Step 4: Deciding Your Research Question
A research question is not a topic with a question mark on the end. This is the step where something you are interested in becomes a single sentence a method can be pointed at.
Methods make no sense until the question is settled
The source is unusually direct about the ordering, and it repeats the point in two separate places. It is worth taking literally.
Both statements point the same way. The question is the input to every method decision, so a method chosen first has nothing to be judged against. It cannot be right or wrong, only familiar or unfamiliar.
This matters more for a working project manager than for a full-time researcher, because you arrive with methods already in hand. You know how to run a survey, pull a register, convene a workshop. The temptation is to reach for the instrument you can execute and then find a question it happens to answer.
The cost of getting this wrong is not detected until analysis, when the data turns out to be evidence for something adjacent to what you wanted to know. The decision that waits on this sentence is set out in Choosing a Research Methodology.
The nine questions, and what each one settles
The source gives nine questions to ask once the topic and context are decided. Each one closes something down. Answer them in writing — the answers become the first draft of your methodology section.
THE NINE QUESTIONS THE SOURCE ASKS BEFORE THE QUESTION IS FIXED
| The source's question | What it settles | The answer that means you are not ready |
|---|---|---|
| What do I want to know? | The unknown at the centre of the project, stated as an absence of knowledge rather than a subject area | A noun phrase. "Stakeholder engagement" is a territory, not something you want to know. |
| What are the specific questions I need to ask? | The sub-questions that together answer the main one | A list that does not add up to the main question, or one that answers something wider than it |
| What do I need to find out to answer the questions? | The facts or accounts that would constitute an answer | "Everything I can get." If any finding would do, no finding will settle anything. |
| Where can I find it? | The location of the evidence — people, records, documents, systems | A source you have never actually looked at, assumed to hold what you need |
| What sort of information do I need — anecdotal, historical, multiple sources? | The character of the evidence, and therefore the kind of claim you will be able to make | No preference. The character of the evidence has to be chosen, not discovered afterwards. |
| Will I use primary and/or secondary sources of data? | Whether you are generating evidence or working with evidence others generated for other purposes | "Both", with no split. See Primary and Secondary Data Sources. |
| How much data or information is enough? | Your stopping rule, and much of your schedule | Silence. An unanswerable quantity question usually means the question itself is not tight enough. |
| What authority do I need to collect this data or information? | Whether you can lawfully and legitimately obtain what the design assumes | "I can already see it." Access as an employee is not authority to use it as a researcher. |
| How current does this information need to be? | The time boundary of the study, and whether existing data can serve | No boundary, which quietly commits you to whatever happens to be available |
Column one is the source's list of nine, reproduced in full. Columns two and three are synthesis added here; the source poses the questions without commentary.
Three questions about knowledge, six about evidence
The nine split cleanly, and the split is worth making because the two halves fail differently. The grouping is added here; the source presents the nine as a single list.
Questions 1-3: what you want to know
- What do I want to know?
- What are the specific questions I need to ask?
- What do I need to find out to answer the questions?
- Failure mode: a question so broad that no result could be wrong
Questions 4-9: whether you can get it
- Where can I find it?
- What sort of information do I need?
- Primary and/or secondary sources?
- How much is enough? What authority do I need? How current?
- Failure mode: a question that is perfectly formed and cannot be answered by you, this year, with this access
The second group is a feasibility screen wearing a different hat. The source raises feasibility as a separate consideration — how long the research will take, ethical constraints, the cooperation you can obtain, and cost — and warns that failing to consider any of them can mean disaster later. Questions 4 to 9 are where that warning bites at the level of a specific sentence; see Assessing Research Topic Feasibility.
The authority question deserves separate attention. It is the one most often answered by assumption, and the one most likely to stop a project after work has begun. Confidential or personal information, in particular, triggers obligations the design has to absorb before collection starts; see Step 6: Seeking Ethics Clearance.
The one-sentence rule
The instruction is to state the research question in one sentence. The value of it is diagnostic. Almost every defect in a research question shows up as a difficulty in compressing it, and the difficulty tells you which defect you have.
Reading the compression failure
The last of those is the most common, and the fix is not in this step. Concepts carry a great deal of meaning, and a question built from unpinned concepts cannot be tested; see The Conceptual and Theoretical Framework.
From question to research statement
The source does not stop at the question. It adds one further instruction, and that instruction sends you backwards before it sends you forwards.
- Topic and context
- Focused research question
- Further reading
- Research statement
The loop back into the literature is deliberate and consistent with the source's view that the process is ever-changing rather than a fixed set of steps. A question that has just been focused is exactly the point at which reading becomes efficient, because you finally know what you are looking for. The reading routine is set out in Step 3: Checking the Literature.
A worked interrogation
Starting point: an interest in why the handover from construction to operations goes badly. That is a territory, not a question. Running the nine questions over it does two things — it narrows the territory, and it exposes which parts of the intended study were never going to be possible.
