Primary research is the most tempting part of an Extended Project Qualification — and the most easily wasted. Collecting your own data feels like real research in a way that reading books does not, and examiners do notice it. But a weak survey or a handful of casual interviews does not become strong evidence simply because you collected it yourself. Done properly, primary research can sharpen a project in ways library research cannot: it can test whether published claims hold for a real group of people, and it can give an artefact project its audience evidence. Done carelessly, it produces impressive-looking charts that prove nothing and can cost marks through weak method and poor ethics. This guide covers only primary research: choosing a method, sampling honestly, handling ethics, and using findings in the project. For the wider picture — formats, grading and timelines — see our complete EPQ guide.
One point of honesty first. An EPQ is the student’s own undertaking — supervised at your school or college and signed off as work you produced yourself. Nobody can legitimately collect your data or write up your findings for you. What good help can do — and what this guide tries to do — is teach you the method: designing a survey or interview study that a small-scale student project can actually defend.
Start from the question, not from the survey
The first decision is not which method to use but whether you need primary research at all. Primary research means new data you collect yourself — surveys, interviews, focus groups, observations, experiments — as distinct from primary sources, which are original evidence created by someone else that you analyse, such as historical documents or published datasets. The distinction matters, because many EPQ questions are answered perfectly well from published sources, and adding a survey to such a question adds work without adding evidence.
Ask three questions before collecting anything. Does your research question actually require new data — for example, what a specific audience thinks, wants or does, where no published study covers that group? Does your chosen method match the question — a survey for breadth across many people, an interview for depth with a few? And can you, as a sixth-form student with limited time and access, collect data of sufficient quality to support a claim, rather than data that merely decorates one? If the answer to any of these is no, a library-based project is the stronger choice, and examiners will not penalise you for it. Primary research is not automatically more sophisticated than a well-designed investigation based on existing evidence; a weak survey is weaker than a strong literature-based argument.
If primary research is justified, state its role precisely in your project plan: what claim will it support, and what would count as evidence for or against that claim? That sentence, written before you design a single question, prevents the most common failure — collecting data first and inventing a use for it afterwards.
Surveys: design questions that measure what you think they measure
A survey collects comparable answers from a number of people, which makes it the right method when you want breadth: patterns across a group, proportions, comparisons between subgroups. Its strength is also its trap: a survey’s results look authoritative — percentages, charts, neat tables — even when the questions were leading, ambiguous or unanswerable.
Write questions the way a respondent will actually read them. One idea per question, plain wording, no double negatives, no leading phrasing that steers the answer (“Don’t you agree that…?”), and no double-barrelled questions that ask two things at once (“Do you enjoy and regularly use…?”). Offer response options that cover the realistic range, including a neutral or “not applicable” option where one genuinely exists, and pilot the survey on a few people first: ask them what they thought each question meant, and rewrite any question they understood differently from your intention. A pilot with five people catches more errors than a hundred responses to a flawed questionnaire.
Keep the survey short. Response quality falls as length rises, and a twenty-question survey from a student project will often be abandoned halfway or answered carelessly at the end. Ten well-designed questions, answered thoughtfully by a defined group, beat thirty questions answered at random. And decide your analysis before you launch: if you plan to compare two groups, you need enough respondents in each group for the comparison to mean anything; if you plan open-ended questions, you need time to code and theme the answers rather than quoting a few that happen to agree with you.
Interviews and focus groups: depth, not anecdote
Interviews suit questions where depth matters more than breadth: how people understand an experience, why they made a decision, what a process felt like from the inside. A semi-structured interview — a prepared set of open questions, with freedom to follow up interesting answers — is usually the right balance for a student project: fully structured interviews sound like surveys read aloud, and fully unstructured ones wander beyond the question.
Prepare an interview guide, not a script. List the topics and the opening question for each, plus prompts you can use if an answer is thin (“Can you give me an example of that?”). Ask open questions that invite explanation rather than yes-or-no answers, and avoid leading prompts that supply the answer you want. Record the interview with permission, transcribe it accurately, and analyse the transcript systematically: code recurring themes across interviews, note disagreements between participants, and report the range of views, not only the quotations that support your argument. A single interview quoted at length is anecdote, not evidence; three or four interviews, thematically analysed, begin to be.
Focus groups follow the same principles with an added dynamic: participants respond to each other as well as to you, which can surface shared assumptions and disagreements a one-to-one interview misses. They also need firmer moderation — one dominant voice can silence a group — and the analysis must distinguish group consensus from individual views. For most EPQ projects, individual interviews are the safer, more manageable choice.
Sampling: say honestly who your data speaks for
Sampling is how you select participants, and its honesty determines how far your findings can be generalised. A sample’s value lies in representativeness: how far it reflects the characteristics of the population you want to speak about. Student projects almost always use convenience sampling — people who are easy to reach, such as classmates, friends or family — and that is acceptable, provided you say so and limit your claims accordingly. A survey of thirty classmates can support a claim about those classmates’ views; it cannot support a claim about “young people” or “students” in general.
