Methodological Coherence: Making Every Step Fit
Coherence is the quality a reviewer tests first and names last. This page turns the source's fit requirements into a grid you can run over your own design, section by section, before anyone else does.
The two fit rules the source states
The supplied teaching source makes coherence an explicit requirement twice, in two sentences that do different work. The first governs the internal logic of the research. The second governs its relationship with the organisation it is conducted in. You need both to pass.
Note the scope words. The first rule applies to every step, the second to every aspect. The source is not describing a section of your write-up; it is describing a property of the whole design.
Rule one: internal fit
- The reference point is the purpose and focus of the research
- Applies to every step in the process as you designed it
- Failure looks like a method that answers a question you are not asking
- Tested by tracing each step back to the research question
Rule two: contextual fit
- The reference point is the philosophical framework within the organisation the project is situated in
- Applies to every aspect of the project as you develop it
- Failure looks like findings the organisation cannot recognise as evidence
- Tested by asking who will act on this and what they will accept
The second rule is the one practitioners under-read. The source states that “every research project is situated within a particular philosophical framework” and that this framework “reflects the organisational context in which it is undertaken.” For a project manager researching inside their own organisation, that is the difference between a report that changes a delivery decision and one that is filed.
The four things every step must be appropriate to
Alongside the fit rules, the source gives a four-item list. Every step in the research project should be appropriate to the nature of the study, the aim of the study, the population of the study and the context of the study.
THE FOUR APPROPRIATENESS CRITERIA AS WORKING QUESTIONS
| Criterion | The question it puts to each step | What a failure sounds like in review |
|---|---|---|
| Nature of the study | What kind of study is this — describing what is, explaining how or why, or evaluating against a benchmark? Does this step belong to that kind of study? | You said this was an evaluation, but there is no benchmark anywhere in the design. |
| Aim of the study | What is the study for, in one sentence? Does this step move the study towards that aim, or is it there because the data was available? | This whole instrument produces information you never use in your conclusions. |
| Population of the study | Who or what is being studied, and can this step actually reach them? Do you have access, authority and enough of them? | Your question is about frontline delivery, but every respondent is a manager. |
| Context of the study | Where is this being conducted, under whose sponsorship, and under what constraints of time, money and access? | The design is sound and there is no way to execute it inside your timeframe. |
The four criteria are the source's. The questions and failure signatures are a practical reading offered by this page.
The nature question connects directly to the three kinds of study the source identifies — descriptive, explanatory and evaluative. If you cannot say which one you are doing, the other three criteria have nothing to test against. That distinction is set out in Descriptive, Explanatory and Evaluative Research.
Why coherence is structural rather than stylistic
Coherence is not a writing problem. It is a consequence of how the source models a research project: as a stack in which each layer is embedded in the one below it.
- Ontology and epistemology
- Theoretical perspective
- Methodology
- Data collection methods
- Data as evidence
- Findings
Because the layers are embedded rather than merely adjacent, an inconsistency does not stay where you introduced it. Change your assumption about what counts as knowledge and every layer above it is now unsupported, whether or not you rewrite them. This is why coherence problems tend to surface late, in the results chapter, as an inability to say what the data means. The stack itself is described in The Methodological Pyramid and its individual layers in The Five Basic Elements of a Research Project.
The source attaches an obligation to the chain: “the researcher must explain and justify the methodology and methods used in the research project.” Explaining is describing what you did. Justifying is showing why it followed from what sits beneath it. Weak methodology chapters do the first and skip the second.
The coherence audit grid
Run this over your own design as a document review. For each row, find the place in your proposal where the answer is written down. If you can only answer from memory, it is not yet in the design.
