A student opens a document called “My EPQ” and discovers that choosing an interesting subject was the easy part. There are articles to read, a question to refine, decisions to record and an eventual audience to persuade. The blank page does not necessarily mean the student lacks ideas. Often the missing piece is a workable relationship between a question, evidence and a conclusion.
A useful Extended Project Qualification plan starts with a bounded question the student can investigate, identifies evidence that could change the answer, and preserves the reasoning behind important decisions. Build the report or artefact from that investigation. Use the production record to show how the work developed, and practise explaining both the conclusion and its limits in the student's own words.
This guide uses AQA's Level 3 Extended Project Qualification, code 7993, as its specific qualification example. The specification and 2026/27 coordinator guidance were checked on 21 October 2026. Other awarding organisations and international project qualifications have their own instructions. Ask the school to confirm the board, submission cycle and current documents before applying any administrative detail.
The teaching cases below are fictional. They demonstrate research decisions, not model submissions to copy or promises of a particular grade. For the wider educational principle, read Learning That Transfers into Life: a project becomes useful when the learner can explain the thinking and adapt it after the original question changes.
eduKate · Education · Qualifications, inquiry and progression
Find the next useful step
Choose the route closest to your question. Every teaching chapter stays open; the grouped index can be expanded when you need it.
ROUTE 1 · CHAPTERS 1–3
Define a workable investigation
Identify the qualification, narrow the question and diagnose the actual bottleneck.
ROUTE 2 · CHAPTERS 4–6
Build and compare evidence
Follow sources, compare quantities and record justified decisions.
ROUTE 3 · CHAPTERS 7–9
See the project develop
Work through report and artefact cases, then organise an argument.
ROUTE 4 · CHAPTERS 10–12
Prepare and transfer
Plan the work, use supervision and test the reasoning in changed cases.
ROUTE 5 · CHAPTERS 13–14
Questions and next actions
Resolve common assumptions and prepare a focused proposal.
Full chapter index · Independent learning check · Education parent guide
Full chapter index
Define a workable investigation · Chapters 1–3
Build and compare evidence · Chapters 4–6
See the project develop · Chapters 7–9
Prepare and transfer · Chapters 10–12
Questions and next actions · Chapters 13–14
CHAPTER 1 OF 14 · Define a workable investigation
1. Identify the qualification and the work being assessed
An EPQ is a formal qualification with supervised processes. A personal reading project can be valuable without meeting those processes. Conversely, completing the paperwork does not by itself establish a strong investigation. The student needs to attend to both the award requirements and the intellectual work.
AQA's current 7993 specification describes a standalone Level 3 qualification in which students choose and develop a project beyond the material of their other qualifications. It includes taught skills and supervision alongside independent work. Independence therefore means making and explaining the relevant decisions with appropriate support, rather than refusing every conversation with a teacher.
Keep the formal instructions somewhere accessible. Record the qualification code, school deadlines, approved proposal, required evidence and contact person. A deadline copied from a friend at another centre may concern a different assessment cycle. A general online checklist may omit something the school needs to authenticate the work.
The official specification PDF distinguishes a research-based report, approximately 5,000 words, from an artefact accompanied by a research-based written report of at least 1,000 words. It also requires evidence of the development process and a presentation. These are AQA instructions, not universal definitions of independent research.
Neither route is automatically easier. An artefact can involve difficult design decisions, testing and documentation. A written investigation can involve difficult source selection, comparison and synthesis. Choose the form that can reveal the intended learning, then check its feasibility with the supervisor.
At the first meeting, a useful student can say: “Here is my proposed question, the evidence I think is available and the part I cannot yet judge.” That gives the supervisor something specific to discuss. “Is this topic good?” is a reasonable beginning, but it needs to develop into a more inspectable proposal.
CHAPTER 2 OF 14 · Define a workable investigation
2. Turn an interest into an answerable question
An interest names an area. A research question identifies something uncertain within it. “Transport” is an area. “Electric buses” is a narrower topic. “Under which operating conditions would replacing a particular diesel bus service with electric buses reduce its operational emissions?” identifies a comparison, a setting and conditions that need evidence.
