---
name: academic-research-topic-selection
description: >
  Turns an area of interest into a researchable, feasible, assessable topic for an
  undergraduate or honours research project. Use for "help me choose a dissertation
  topic", "what should I write my final year project on", "my topic is too broad",
  "narrow my research topic", "is this topic feasible in the time I have",
  "has this already been done", "is my topic original enough", "I have three ideas
  and cannot decide", "my supervisor says my topic is too big".
category: 15 Academic University Research
ref: "15.01"
tier: 3
inherits: [K2, K3, K4, K5]
---

# Academic Research Topic Selection

## 1. One-line description

A structured method for moving from an area of interest to a single researchable question that can be answered honestly within the words, months, access, skills and ethics timeline actually available, and that will be marked against the criteria the institution is actually using.

## 2. What this skill is used for

**The research problem it solves.** Most weak undergraduate and honours projects are not badly executed. They are competently executed on a topic that could never have gone well. The topic is chosen at the moment of least knowledge and greatest anxiety, quickly, because a form is due, and it silently fixes the ceiling on the mark before a single source has been read. Four failures do most of the damage: a topic too broad to answer in the words available, so the project describes rather than argues; a topic that is interesting but not researchable, because no evidence could settle it in the time; a topic whose data or participants were never obtainable; and paralysis about originality, in which a student searches for months for an idea nobody has had, against a standard that does not apply to their degree. This skill installs a narrowing procedure, a feasibility screen with fatal-flaw logic, an honest account of originality at this level, and a commitment point.

**Where it sits in the research lifecycle.** First. Before the search strategy, before the proposal, before any reading that is not a targeted check. Everything downstream inherits the topic decision, and the cost of changing it rises steeply with every week that passes.

**Typical use cases.**
- A final-year or honours student with an area of interest and no question.
- A student whose supervisor has said the topic is too broad, without saying how to narrow it.
- A student holding three ideas who cannot choose and is losing weeks to the choice.
- Testing whether a topic that sounds exciting can actually be researched with the access available.
- Checking whether a question has already been answered, before committing to it.
- Re-scoping a topic that has failed at the proposal stage or lost its data access.

**Who uses it.** Undergraduate and honours students beginning a research project or dissertation, often with no prior experience of independent research; supervisors running early topic conversations at scale; academic skills and writing centre staff. Assume the user does not yet know the vocabulary of research design and should not be required to.

## 3. When to use it

- You have an area of interest and no question, or a question so broad that any answer would be a summary rather than a finding.
- A topic approval form, registration or supervisor meeting is coming and you need a defensible topic rather than a good intention.
- You have been told the topic is too big, too vague, or "not a research question", and need a procedure rather than another opinion.
- You are choosing between several ideas and the choice is costing you time you need for the research itself.
- You suspect the data, participants or documents your idea needs may not be obtainable, and you want to find that out now rather than in month four.
- You are worried the topic has already been done, and need to check without conducting a full literature review first.
- You are stuck on originality and cannot start because nothing feels new enough.
- A topic has failed at proposal stage and needs re-scoping rather than replacing.

## 4. When NOT to use it

- **The topic is already fixed by the institution, the supervisor or a funded project.** Where you have been allocated a question, or the project sits inside an existing study with its own protocol, topic selection is not the task. Take the given question, clarify its boundaries, and move to **15.02 Academic Literature Search Strategy** and **15.04 Research Proposal Writing**. Attempting to re-open a settled topic wastes the weeks you have.
- **The real problem is the search, not the topic.** If you have a defensible question and cannot find literature on it, that is usually a vocabulary and database problem rather than a topic problem, and re-selecting the topic will not fix it. Go to **15.02**. Change the topic only after a competent search has failed.
- **The work is a taught assignment with a set brief.** An essay, a case analysis or a report answering a question the module has supplied is not a research project. It needs argument and structure, which is **15.05 Academic Writing Structure and Argumentation**, not topic selection.
- **The project is a systematic review with a registered protocol.** Question formulation there is governed by the protocol's own conventions and is tightly coupled to eligibility criteria and the search. Use **15.06 Systematic Literature Review**.
- **The decision is really about supervision, funding or personal circumstances.** A topic cannot be selected around an unresolved question of whether you will have a supervisor in that area, whether you can afford the fieldwork, or whether you will be in the country during data collection. Resolve those first; they are constraints, not preferences, and a topic chosen against unknown constraints will be re-chosen later. `RESEARCHER DECISION REQUIRED` (K5 §2.7).
- **The topic requires access to people or records that you have no route to, and no alternative population exists.** Do not proceed on the hope that access will appear. Reformulate the question around a population you can reach, or change the topic. Optimism about access is the single most expensive error at this stage.
- **The task is a commercial or applied scoping exercise rather than an assessed academic one.** Where the output is a decision for an organisation rather than a question for examination, the criteria are decision usefulness and cost, not contribution and assessability. Use **01.02 Business Problem to Research Question** and **01.04 Research Method Selection**.
- **The user is asking for a topic to be produced for them to submit as their own choice, unexamined.** See the academic integrity guardrail in §12.1. This skill generates candidates, criteria and interrogation. The selection, the justification and the words are the student's.