Understand the basic alternatives so you can choose deliberately. Random sampling gives every member of a defined population an equal chance of selection and removes researcher bias, but it requires a list of the population and access to it, which students rarely have. Stratified sampling divides the population into subgroups — by year group, subject, gender — and samples from each, which improves representativeness but takes more time. Systematic sampling takes every nth person from a list. Quota sampling fills set numbers per subgroup quickly but is likely to be biased, because the researcher still chooses who fills each quota. Whatever you use, report it plainly: who was invited, how many responded, who they were, and who is missing. A methods paragraph that hides its sample is a paragraph examiners will distrust.
Size is not a magic number. There is no EPQ rule that a survey needs a particular number of respondents, and a small, well-defined sample analysed honestly outscores a large, undefined one reported as if it were representative. What matters is that the sample fits the claim: the narrower the claim, the smaller the sample it can rest on.
Ethics: agree them before you collect anything
Ethics must be addressed before data collection begins, in discussion with your supervisor, and recorded in your production log. The core principles are consistent across student research. First, minimise the risk of harm: be sensitive to physical and emotional harm, avoid questions likely to distress participants, and do not pursue topics involving vulnerable participants without explicit supervisor approval. Second, obtain informed consent: participants must understand what the research is about, what they are being asked to do, and why, before they agree — and for participants under sixteen, parental consent is normally required. Third, protect anonymity and confidentiality: do not identify participants in your write-up unless they have explicitly agreed to be named, and store any identifiable data securely and separately from your analysis. Fourth, avoid deception: be honest about the research’s purpose; do not claim to study one thing while studying another. Fifth, give the right to withdraw: participants must be free to stop, or to withdraw their data, without penalty, and must know that in advance.
Put these principles into a short participant information note and a consent step at the start of every survey or interview: what the project is, what participation involves, that participation is voluntary, that answers will be anonymised, and how to withdraw. Keep the signed or recorded consents with your project records. If your topic touches sensitive issues — health, personal experience, political views — discuss it with your supervisor before approaching anyone; some topics are unsuitable for a school project, and discovering that after data collection is too late.
Using findings in the project: evidence, not decoration
Findings earn marks when they do analytical work in the outcome, not when they appear as an appendix nobody refers to. Present survey results clearly — tables or charts with the sample size stated, not percentages from tiny samples dressed up as population figures — and then interpret them: what pattern appears, how does it compare with the published literature, and what does it add to your argument? Present interview findings thematically, with short quotations that illustrate a theme, and with disagreements reported alongside agreements.
Be explicit about limitations, in the outcome itself rather than only in the evaluation. State the sample, its limits, and how far the findings generalise. A sentence such as “This survey of thirty Year 12 students at one school suggests, but cannot establish, that…” is stronger, in examiners’ eyes, than a sweeping claim from the same data. And never fabricate, selectively report or tidy up results: report the inconvenient finding, explain it if you can, and let it qualify your conclusion. Examiners trust a project whose data sometimes disagrees with its argument far more than one whose data agrees perfectly.
Finally, keep the balance right. Primary findings should supplement a project grounded in published research, not replace it. A project whose only evidence is its own small survey has built its argument on the weakest available foundation; a project that tests published claims against its own small, honestly reported dataset has used primary research exactly as it should be used.
How to get guidance on your EPQ primary research
If your survey questions keep steering their own answers, your sample will not support the claim you want to make, or your ethics plan has gaps you are not sure how to close, help before you collect data is worth far more than rescue afterwards — once data is collected badly, it usually cannot be repaired. Our support here is coaching, not substitution: we review your research design, question wording, sampling plan and ethics note, and comment on drafts you have written, with model material you can study and learn from — the data collection and the project itself stay yours, as they must. Support is priced from £20 per 1,000 words of material reviewed or prepared. To start, send your brief — your question, your proposed method, your sample and your deadline — and place an order through the site, email team@unitedacademichelp.uk, or message us on WhatsApp for a quick reply about what help would genuinely strengthen your research design.
Frequently asked questions
Do I have to do primary research for the EPQ? No. Many strong EPQ projects are library-based. Primary research is justified only when your question genuinely requires new data that published sources cannot supply.
What is the difference between primary research and primary sources? Primary research is new data you collect yourself, such as surveys or interviews. Primary sources are original evidence created by someone else that you analyse, such as documents or datasets.
How many survey responses do I need? There is no set number. What matters is that your sample fits your claim: a small, well-defined sample reported honestly outscores a large, undefined one presented as representative.
What are the key ethics principles? Minimise harm, obtain informed consent (including parental consent for under-16s), protect anonymity and confidentiality, avoid deception, and give participants the right to withdraw — all agreed with your supervisor before collection begins.
Can I survey my friends? Yes, as convenience sampling, provided you report it honestly and limit your claims to the group actually surveyed rather than generalising to a wider population.
Can someone collect my data or write up my findings for me? No. Data collection and the project itself must be your own supervised, authenticated work. Legitimate help is guidance — research design, question wording, sampling and ethics review, and feedback on drafts you have written.