That question is still too ambitious for some students. It might require operational data they cannot obtain. A more manageable investigation could compare how two published studies estimate emissions for a defined kind of service. The learner would investigate assumptions and methods rather than pretend to have measured an entire transport system.
Try a sentence with four parts: what is being compared, for whom or where, over which boundary, and by which criterion. The criterion matters. The “best” transport option could mean lowest emissions, lowest cost, greatest accessibility or least disruption. One project cannot quietly switch among those meanings whenever the evidence becomes inconvenient.
Write an initial answer before reading deeply, but label it a hypothesis or starting expectation. This reveals the belief the evidence must challenge. A student who expects electric buses always to produce lower emissions can then ask which electricity supply, vehicle use and measurement boundary might complicate that expectation.
A question does not need to sound grand to be demanding. Comparing two estimates carefully can require more analysis than describing a whole industry's history. The standard is the quality of the reasoning the question makes possible, not how many countries appear in its title.
Parents can help by listening to the question and asking what an answer would look like. They should avoid replacing it with their own professional research agenda. A project that depends on an adult's contacts, technical vocabulary and judgement can leave the student managing an impressive subject they do not yet understand.
Before approval, test whether the student can explain the question without reading a memorised paragraph. If the explanation collapses when one term is questioned, teach that term or reduce the scope. A clear question gives the project somewhere to go.
CHAPTER 3 OF 14 · Define a workable investigation
3. Diagnose the bottleneck before adding more work
“I need to do more research” can hide several different problems. The student may not know what evidence would answer the question. They may understand the question but lack access to a crucial document. They may have sufficient evidence but be unable to compare it. Each problem needs a different response.
Ask the student to show one claim they currently want to make and the source supporting it. If no claim can be identified, return to the question. If the source merely discusses the broad topic, practise matching evidence to the exact claim. If the source is relevant but difficult, work on the concept or method that blocks understanding.
Consider a student who has collected fifteen articles about urban trees. She cannot begin writing because every paragraph seems important. Her problem may be selection. Give her three possible claims and ask which source supports each. An article about summer air temperature does not automatically establish winter energy savings or improved mental health.
Another student finds a strong paper but copies its abstract because the statistical method is unfamiliar. The next step is not to collect twenty more papers. The student needs to understand enough of the method to explain what the study measured and what its conclusion can support. If that is beyond the available preparation, the question may need adjustment.
A third student can discuss sources fluently but produces a disorganised draft. That suggests a writing-structure problem. Ask for a paragraph containing a claim, relevant evidence, an explanation and a limitation. Repairing this small unit can be more productive than repeatedly redrafting the introduction.
Make the diagnosis temporary and testable. “I cannot yet compare studies using different denominators” identifies a learning task. “I am bad at research” turns a task into an identity. The first description suggests an exercise and a way to check improvement.
At each review, agree one next action that would produce useful evidence: locate a missing methodology section, calculate a comparable rate, rewrite a claim or explain a difficult graph. The project advances when uncertainty becomes something the student can investigate.
| Observed difficulty | Likely question to investigate | Useful next evidence |
|---|---|---|
| Many sources, no clear direction | What claim would answer the question? | A bounded question and one relevant source. |
| Copied abstract, weak explanation | Which concept or method is unclear? | A short explanation in the student's own words. |
| Numbers disagree | Are units, denominators and boundaries comparable? | A common-basis comparison with stated assumptions. |
| A polished but disorganised draft | What reasoning should the reader follow? | An outline arranged by claims rather than reading order. |
CHAPTER 4 OF 14 · Build and compare evidence
4. Build a source trail that supports a judgement
A bibliography records what was consulted. A working source trail also records why a source matters. For each substantial source, note the question it addresses, its relevant evidence, its method, its limitations and the claim it may support. Keep the original reference and page or section location so another reader can find the material.
Authority depends partly on the claim. A government publication can establish what a policy says. It may not independently establish how well that policy worked. A company's technical document can explain the product it designed. It may leave comparisons with competitors or long-term outcomes unresolved.