## 5. Required inputs

**Required.**
- **The area of interest, in the student's own words.** Even a vague one. If there is no area at all, do not invent one: work backwards from modules the student did well in, questions that annoyed them, and any data or setting they already have access to.
- **The constraint envelope**: degree level, discipline, word count for the final artefact, submission date, and whether primary data collection is permitted. Without these, feasibility cannot be assessed and the narrowing has no target. If they are not known, stop and get them from the handbook, not from assumption.
- **The marking criteria for this assessment.** The topic will be marked against these, not against what makes an interesting question in general. If unavailable, ask; if genuinely unavailable, proceed and flag the resulting judgement as unverified (K3 §3.4).

**Optional, and what each one adds.**
- **The institution's ethics process and typical turnaround.** Converts "will need ethics approval" from a vague worry into a date, which is what actually decides whether a topic involving people is feasible.
- **The supervisor's own research area and publication record.** Determines which candidate topics come with expert supervision and which come with a supervisor learning alongside you, which is a real difference in the quality of the guidance available.
- **Any pre-existing access**: an employer, a placement, a community organisation, a dataset the department holds, a family business. This is the most underused asset in undergraduate research and it frequently determines which of several equally good topics is the one that can actually be done.
- **Examples of high-marked projects from the same programme.** Calibrates expected scope better than any abstract description of standards. Two examples tell a student more about the size of an honours question than a page of guidance.
- **The student's methodological comfort**: whether they have completed a statistics module, whether they have ever run an interview, what software they can use. Prevents choosing a design the student cannot execute, which surfaces at analysis when it is too late.
- **A date by which the topic must be fixed.** Turns Step 10 from advice into a deadline, which is the only thing that reliably ends the searching.

## 6. Questions to ask before starting

1. **What degree level, what discipline, and what does the handbook call this artefact?** Level sets the originality standard, discipline sets what counts as a legitimate question and method, and the artefact name (dissertation, capstone, research report, extended essay, thesis) usually signals expected scope. Default if unanswered: ask, and do not proceed. Every judgement in this skill is level-dependent and discipline-dependent, and guessing produces confident advice for the wrong degree.
2. **How many words, and how many weeks until submission?** These are the two hard walls. A question that needs four studies does not fit 8,000 words; a question needing ethics approval, recruitment and transcription does not fit eleven weeks. Default: ask. If truly unavailable, use the most conservative plausible figure and state it as an assumption at the point it bites.
3. **Is primary data collection permitted, expected, or excluded?** Some programmes require it, some prohibit it, some allow either. This eliminates whole classes of topic immediately. Default: assume secondary or documentary work is acceptable, state the assumption, and flag that it must be confirmed before the proposal.
4. **What does the ethics process require for this kind of project, and how long does it take?** Determines whether a topic involving people, patients, children, employees or sensitive subjects is achievable inside the timeline. Default: assume any project involving human participants requires formal approval and budget six to ten weeks for it, stated explicitly as an assumption to be replaced with the real figure.
5. **What access do you already have that most students do not?** Frequently the single best predictor of a project that finishes. Default: assume none, and prefer topics that rely on publicly available data or documents.
6. **What is your institution's AI use policy for this assessment, and what must be declared?** This governs how this skill may be used at all, and it varies by institution and by assessment. Default: assume assistance must be declared and that no generated text may be submitted, and work in the interrogative mode described in §12.1.
7. **What would make you still be interested in this in five months?** Not sentimentality. Topic abandonment mid-project is common and expensive, and boredom is the usual cause. Default: prefer, among feasible candidates, the one connected to something the student has already voluntarily read about.