THE AUDIT GRID — SEVEN ROWS, ONE PASS
| Layer | The question | Evidence you should be able to point at | Incoherence signature |
|---|---|---|---|
| Purpose and focus | Can you state the purpose of the study in one sentence without using the word and? | A single stated aim, and a research question stated in one sentence | Two aims joined by and, each pulling towards a different design |
| Ontology | What are you treating as real and knowable — measurable facts, or accounts people give? | An explicit statement of assumptions about the nature of what you are studying | The assumption is never stated, so no later choice can be justified against it |
| Epistemology | How do you hold that knowledge about it is created, and what does that make valid evidence? | A named position with a citation, and a statement of what will count as evidence | A position named in one paragraph and contradicted by the instrument three pages later |
| Methodology | Does the strategy for the study follow from that position, and does it suit the kind of question? | A named methodology, justified against both the question and the epistemological position | The methodology is justified only by convenience, precedent or supervisor preference |
| Methods | Does each data collection method produce the kind of evidence your position accepts? | Each instrument mapped to the question or sub-question it answers | An instrument that produces data you have no plan to analyse |
| Population and access | Can these methods reach the population, and do you have authority to collect from them? | A defined population, a sampling or selection approach, and confirmed access | The design assumes cooperation that has never been requested |
| Organisational context | Will the organisation that has to act on this recognise your evidence as evidence? | A stated understanding of what the sponsor treats as a credible basis for a decision | Rigorous findings in a form the decision-maker cannot use |
The last row is the one the source insists on and most templates omit. Contextual fit is not about pleasing a sponsor; it is about whether the knowledge you produce can enter the organisation at all. Findings the commissioning organisation cannot use have failed the second rule however well they satisfy the first.
Running the audit in six passes
Do not attempt this in one read. Each pass looks for a different kind of break, and mixing them is how you talk yourself past problems.
Six passes over your own design
Pass one — downward trace
Start at the research question and work down. For each layer, ask what the layer above required of it. Mark every place where a choice appears without a reason. Unreasoned choices are not necessarily wrong, but each is a place where the chain is only implied.
Pass two — upward trace
Start at your data collection instruments and work up. For each question on the instrument, name the sub-question it answers and then the aim it serves. Anything that cannot be traced to the top comes out. This pass usually shortens the instrument, which is a benefit in itself.
Pass three — the four criteria
Take each major step and put all four appropriateness questions to it: nature, aim, population, context. It is a short exercise with a high yield, and the only pass that is directly a source requirement.
Pass four — the contradiction hunt
Read only the methodology chapter, looking for sentences that cannot both be true. A stated interpretive position beside a claim of generalisable results is the classic pairing. So is a stated intention to explore alongside a hypothesis to be tested.
Pass five — the feasibility check
Ask the four practical questions the source raises about feasibility: how long the research will take, whether there are ethical constraints, whether you can achieve the cooperation you need, and how significant the costs are. A design that fails here is not coherent with its own context, however elegant it looks.
Pass six — the hostile read
Read your design as the person most likely to dislike the answer. The source's own framing supports this: it asks in whose interest the questions are being asked, and how the findings could be used or interpreted by other people.
A worked audit
A delivery manager wants to know whether a stage-gate assurance process introduced across a portfolio has improved delivery outcomes. The proposal states an interpretive position, because the manager is also interested in how project teams experience the gates. The method proposed is a structured questionnaire to all project managers, analysed by counting responses in each category before and after the process was introduced.
The upward trace breaks immediately. Counting categorical responses and comparing them across two periods is a design for estimating the size of a phenomenon, which the source associates with structured instruments that fit diverse experiences into predetermined categories. The stated interpretive position holds that people provide their own accounts and explanations. The instrument cannot produce the kind of evidence the position treats as knowledge, a mismatch worked through in Qualitative and Quantitative Research Compared.
The four criteria then locate the deeper problem. The aim as written contains two studies: whether outcomes improved, which is evaluative, and how teams experience the gates, which is descriptive and interpretive. Nothing downstream can be appropriate to both. The population criterion adds a third break — the question is about teams, but only project managers are being asked.