Use primary sources where they answer the actual question. Research papers, official datasets, original documents and first-hand creative works can provide direct evidence. A good secondary source can help a beginner understand the field and locate primary material. The distinction concerns the source's role in this investigation, rather than a simple good-source/bad-source label.
Suppose a project examines whether a city policy reduced traffic. A council announcement establishes the policy's intended start and design. Traffic counts may show a change. To infer that the policy caused it, the student must also consider other changes, measurement locations and comparison periods. One official document cannot perform all three evidence jobs.
Searches should follow the emerging question. If one study's conclusion depends on a definition of “peak period”, investigate that definition. If two sources disagree, check whether they measured the same population and outcome. Repeating a broad search can produce more reading without resolving the reason for disagreement.
Do not use the number of references as the main quality test. Several sources can repeat the same original claim. Trace significant numbers back to their origin where possible. Five articles quoting one report do not become five independent measurements.
Leave a clear marker when a source is unavailable or a claim remains uncertain. A gap can lead to a narrower conclusion. It should not lead to an invented quotation, an unsupported statistic or a confident statement that the student cannot verify.
Two numbers can look comparable while answering different questions. Before subtracting or ranking them, identify the numerator, denominator, units, population and time period. This is a research skill a student can practise with small examples before applying it to complex material.
Imagine two fictional transport studies. Study A reports 80 units of emissions per vehicle-kilometre. Study B reports 12 units per passenger-kilometre. It would be wrong to conclude that B is much cleaner simply because twelve is smaller than eighty. The denominator changed.
If the vehicle in A carries an average of ten passengers under the stated model, 80 divided by 10 gives eight units per passenger-kilometre. On that limited common basis, A's estimate is lower than B's twelve. The conclusion remains conditional on the occupancy estimate and comparable measurement boundaries.
Now change average occupancy to five. The same vehicle figure becomes sixteen units per passenger-kilometre. The ranking reverses. This does not prove either study is unreliable. It shows why occupancy is part of the relationship rather than a small detail to hide in a footnote.
The student's explanation should distinguish conversion from discovery. Dividing by an assumed occupancy produces an illustrative model; it does not measure actual passengers. If the evidence does not establish occupancy, the student can show several plausible hypothetical values and explain which additional observation would matter.
Return to the exact project question. If it concerns emissions per service delivered, one denominator may be appropriate. If it concerns total city emissions, the amount of travel and changes in demand become relevant too. A correct calculation can still answer the wrong policy question.
For a writing check, ask the student to complete this sentence: “These figures can be compared for this purpose because…; the comparison remains limited by…”. If either half is missing, investigate before making a strong ranking. Careful comparison often makes an argument more interesting because it explains why the answer changes.
CHAPTER 6 OF 14 · Build and compare evidence
6. Use a decision record rather than reconstructing a success story
Research rarely develops in a straight line. A source may undermine the starting expectation. A proposed interview may be unsuitable. A design may fail a test. The useful record preserves what happened, what the student noticed and why the next decision followed.
AQA's 2026/27 coordinator guidance describes the candidate record, production log and assessment record, including proposal, planning, review and reflection stages. Use the centre's current forms and instructions. A private notebook can support thinking but does not automatically replace required assessment evidence.
A useful working entry might read: “I intended to compare total annual emissions. The two accessible studies use different city boundaries. I will first compare their methods, then limit the numerical comparison to the common component.” This shows an observed problem and an intellectual response.
“Did research for three hours” shows activity. It does not reveal what changed. Time records may help planning, but connect them with a decision or discovery. What became clearer? What assumption failed? What will be done differently?
Record substantive support honestly. If a supervisor explained a method or suggested a source category, preserve that advice in the way the centre requires. The student should remain able to explain the subsequent reasoning. Hiding help does not make a project independent; it makes its history less accurate.
Parents can ask the student what changed this week rather than asking only how many words were written. A week spent correcting an unsuitable comparison may be productive even if the draft becomes shorter. The relevant progress is a more defensible investigation.
Avoid writing an artificial story in which every early decision anticipated the final conclusion. A thoughtful revision is evidence of learning. The record should make the real development understandable without turning every minor action into a diary entry.