## 7. Step-by-step methodology

**Step 1. Fix the constraint envelope before discussing the topic at all.**
Write down, from the handbook and the module page rather than from memory: degree level; discipline and any sub-field conventions; word count for the final artefact; submission date and therefore weeks remaining; whether primary data collection is permitted; the ethics route and its stated turnaround; whether the supervisor is allocated or chosen and by when; the marking criteria and their weightings; and the institution's AI use policy for this assessment. Name the source of each item. Where something is genuinely unknown, write "unknown" rather than a plausible number, because a guessed ethics turnaround is the most common reason a feasible-looking topic fails. *Correct result: a one-page constraint sheet, sourced item by item, with unknowns visible.* Every judgement that follows is made against this sheet, and no topic can be called feasible except relative to it.

**Step 2. Separate area, topic, problem and question, and locate where you currently are.**
These four are routinely collapsed, and the collapse is what produces a project with nothing to argue. An **area** is a field of study. A **topic** is a bounded subject inside it. A **problem** is a specific tension, inconsistency, gap or unresolved difficulty, and it is stated as a sentence containing a "but", a "yet" or a "despite". A **question** is an interrogative sentence naming a population, a phenomenon and a context, whose answer would resolve part of the problem. Most students arrive holding an area and believing they hold a question. Write your current formulation, classify it honestly, and then write the missing levels beneath it. The problem is the level students most often skip, and skipping it is why a literature review ends up as a summary: with no problem, there is nothing for the literature to be evidence about. *Correct result: a four-line ladder from area to question, with the problem stated as a sentence containing a tension.*

**Step 3. Narrow on four axes, one at a time.**
Broad topics are narrowed by specifying: **population** (who exactly, with what boundary), **phenomenon or construct** (what precisely is being examined, defined so that two people would agree on it), **context or setting** (where, in what institution, sector or medium), and **comparison or period** (against what, or when). Narrow one axis, re-read the question, and check whether it is now answerable. Narrowing all four in one move is the second commonest error after not narrowing at all, because it produces a question so specific that no existing literature speaks to it and there is nothing to build an argument from. The right endpoint is a question narrow enough to answer and broad enough that a body of work exists against which the answer means something. *Correct result: a question with all four axes visible, reached by named moves you can justify to a supervisor.*

**Step 4. Write the imagined finding, and then its opposite.**
In two or three sentences, write a result you might plausibly get, phrased the way your discipline would report it. Then write the opposite result. This test does more work than any other in this skill. If you cannot write a sentence that would count as an answer, you do not yet have a question. If the opposite finding is impossible, or would surprise nobody, the question is either a foregone conclusion or a description task rather than a research question. If both versions are interesting and either could plausibly be true, the question is live. *Correct result: two written candidate findings, both credible, and a note of what evidence would distinguish them.*

**Step 5. Run the prior-answer check, which is not a literature review.**
Three targeted probes, each time-boxed. First, search the exact question in the vocabulary of the field and read titles and abstracts only, looking for a study that answers it in your population and context. Second, find one or two recent review articles or handbook chapters in the area and read their "future research" or "limitations" sections, which are the most efficient existing statement of what remains open. Third, check whether the question is answered in a different population or setting, which is a finding in its own right and often converts an unoriginal question into a defensible one. The outcome is one of four verdicts: **unanswered**, **answered elsewhere** (transferable to your context), **answered here** (choose again), or **contested** (the strongest position for an undergraduate project, because the argument is already framed for you). *Correct result: a verdict with the specific items that support it named, and an honest note that this is a targeted check, not a systematic search.* The full search follows in **15.02**.