THE NINE QUESTIONS RUN OVER ONE TOPIC
| Question | Answer for this study | What the answer forced |
|---|---|---|
| What do I want to know? | Whether operations readiness activities that start before practical completion change the number of defects carried into the first year of operation | Narrowed from "why handover goes badly" to one activity and one outcome |
| What are the specific questions I need to ask? | When readiness activity started; who was involved; what defects were recorded and when | Three sub-questions, each answerable from a different source |
| What do I need to find out to answer the questions? | Start dates for readiness activity, and defect counts by period for the same assets | Established that two datasets have to be joinable at asset level |
| Where can I find it? | Project records for the dates, the maintenance system for the defects | Revealed that the two systems do not use a common asset identifier |
| What sort of information do I need? | Historical records, plus accounts from participants to explain the pattern | Committed the study to more than one type of method |
| Primary or secondary? | Secondary for both datasets; primary for the explanatory interviews | Made the provenance of the records a limitation to be written up, not a footnote |
| How much is enough? | Enough completed handovers to see a pattern, and interviews continued until accounts stop changing | Turned an open-ended plan into a stopping rule |
| What authority do I need? | Approval to use maintenance records that name individual technicians, and to interview staff | Identified personal information in what looked like a purely technical dataset |
| How current? | Handovers completed within a period long enough to have a full year of operating data | Set the sampling window, and excluded the most recent projects |
Illustrative. The pattern of the interrogation is the point, not the specific answers.
The sentence that survives: does operations readiness activity commenced before practical completion reduce defects recorded in the first year of operation? One sentence, two nameable concepts, a stated population and a time boundary.
Note what the interrogation cost. A missing common identifier between two systems is the kind of thing discovered three months into collection if it is not asked about at question stage. So is the presence of personal information in a maintenance dataset.
Before you move to design
Two checks. The first is a record you keep; the second takes about a minute and is better done with another person.
The question record
- The question exists as one sentence, dated, with earlier versions retained
- All nine of the source's questions have written answers, however short
- Every concept in the sentence has a stated definition
- The population and the time boundary are visible in the sentence or immediately beneath it
- You can name what the answer would look like, including an answer you would not like
- The literature gap the question addresses is identified and can be pointed to
- The authority to obtain the evidence has been confirmed rather than assumed
- The aim or objective is written down separately, even though the source leaves that step orphaned
With the question fixed, the method decision finally has something to be judged against; see Step 5: Designing Your Research.
The second-pass review
The nine questions above interrogate a question you are forming. Later material supplies a second pass, run when you are finalising rather than exploring, and it asks different things — twelve questions split between what the literature says and whether the data is actually reachable.
The sharpest of them is about access rather than interest: where is that data? Is it in records and reports? Is it in the testimony of people you would reach through interviews or focus groups? Do you have access? A question that survives every other test and fails that one is not answerable by you.
What to carry forward
- Methods make no sense until the question is decided. Resist choosing the instrument you already know how to run.
- Answer the source's nine questions in writing; they are the first draft of your methodology section.
- Three of the nine are about knowledge and six are about evidence supply. The second group is where most projects fail.
- Compress to one sentence, and read the difficulty of compressing as a diagnosis rather than a nuisance.
- The source never defines the research statement it tells you to formulate, and orphans the aim-or-objective step. Write both, and take the format from elsewhere.
- Version and date the question. Evolution is expected; silent drift is not.
Frequently asked questions
Why does the source insist the question comes before the methods?
Because methods are only as good as the answers they give to research questions, so a method chosen before the question has nothing to be evaluated against. The source states plainly that deciding on research methods makes no sense until you have decided on your research questions. Choosing the instrument first tends to produce a study that is executable but answers a question nobody needed answered.
Does the research question really have to be one sentence?
The source's instruction is to state it in one sentence. Treat it as a test rather than a formatting rule: a question that will not compress is usually two projects, an unoperationalised abstraction, or a set of concepts you have not yet defined. You can add sub-questions beneath it, and the source expects you to.
What is a research statement, and how is it different from the research question?
The source instructs you to formulate a research statement after you have focused the question and read more widely, but it never defines the term or distinguishes it from the question. That is a gap in the supplied material. Take the definition and the format from the text your programme follows, and say which one you used.
Can my research question change once I have started?
Yes, and the source expects it to: research questions evolve over time as you gather information and evidence to refine your focus. What matters is that the change is recorded. Keep a dated version history with one line on what forced each revision, so the final question is visibly connected to the literature gap you set out to address.
How much data is enough?
The source poses the question and does not answer it, and it gives no sample size guidance anywhere in the supplied material. Answer it yourself at question stage, either as a number with a justification or as an explicit stopping rule. If you cannot answer it at all, the question is usually still too broad to design against.
What does the source mean by asking what authority I need to collect the data?
It is one of the nine questions and it asks whether you are permitted to obtain and use the information your design assumes. Being able to see data in the course of your job is not the same as being authorised to use it as research material, particularly where the data is confidential or identifies individuals.
References and source attribution
- 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 — cited in the supplied source on topic selection.
- Trochim, W. M. K. 2006, Research Methods Knowledge Base — the excerpt reproduced in the supplied source on research ideas and the feasibility of a study.
- Bryman, A. 2016, Social Research Methods, 5th ed., Oxford University Press, Oxford.
- The supplied teaching source: consolidated weekly teaching notes and slide material on developing a research topic, used here as the basis for all statements attributed to "the source".
Suggested questions for Ask KEVOS
- Run the source's nine questions over my topic and show me where the answers are missing.
- Compress my research question into one sentence and tell me what the compression exposed.
- My question has two unknowns in it — help me decide which one to keep.
- What stopping rule should I write for "how much data is enough" on an interview study?
- Draft the aim and objective statements that sit under my research question.