Maya's fictional project begins with the question, “Are reusable cups better than disposable cups?” Her first draft lists advantages and disadvantages but keeps changing the meaning of “better”. Sometimes it means waste, sometimes carbon emissions, and sometimes customer convenience.
With her supervisor, she narrows the investigation to how assumptions about reuse and washing affect a simplified emissions comparison. She will use published evidence to discuss the assumptions, while keeping her own numerical model clearly illustrative. The project will not claim to measure a particular café's actual footprint.
In her invented model, manufacturing a reusable cup produces 600 units of emissions and each use, including the defined washing process, produces four units. A disposable alternative produces twenty units per use. After n uses, the model compares 600 + 4n with 20n.
Equality occurs when 600 = 16n, so n = 37.5. Under this model, the reusable option is lower from the thirty-eighth whole use. That is a conditional break-even calculation, not a universal number for reusable cups.
Maya then changes the washing assumption to ten units per use. The equality becomes 600 + 10n = 20n, giving sixty uses. She realises that her conclusion depends strongly on an operational assumption. She must explain where a real estimate of washing emissions would come from and which activities it includes.
Her report now has an argument. It explains the comparison boundary, evaluates source assumptions, uses the model to reveal sensitivity and concludes within those limits. The calculation serves the investigation rather than decorating a general essay about sustainability.
For the independent check, Maya receives a model in which the recurring reusable-use burden is twenty-two units while the disposable burden remains twenty. She explains that the reusable option never breaks even within that simplified model because it starts higher and increases faster. Recognising that change is stronger evidence of understanding than memorising thirty-eight.
Daniel wants to create a board game that helps younger pupils understand fractions. His first aim is to make the game enjoyable. That is worthwhile, but it does not specify the mathematical learning or explain how the design will be evaluated.
He narrows the intended capability to recognising equivalent fractions using area representations. He identifies a possible misconception: pupils may compare the number of shaded pieces without attending to the size of the whole. His design therefore includes cards that look superficially similar but require careful comparison.
Before making a polished board, Daniel produces a paper prototype. He can explain why two quarters and one half represent the same proportion when the whole is consistent. He also creates a counterexample using different-sized wholes, so that “the same fraction” is not confused with “the same absolute amount”.
His research informs design choices. It does not need to become an unrelated essay attached to an attractive object. Each important feature should have a reason: what problem does it address, which evidence informed it and what happened when it was tried?
Any testing involving younger pupils needs appropriate school approval, supervision and consent arrangements. A student should not independently collect children's personal data or make claims about learning gains from an informal game session. The supervisor should help establish a safe, feasible evaluation before recruitment begins.
Daniel can first test the mathematical consistency himself and ask an appropriate adult to review instructions. If authorised participant testing becomes possible, observations may reveal unclear wording or unanticipated strategies. A small convenience sample would not establish that the game improves attainment for all children.
The final account explains the design boundary, successive changes and remaining limitations. Photographs or other evidence can show stages of development where appropriate. A beautiful finished object is welcome, but the educational achievement also lies in the student's explanation of the decisions behind it.
CHAPTER 9 OF 14 · See the project develop
9. Give the report an argument rather than a reading diary
A reading diary follows the order in which sources were encountered. An argument follows the question. The student may have read the most useful source last, yet it belongs near the beginning of an explanation if it defines the central comparison.
Begin a draft outline with claims rather than chapter names. “Background” does not say what the section establishes. “The two studies define a journey differently” identifies an issue that the reader must understand before interpreting the numbers. A precise working claim makes source selection easier.
Build a paragraph around one move. State the point, introduce the evidence, explain the connection and qualify the conclusion where needed. Not every paragraph must mechanically follow four identical sentences, but the reader should be able to recover that reasoning.
For example, “Study A found a smaller figure” is incomplete. Smaller per what, under which conditions, and relevant to which decision? The paragraph improves when it explains why the figure supports a limited comparison and why it cannot settle the broader question alone.
Use transitions that preserve logic. “However” should introduce a qualification or tension. “Therefore” should follow from what was established. If the student inserts these words only to sound academic, ask them to explain the relationship aloud before rewriting it.