**Step 6. Calibrate originality to the level, explicitly.**
This step exists to remove a misunderstanding that paralyses more students than any other. Doctoral originality requires a distinct contribution to knowledge in the field. Undergraduate and honours originality does not, and no examiner expects it. What is required at this level is that the work is your own, that the question has not been trivially answered, and that you demonstrate competent, defensible research practice on a question with a real answer. Legitimate forms of contribution at this level include: a known question examined in a new population, setting, sector or country; a replication in a context where the original may not hold; new primary data on a question previously studied only with secondary data, or the reverse; an established framework applied to a case it has not been applied to; a synthesis that brings together two literatures that do not usually cite each other; a methodological variation on an existing study; and a well-executed descriptive account of something not previously documented. Choose the form deliberately and write it in one sentence. *Correct result: the contribution named as one of these forms, in a sentence, at the correct level for the degree.* Say plainly that searching for an idea nobody has ever had is the wrong search, and that it is the most common reason topic selection takes months.

**Step 7. Screen feasibility on seven dimensions, with fatal-flaw logic.**
Score each dimension green, amber or red, and do not average. One red is disqualifying, however strong the rest.

| Dimension | The question it answers | Red |
|---|---|---|
| Data access | Can you actually obtain the data, documents or records? | Access depends on a permission you do not have and cannot request |
| Sample availability | Do enough of the right people exist, reachable by you, in time? | The population is small, gatekept, or reachable only through a body that has not agreed |
| Ethics class and timeline | What approval is needed, and does it fit before fieldwork must start? | Vulnerable participants or sensitive data, with an approval timeline longer than the weeks available |
| Method skill | Can you execute the analysis this question requires? | The design needs a technique you have never used and no time to learn |
| Supervisor expertise | Is there anyone who can supervise this competently? | No supervisor in the department works anywhere near it |
| Time | Does the full sequence fit, with slack? | The plan works only if nothing goes wrong |
| Scope against word count | Can this be answered, not merely described, in the words allowed? | The question needs a chapter per axis and you have three chapters |

Cost the time dimension by sequence rather than in the abstract: ethics, recruitment, collection, transcription or cleaning, analysis, writing, and revision, each with a date. *Correct result: a scored table, any red named, and a decision to proceed, re-scope or drop.* Re-scoping a red is usually preferable to abandoning the topic, and the axes in Step 3 are the tool for it.

**Step 8. Test the topic against the criteria that will actually mark it.**
Take the marking rubric and, for each criterion, write one sentence describing how this topic gives you something to write for it. A topic that offers no route to the criterion carrying the largest weight is the wrong topic however interesting it is. The common failures are visible immediately here: a topic with no methodological choice to justify scores nothing on a methodology criterion; a purely descriptive topic scores nothing on critical analysis; a topic with no engagement with existing work scores nothing on the literature criterion. *Correct result: a criterion-by-criterion note, with any criterion the topic cannot serve flagged.*

**Step 9. Compare three candidates rather than defending one.**
Carry at least three formulations through Steps 3 to 8, not one. A single candidate invites justification rather than assessment, and students defend a first idea for weeks past the point where the feasibility screen has told them to move. Comparison also reveals that the differences that feel decisive (which topic is more exciting) are usually smaller than the differences that decide the outcome (which one has data you can get). Present the three as a table with the scored dimensions, the contribution form, and the rubric fit. *Correct result: three comparable candidates, with the reasons one is preferred stated in terms of feasibility and assessability, not enthusiasm.* `RESEARCHER DECISION REQUIRED`: the choice is the student's and their supervisor's, and it rests on knowledge of the programme, the department and the student's own circumstances that this skill does not have (K5 §2.1, §2.7).

**Step 10. Commit, and write the commitment note.**
Fix a date, choose, and write a short note recording: the chosen question, the problem it addresses, the contribution form, the feasibility reds accepted and how they are mitigated, the axes that were narrowed and why, and the two candidates rejected with their reasons. This note is the raw material for the rationale section of the proposal, it is the answer when a supervisor asks why not something else, and it is the thing that stops the topic being reopened in week six. State plainly that the topic will still change in detail, and that changing detail is normal, while changing the question entirely after the search has begun is expensive. *Correct result: a dated one-page commitment note, and the search in 15.02 beginning the same week.*

## 8. Analytical framework

Two structures do the work, and they run in sequence.

