the analysis section in full before data collection and lodging it with the supervisor produces a dated record of what was planned. This makes the planned-versus-exploratory distinction in the results chapter defensible rather than assertable, and it takes an afternoon.

**Mixed methods integration written explicitly.** Where a study is mixed, the chapter must state at which point the strands meet: at design (one strand shapes the other), at collection, at analysis, or only at interpretation. It must also state what happens if the strands disagree, decided in advance, since deciding afterwards invites the reader to suspect the choice was made to suit the result. See **07.06 Qualitative and Quantitative Integration**.

**Turning the chapter into a methods section.** For publication, the chapter compresses to perhaps a tenth of its length. Doing the compression early is a useful test: what survives is the load-bearing content, and what does not survive should be examined for whether it was serving the argument or the word count. See **15.24 Journal Article Development**.

**Where the standard approach does not fit.** Practice-based, design-science and action research designs do not decompose neatly into these layers, and forcing them into a survey-shaped chapter distorts them. The requirement that survives is the justification chain: whatever the structure, each choice must be traceable to the research question, and each claim must be bounded by what the design can support. Say what structure you are using and why, and hold the chain.

## 16. Skill chain

**Recommended previous skills:**
- **01.04 Research Method Selection.** Hands over the chosen method and the comparison against alternatives that becomes the chapter's justification.
- **15.07 Theoretical and Conceptual Framework Development.** Hands over the constructs, the operationalisation table and the traceability matrix the instrument section is built on.
- **01.06 Sampling Strategy.** Hands over the sampling decision, its logic and its consequences for inference.
- **02.01 Survey Questionnaire Design** or **02.03 Interview Question Development.** Hand over the instrument the chapter justifies.

**Recommended next skills:**
- **15.09 Ethics Clearance Application.** Takes the procedure and turns it into an ethics submission, which must match it.
- **15.10 Data Analysis Chapter Development.** Takes the specified analysis plan and executes and reports it.
- **15.11 Discussion and Contribution Development.** Takes the limitations and the claim bounds established here.
- **15.12 Thesis Defence and Viva Preparation.** Reuses the rejected alternatives directly as viva answers.

**Runs well alongside:**
- **01.07 Analysis Plan Development**, for specifying the analysis before data exist.
- **13.05 Research Ethics and Consent Design**, for the substantive ethics work behind Step 11.
- **15.18 Thesis Architecture and Chapter Coherence**, which runs the coherence audit across the whole document rather than one chapter.

---
A Yazi Supplied Skill and resource.