Keep conclusions proportionate. An investigation comparing two methods may conclude that the estimated answer is sensitive to a boundary definition. That is a useful conclusion. It need not pretend to solve the entire subject or recommend a national policy.
Revision becomes more efficient when each pass has a purpose. First check the argument's order. Then check whether each substantial claim has adequate evidence. Then improve sentence clarity and references. Editing commas in a paragraph that does not belong in the argument is a poor use of a tired student's attention.
CHAPTER 10 OF 14 · Prepare and transfer
10. Plan backwards while leaving room to learn
A project timetable should include thinking, checking and revision. If the last source is scheduled to be read the night before submission, there is no realistic opportunity for it to change the argument. The plan assumes research will produce only convenient answers.
In an illustrative twelve-week plan, a student might protect early time for question refinement and access checks, a middle period for investigation, and a later period for drafting, review and presentation practice. This is a planning example, not an AQA delivery timetable or recommended total workload.
Identify dependencies. An interview cannot be analysed before it is approved and conducted. A model cannot be interpreted before its units and assumptions are understood. A report can be outlined before every source has been read, but its strongest conclusions must wait for adequate evidence.
Estimate work in visible tasks. “Finish research” is difficult to review. “Compare how three sources define the outcome and write a short account of the disagreement” has an observable result. A student can discover that it takes longer than expected and adjust the remaining scope.
Protect the other subjects too. A project is one part of the student's education. If the plan repeatedly displaces essential work elsewhere, discuss the load with the supervisor. Reducing an oversized question can preserve the quality of the investigation and the wider programme.
Use small review points before expensive commitments. A short source comparison can reveal that the question is unanswerable with available material. A paper prototype can reveal an instruction problem before hours are spent manufacturing the final artefact. Early inspection makes revision less costly.
A missed milestone is information. Ask whether the obstacle was access, an underestimated task, a concept gap or a competing demand. Match the repair to that cause. Repeating “manage your time better” does little if the student is waiting for evidence that will never become available.
CHAPTER 11 OF 14 · Prepare and transfer
11. Make supervision, integrity and presentation support learning
Supervision is most useful when the student brings evidence of an attempt. That might be an outline, a source comparison, an unsuccessful design or a question about an assumption. The supervisor can then focus on the next intellectual move rather than take over the project.
Keep authorship clear throughout the work. The student should be able to explain the vocabulary, calculations, source choices and conclusions in the submission. Polished language supplied by someone else can hide the very learning the project is intended to demonstrate.
Where digital or AI tools are considered, follow the awarding organisation's and centre's current integrity instructions. Record permitted use as required. Never submit generated references without checking the actual source, and never treat a fluent answer as evidence that a quotation or statistic exists. If permitted assistance cannot be explained honestly, stop and ask the supervisor.
Prepare the presentation by selecting the project's central reasoning. What was the question? Why was the method appropriate? Which finding changed the student's view? What remains uncertain? A short explanation of those moves can communicate more than slides filled with every paragraph of the report.
Practise responding to a genuine question. If someone asks why a different source was excluded, the student should explain the selection criterion rather than invent a convenient reason. If an answer is unknown, describe what further evidence would be needed. Recognising a limit is part of responsible communication.
Parents can be an audience without becoming co-authors. They can say which explanation was unclear and ask the student to clarify it. They should not rewrite a specialised answer until it sounds like their own professional presentation.
After practice, check the report again. A question that repeatedly produces confusion may expose an unstated assumption or a missing definition. Presentation preparation can therefore improve understanding, rather than serve only as rehearsal for appearing confident.
CHAPTER 12 OF 14 · Prepare and transfer
12. Test whether the research learning transfers
Try the following original checks without referring to the worked cases. They are learning exercises, not official EPQ assessment questions or a predictor of marks. The purpose is to see whether the student recognises the underlying research decision when its surface changes.