**The specification ladder**, which converts interest into a question:

    Area → Topic → Problem → Question → Sub-questions → Imagined finding

Read it downward to narrow and upward to check. Downward, each level adds a boundary. Upward, each level must justify the one above: the sub-questions must together answer the question, the question must address the problem, and the problem must be a real tension inside the topic rather than a restatement of it. The upward reading is the one students skip, and it is where misalignment is caught before it reaches the proposal, where markers reliably find it.

**The feasibility gate**, which runs on each candidate question:

    Question → Data required → Access route → Ethics class →
    Method skill needed → Time sequence → Word budget → Verdict

Applying it: the gate is conjunctive, not compensatory. Every stage must pass; strength at one does not offset failure at another. A brilliant question with no access route is not a better project than an ordinary question with a clear one, because the brilliant version does not get written. Where a stage fails, return to the narrowing axes rather than to a new topic: most failures at the gate are failures of scope, and re-scoping preserves the reading already done.

## 9. Output format

**1. Constraint sheet.** Degree level, discipline, word count, submission date, weeks remaining, primary data permitted or not, ethics route and turnaround, marking criteria and weightings, AI use policy. Each with its source. Unknowns marked as unknown.

**2. The ladder, for each candidate.** Area, topic, problem (as a sentence containing a tension), question, and two or three sub-questions.

**3. Imagined findings.** Two contrasting plausible results per candidate, and what would distinguish them.

**4. Prior-answer check.** Verdict per candidate (unanswered, answered elsewhere, answered here, contested), the specific items consulted, and an explicit statement that this is a targeted check rather than a systematic search.

**5. Contribution statement.** One sentence per candidate, naming the form of contribution, at the level of the degree.

**6. Feasibility table.**

| Candidate | Data access | Sample | Ethics and timeline | Method skill | Supervisor | Time | Scope vs words | Reds | Verdict |
|---|---|---|---|---|---|---|---|---|---|

**7. Rubric fit.** Criterion by criterion, one sentence each, with unserved criteria flagged.

**8. Comparison and recommendation.** The three candidates side by side, with a preference stated in feasibility and assessability terms, marked `RESEARCHER DECISION REQUIRED`.

**9. Commitment note.** Dated. Chosen question, rejected candidates with reasons, accepted risks and mitigations, and the first three actions with dates.

**When the inputs are thin, the format must not force fabrication (K4 §1).** If the constraint sheet has unknowns, the feasibility table shows "cannot assess" rather than a guessed score, and the recommendation is explicitly conditional on the missing items. If the prior-answer check could not be run because no sources were accessible, say so and record what was attempted; do not produce a verdict on whether the question has been answered from general impression. If only one candidate exists, output one candidate and say that the comparison was not run, rather than inventing two weak alternatives to fill the table.

## 10. Quality checks

Run before the topic is taken to a supervisor. These sit on top of K4 §8.

1. Is the question an interrogative sentence naming a population, a phenomenon and a context?
2. Could two people read the question and agree on what would count as an answer?
3. Have both an imagined finding and its opposite been written, and are both plausible?
4. Is the problem stated as a tension, rather than as the topic restated?
5. Do the sub-questions, taken together, actually answer the main question, with nothing left over and nothing missing?
6. Has every feasibility dimension been scored, and is every red either mitigated or accepted in writing?
7. Is the ethics timeline a real figure from the institution, or an assumption marked as one?
8. Has the time sequence been costed by stage with dates, rather than estimated as a total?
9. Is the contribution stated at the level of the degree, and not at doctoral level?
10. Does the topic give something to write for the highest-weighted marking criterion?
11. Was the prior-answer check actually run against real, opened sources, and is it labelled as a targeted check?
12. Are all three candidates genuinely different, or are they one topic in three phrasings?
13. Is there a commitment date, and has the searching stopped?
14. Would the student be able to explain, unaided and in two minutes, why this question and not another?