Two coherent designs, not one incoherent design
Note what the audit did not do. It did not find a bad method. The questionnaire is a perfectly good instrument, and counting categories is a perfectly good analysis. It found a method that had been chosen at a different level from the one the philosophical position was chosen at, which is what incoherence almost always turns out to be. Choosing the strategy deliberately, rather than inheriting it, is covered in Choosing a Research Methodology.
Failure signatures and the pre-flight check
The last one is worth expecting rather than fearing. The source describes the research process as an ever-changing process rather than a fixed set of steps, and says your research questions will evolve over time as you gather information and refine your focus. Drift is normal. Undeclared drift is what breaks coherence, because the earlier chapters keep asserting a design you have quietly stopped following.
Pre-flight before you submit or defend
- The purpose of the study is stated in one sentence, and the research question in one sentence.
- Every data collection instrument maps to a stated sub-question, with nothing orphaned in either direction.
- The methodology is justified against the research question and against the stated philosophical position, in that order.
- Each of the four appropriateness criteria has been put to each major step, and you can say where.
- Any change of direction since the proposal is written into the front matter, not just into the method.
- You can name the decision the organisation will make with your findings, and the form it needs them in.
What to carry forward
- Two rules, both from the source: every step fits the purpose and focus of the research, and every aspect fits the philosophical framework of the organisation the project sits in.
- Four criteria to apply to each step: the nature, aim, population and context of the study. The source names them and defines none of them.
- Coherence is structural because the layers are embedded, not adjacent. A change low in the stack silently invalidates everything above it.
- Audit in separate passes — downward, upward, the four criteria, contradictions, feasibility, hostile read — and never in one sitting.
- Most incoherence is a good decision made at the wrong level, not a bad decision. Declared drift is fine; undeclared drift is what a reviewer will find.
Frequently asked questions
What exactly does the source require every step to fit with?
Two things. Every step in the research process should be appropriate to, and fit with, the purpose and focus of the research; and every aspect of the project should fit with the philosophical framework within the organisation in which the project is situated. The source also lists four criteria each step must be appropriate to: the nature, aim, population and context of the study.
Does the source define the nature, aim, population and context of a study?
No. It lists the four criteria in a single sentence and does not gloss any of them, rank them, or say what to do when two conflict. Working questions for each criterion are offered on this page as a practical reading, and should be cited as such rather than as source definitions.
How is methodological coherence different from just having a good method?
A method can be technically excellent and still incoherent, because coherence is about the relationship between layers rather than the quality of any one of them. A well-built questionnaire is a good method; it becomes an incoherence when it produces evidence your stated epistemological position does not accept as knowledge.
When should I run a coherence audit?
Run the first one before the proposal is submitted, when changes are cheap. Run a second after any change of direction during fieldwork, because the source expects questions and focus to evolve and the risk is that the earlier chapters keep describing the design you abandoned. Run a third before you defend.
My design changed halfway through. Have I broken coherence?
Not necessarily. The source describes the research process as ever-changing rather than fixed, and expects questions to evolve as evidence accumulates. What breaks coherence is leaving the earlier sections asserting the original design, so update the aim, question and methodology sections to match what you actually did and say why the change was made.
References and source attribution
- Quinlan, C. 2011, Business Research Methods, 1st ed., Cengage Publishing.
- Bryman, A. 2016, Social Research Methods, 5th ed., Oxford University Press, Oxford.
- O'Leary, Z. 2017, The Essential Guide to Doing Your Research Project, 3rd ed., Sage Publications, London.
- Naoum, S. G. 2013, Dissertation Research & Writing for Construction Students, 3rd ed., Routledge.
- The supplied teaching source: Week 2 study notes and slide deck on research philosophies, and Week 3 study notes on designing your research.
Suggested questions for Ask KEVOS
- Run a coherence check on my research design: here is my question, methodology and instruments.
- What are the four things the source says every research step must be appropriate to?
- How do I justify my methodology against my stated epistemological position?
- My interpretive study uses a structured questionnaire — is that a problem?
- How do I document a change of research direction without breaking my proposal?
- What questions will a steering committee ask about my research design?