First, one fictional study reports 60 units per vehicle-kilometre and another reports nine per passenger-kilometre. At an assumed average occupancy of five, the first becomes twelve per passenger-kilometre. It is higher on that common basis. At occupancy ten, it becomes six. A secure explanation identifies both the changed ranking and the unmeasured occupancy assumption.
Second, an illustrative reusable item has a production burden of 300 units and a recurring burden of five per use. Its disposable comparator has a burden of fifteen per use. Equality is 300 + 5n = 15n, giving thirty uses. The reusable option becomes lower after thirty uses, so the first whole number giving a lower burden is thirty-one.
Third, a student's three websites all quote the same original survey. How many independent surveys have been found? One. The websites may interpret it differently, but repetition is not independent replication. The student should locate the survey and examine what it actually measured.
Fourth, a source contradicts the project's initial hypothesis. The appropriate response is to evaluate its relevance and method, then revise or qualify the argument if warranted. Excluding it solely because it is inconvenient weakens the investigation.
If a learner gets the arithmetic right but cannot state what a result means, practise interpretation. If they explain the principle but use the wrong denominator, practise identifying quantities before calculation. If they repeat a model answer, change a condition again.
For a later university project, the formal agreement and credit conditions become a separate matter. University Independent Study Agreements explains that next setting. EPQ research habits may help, but the school qualification does not automatically register a university course or award its credit.
CHAPTER 13 OF 14 · Questions and next actions
13. Questions students and parents often ask
Does an EPQ have to be related to a planned degree?
The project needs to fit the qualification and centre approval process. A possible degree interest can help generate questions, but a manageable investigation that develops sound reasoning is more useful than an impressive-sounding title the student cannot investigate. Check any university's actual admissions treatment separately.
Is the artefact route a way to avoid serious writing?
For AQA 7993, an artefact still requires an accompanying research-based written report and evidence of the project process. The appropriate choice concerns the work the student wants to investigate and can explain. Construction, testing and documentation may create substantial demands of their own.
How many sources should a student use?
There is no useful universal number for every question. Start with the evidence needed to investigate the claim and the quality and independence of those sources. A long bibliography cannot compensate for sources that do not support the argument or merely repeat one another.
Can the question change after the project begins?
Research can reveal that an initial question is too broad or that evidence is unavailable. Discuss changes with the supervisor and follow the centre's approval and recording instructions. Explain the reason for the revision rather than concealing it in a retrospective account.
Will completing an EPQ guarantee a lower university offer?
No. Admissions treatment depends on the university, programme, qualification combination, applicant and entry cycle. Use the current programme's official instructions. A qualification's general recognition does not establish a particular offer, and an offer condition is distinct from the student's research learning.
What if the student is working hard but the report is not improving?
Inspect one paragraph or source comparison. Identify whether the gap is understanding the question, interpreting evidence, organising an argument or expressing a relationship. Choose a short repair and test a changed example. Counting additional hours does not identify the missing connection.
CHAPTER 14 OF 14 · Questions and next actions
14. Choose one useful next research action
Bring the supervisor a one-page proposal containing the question, its boundary, two promising sources, an initial method and the main uncertainty. Do not wait for a perfect introduction. A proposal that reveals a real uncertainty gives teaching and feedback somewhere to begin.
The student can protect a small set of habits: preserve source locations, explain important choices, distinguish assumptions from observations and revise a claim when the evidence requires it. The parent can support a manageable routine and remain curious about the reasoning. The teacher can make the next learning move explicit without doing it for the learner.
Return to the official AQA specification, its full assessment specification and the 2026/27 coordinator guidance for formal requirements. These sources were reviewed on 21 October 2026. The examples and diagnostic suggestions here are original teaching tools, not official marking advice.
A good project leaves something beyond a submitted file. The learner can recognise a question worth asking, find and examine relevant evidence, explain a justified conclusion and identify what would change it. That is a capability available for the next unfamiliar task.
Continue with the decision you need
If you are comparing project-based learning within another school qualification, Cambridge AICE Diploma Subject Groups and Award Planning explains the different compulsory core and award structure. The two qualifications have distinct rules; a research habit can travel without their certificates or credits becoming interchangeable.
Contents · Previous chapter · Continue through How Education Works