## 11. Common failure modes

| Failure | How to recognise it | How to prevent it |
|---|---|---|
| **The topic that is an area** | The question could title a textbook; any answer would be a summary | Run the ladder in Step 2 and narrow on the four axes |
| **Over-narrowing in one move** | No literature speaks to the question at all | Narrow one axis at a time and re-check (Step 3) |
| **Interesting but not researchable** | Nobody could say what evidence would settle it inside the timeline | Write the imagined finding and its opposite (Step 4) |
| **The foregone conclusion** | The opposite finding is inconceivable | Same test. A question whose answer is already known is a description task |
| **Optimistic access** | The plan says "I will contact the organisation" and nothing has been contacted | Score access red until a named person has agreed in writing |
| **Ethics discovered late** | Approval timeline first checked after the proposal is submitted | Put the ethics route on the constraint sheet in Step 1 |
| **Originality paralysis** | Months spent searching for an idea nobody has had | Calibrate originality to the level (Step 6) and name the contribution form |
| **Rubric blindness** | A topic with nothing to say for the largest marking criterion | Run the criterion-by-criterion test (Step 8) |
| **Defending the first idea** | Every objection is answered rather than scored | Carry three candidates through the screen (Step 9) |
| **The endless search** | Week eight, still reading, no question | Set the commitment date at the start and hold it (Step 10) |
| **AI-generated topic adopted unexamined** | The student cannot explain why this question rather than another | The student writes the ladder and the commitment note; see §12.1 |
| **Confident claims about the literature** | "This has not been studied" after no real search | Absence is a statement about the check performed, and must name it (K4 §2.4) |
| **A method chosen before a question** | "I want to do a survey" precedes knowing what is being asked | Method follows question; a design that fits any question fits none |

## 12. AI guardrails

Skill-specific. The universal prohibitions in K4 apply in full and are not repeated.

1. **Academic integrity.** 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: candidate questions may be generated as material to react to, but the student writes the ladder, the contribution statement and the commitment note in their own words, and must be able to defend the choice unaided. If a student cannot explain their topic without the transcript in front of them, the topic is not theirs and the selection has failed regardless of how good the question is.

2. **Never assert that a topic is original, unstudied or under-researched.** You cannot know this. The honest statement is what the targeted check in Step 5 found and where it looked, phrased as "the sources consulted did not include a study of this question in this population", never as "this has not been studied". Absence of retrieved evidence is a fact about the check, not about the literature (K4 §2.4).

3. **Never name a study, author, dataset or "existing literature" from model knowledge in support of a topic decision.** A topic recommendation resting on a half-remembered body of work is a recommendation resting on nothing. Where you believe relevant work exists, describe the kind of work and where to look for it, explicitly labelled as a search lead, and never in reference format. Full rule in **15.03 Referencing and Citation Management** §12.

4. **Never estimate an ethics approval timeline, an institutional word count, a submission requirement or a marking weighting.** These vary by institution, faculty and year, and a plausible wrong number here changes the feasibility verdict. Mark them unknown and say where the student can find them.

5. **Never assume the discipline, the country, the institution or the degree structure.** What counts as a legitimate research question, an acceptable method and an adequate contribution differs substantially between fields and systems. Ask; where you cannot ask, handle the common cases explicitly and say which one you are assuming.

6. **Do not resolve the choice between candidates on the student's behalf.** Present the comparison, name the trade-offs, and mark the decision as the student's and the supervisor's (K5 §2.1, §2.7). The knowledge that decides it, about the department, the supervisor, the student's circumstances and what the programme rewards, sits outside this analysis.

7. **Do not let enthusiasm override a red on the feasibility screen.** Where a student wants a topic that fails on access, ethics or time, the correct response is to say plainly what fails, offer the re-scoped version that passes, and not to soften the screen. A topic that cannot be completed is not a kinder answer than a topic that can (K4 §9).

8. **Where the constraint sheet has unknowns, cap the confidence of the feasibility verdict at low and say which unknown is doing the work** (K3 §3.4). A verdict resting on a guessed ethics turnaround is not a verdict.

## 13. Best-practice principles

- **The topic sets the ceiling; execution decides where under it you land.** Time spent on selection is repaid at a higher rate than time spent anywhere else in the project, and it is the only stage where the whole project can still be made easier.
- **Narrow by axis, not by instinct.** "Make it smaller" is not actionable. Population, phenomenon, context and comparison are, and they let a student justify the narrowing to a supervisor rather than merely performing it.
- **The tension is the topic.** A project without a "but" in its problem statement will produce a literature review that is a list and a discussion that is a summary. Find the disagreement, the anomaly or the untested assumption early, because everything you will later argue hangs from it.
- **Contested beats unanswered at this level.** A question where the published work disagrees hands an undergraduate a ready-made argument structure and a legitimate contribution, and it is easier to write well than a question nobody has approached.
- **Access first, interest second, among feasible candidates.** Projects fail on access far more often than on interest. Where two topics are both interesting, the one with a confirmed route to data is the better project, and it is not a compromise.
- **Cost the time in stages with dates, never as a total.** "Six months" hides the fact that ethics takes eight weeks and transcription takes four. The sequence is what reveals infeasibility, and it reveals it while there is still time to change.
- **Originality at this level is about the work being yours, not about the idea being unprecedented.** Say this early and say it plainly. It is the single most useful thing a supervisor tells a stuck student, and it is usually told too late.
- **A question you cannot imagine answering is not a question.** The imagined-finding test takes five minutes and eliminates more bad topics than any amount of reading.
- **Three candidates, always.** One candidate turns assessment into advocacy. Three make the trade-offs visible and make the decision defensible.
- **Committing is part of the method, not the end of it.** Searching feels like progress and is not. Set the date before you start, and treat passing it as a failure of process rather than a sign the topic was not good enough.
- **Expect the question to move, and say so.** Detail changes as the literature is read; that is the process working. What must not change after the search has begun is the question itself, and knowing the difference prevents both premature rigidity and endless drift.

## 14. Worked example

Generic fictional scenario, academic, social science.

**INPUT**

A final-year undergraduate on a three-year social science programme has to produce a 9,000 word research project, submitted in twenty-two weeks. Their stated area is "social media and mental health in young people". Primary data collection is permitted. The institution requires ethics approval for any project involving human participants, with a stated turnaround of six to eight weeks and a further two weeks if amendments are requested. The marking criteria weight critical engagement with literature at 30%, methodology and its justification at 25%, analysis at 25%, argument and structure at 15%, and presentation at 5%.

**PROCESS**

*Step 1.* The constraint sheet immediately produces the decisive number: with ethics taking up to ten weeks and writing needing at least six, fieldwork has a window of roughly five weeks. Any design requiring recruitment of a hard-to-reach group is already in difficulty, and this is known in week one rather than week twelve.

*Step 2.* The ladder is written. Area: social media and wellbeing. Topic: social media use and self-reported anxiety among university students. Problem, first attempt: "students use social media a lot and it may affect anxiety". This is a restatement, not a tension. Second attempt: "Cross-sectional studies consistently report an association between time spent on social media and self-reported anxiety, yet studies distinguishing active from passive use report much weaker associations for active use, and the undergraduate population is rarely separated out." That contains a "yet" and it is now a problem.

*Step 3.* Narrowing. Population: full-time undergraduates at one institution. Phenomenon: passive versus active use, defined by a published usage typology rather than by the student's own definition. Context: one platform type rather than social media in general, because pooling platforms is part of what makes the existing evidence inconsistent. Comparison: between-use-type rather than over time, because a longitudinal design does not fit twenty-two weeks.

*Step 4.* Imagined findings. "Passive use is associated with higher self-reported anxiety than active use of comparable duration." And its opposite: "Anxiety differences by use type disappear once total time is controlled." Both are plausible, and the second would be the more interesting result. The question is live.

*Step 5.* Prior-answer check. Two review articles are located and their future-research sections read. Verdict: **contested**, with active and passive use reported inconsistently, and the undergraduate population rarely isolated. This is the strongest available position and it is recorded as such, with the explicit note that this is a targeted check and not a systematic search.

*Step 6.* Contribution form named: an established distinction (active versus passive use) tested in a population and institution where it has not been separately examined, using an existing validated typology. That is an entirely legitimate honours-level contribution and it takes one sentence.

*Step 7.* Feasibility. Access is green, since students are reachable through the institution. Sample is amber: a survey needs enough responses in five weeks and recruitment is the risk. Ethics is amber: low-risk, but anxiety is a sensitive construct and the timeline is tight. Method skill is amber: one statistics module, enough for group comparisons but not a mediation model. Supervisor green, scope against words green, time amber, workable only if the ethics application goes in by week three.

*The judgement call.* The student's preferred version of the project involved interviewing students who had deleted social media accounts after a mental health difficulty. It is a better question. It is also red on two dimensions: the population is not reachable in the time, and the combination of a sensitive topic with a potentially vulnerable group pushes the ethics application into a higher-scrutiny category the timeline cannot absorb. The temptation is to keep it and hope. Instead the reds are named, and the axes from Step 3 are used to re-scope: the same intellectual interest is retained by measuring use type rather than by recruiting a vulnerable group, and the interview question is recorded in the commitment note as the study the student would do next, which is exactly what a discussion chapter needs at the end.

*Step 8.* Rubric fit. Critical engagement is strong, because the topic is contested. Methodology is strong, because the active-passive measurement decision requires genuine justification. Analysis is adequate and matched to the student's actual statistical ability. No criterion is unserved.

*Steps 9 and 10.* Three candidates were carried (this one, a documentary analysis of institutional wellbeing policy, and the interview study). The interview study fails the gate. The documentary analysis passes but scores weakly on the analysis criterion. The survey study is chosen, committed on a fixed date, and the commitment note records the rejected two with their reasons.

**OUTPUT**

A single committed question, a scored feasibility table with two amber risks and their mitigations (ethics submitted by week three, recruitment through two named channels rather than one), a contribution statement at honours level, a rubric fit note showing no unserved criterion, and a rejected-candidates record that later supplies both the rationale section of the proposal and the future-research paragraph of the discussion.

`RESEARCHER DECISION REQUIRED` on the final choice between candidates, which rests on supervisor availability and programme norms this analysis does not hold (K5 §2.1).

## 15. Advanced usage

**Reverse selection from access.** Where a student has unusual access (an employer, a placement, a community organisation, an existing departmental dataset), run the method backwards: start from what the access permits, list the questions it could answer, and screen those against interest and rubric fit. This produces feasible topics faster than forward selection and uses the student's single largest advantage rather than wasting it.

**Selection under a fixed supervisor.** Where the supervisor is allocated rather than chosen, treat their expertise as a constraint on the constraint sheet rather than a preference, and generate candidates inside their area first. A topic supervised by someone who knows the literature is worth more to an undergraduate than a marginally more interesting one supervised by someone reading it for the first time.

**Salvaging a failed topic.** When a topic collapses in week eight because access was refused or ethics declined, do not start again. Keep the problem statement and the reading, and re-run only Steps 3 and 7: change the population, the setting or the data type while holding the tension constant. A refused interview study frequently becomes a documentary or secondary-data study on the same problem, and most of the literature work survives.

**Where the standard approach does not fit.** In practice-based and creative disciplines the question may legitimately be formulated after preliminary practice, and the ladder is run retrospectively to articulate what the work is asking. Where secondary analysis of large public datasets is the norm, the feasibility screen shifts weight from access and ethics to data documentation and skill, and the prior-answer check matters more because the dataset has usually been analysed before.

## 16. Skill chain

**Recommended previous skills:**
- None within this track. This is the entry point for an undergraduate or honours research project.
- **01.02 Business Problem to Research Question** is the commercial analogue, useful where a student's project is set by an external organisation and arrives as a business need rather than an academic question.

**Recommended next skills:**
- **15.02 Academic Literature Search Strategy.** Takes the committed question and builds the documented systematic search that establishes what is known, replacing the targeted check in Step 5.
- **15.04 Research Proposal Writing.** Takes the ladder, the contribution statement, the feasibility table and the commitment note, which between them supply most of the proposal's problem statement, justification, scope and timeline.

**Runs well alongside:**
- **15.03 Referencing and Citation Management**, from the first source read in the prior-answer check, because a reference not captured at the moment of reading is usually a reference lost.
- **15.05 Academic Writing Structure and Argumentation**, which turns the problem statement into a written argument rather than a description.
- **10.01 Literature Review and Desk Research**, where the project is applied or commercially framed and the review must serve a decision as well as an examiner.

---
A Yazi Supplied Skill and resource.