A pilot study is a small-scale trial run of your research conducted before the main data collection, designed to test your instruments, procedures and logistics on a handful of participants so you can fix problems while they are still cheap to fix. Run it when you are using a newly designed or adapted questionnaire, interview guide or experimental protocol, or whenever you are unsure whether your plan will work in practice. In short, a pilot study is a rehearsal: it does not test your hypotheses, but it makes sure that everything you need for the real study actually works.
Most supervisors expect to see some form of piloting in a dissertation that involves primary data, because a flawed instrument cannot be rescued after the data are in. This guide explains exactly what a pilot study is for, how it differs from a feasibility study, what and how much to pilot, the steps to follow, and how to report the outcome in your methodology chapter.
What is a pilot study?
A pilot study (sometimes called a pilot test, pre-test or trial run) is a miniature version of your planned research, carried out on a small group of people who resemble your target sample. The purpose is not to produce findings but to de-risk the main study. You administer the questionnaire, run the interview, or apply the experimental procedure exactly as you intend to later, then ask: did it work, was it understood, did it take the time I expected, and would I change anything before going live?
Piloting belongs to the planning stage of any project that collects primary data, and it sits naturally alongside your wider decisions about methods of data collection and design. A pilot is where the abstract plan in your proposal meets the messy reality of real participants for the first time.
What is the purpose of a pilot study?
A good pilot study earns its place by reducing uncertainty across several fronts. Its core purposes are:
- Test the instrument. Check that every question is clear, unambiguous and answerable. You catch double-barrelled items, leading wording, confusing scales, missing response options and jargon that participants do not understand.
- Check timing and burden. Measure how long the questionnaire or interview actually takes. A survey that looks like ten minutes on paper can take twenty-five in practice, which drives up drop-out.
- Assess feasibility and logistics. Test the practical machinery: does the online survey link work on a phone, do skip-logic routes behave, can you reach participants, is the recording equipment audible, is the lab booking workable?
- Estimate recruitment and response rates. See how easy it is to find and enrol participants, and what proportion respond, so you can judge whether your planned sample is realistic.
- Refine procedures and the protocol. Rehearse your script, consent process, instructions and interviewer prompts so they run smoothly and consistently on the day.
- Trial the analysis plan. Enter the pilot data into your spreadsheet or statistics package and run a dummy analysis. This exposes coding problems, variables you forgot to capture, and questions that produce useless data long before the real dataset arrives.
Trialling the instrument is also your first practical check on measurement quality. While a small pilot cannot establish full reliability and validity, it can flag items that everyone interprets differently or that no one varies on, both of which threaten a sound measure.
“The pilot study is one of the important stages in a research project and is conducted to identify potential problem areas and deficiencies in the research instruments and protocol prior to implementation during the full study.” (Source: Hassan, Schattner & Mazza, 2006)
Pilot study vs feasibility study
Students often use these terms interchangeably, but methodologists draw a distinction. A feasibility study asks the broad question “Can this study be done at all?” It examines whether recruitment, retention, funding, ethics and the overall design are workable. A pilot study is narrower and more concrete: it is a small version of the actual study that tries out the specific procedures and instruments you intend to use. Put simply, every pilot study is a kind of feasibility test, but not every feasibility question requires running a mini-study.
| Aspect | Pilot study | Feasibility study |
|---|---|---|
| Core question | Do the instruments and procedures work as intended? | Can the study be done at all, and should it be? |
| Scope | Narrow: a miniature run of the real study | Broad: recruitment, retention, resources, design, acceptability |
| Typical output | Refined questionnaire, sharper protocol, timing estimates | A go / no-go decision and parameters for a future trial |
| Tests hypotheses? | No | No |
| Relationship | Often one component of a feasibility study | The wider umbrella; may contain a pilot |
For a typical undergraduate or master’s dissertation, what you run is almost always a pilot study in this sense: a small trial of your survey or interview before the full data collection.
What should you pilot?
Pilot anything that has to work flawlessly on the day and that you cannot easily fix once data collection has begun. The usual candidates are:
- Questionnaires and surveys. Wording, response scales, ordering, skip logic, length and the online platform. See our guidance on building a quantitative research questionnaire for the design choices a pilot puts to the test.
- Interview and focus-group guides. Whether questions open up rich talk, whether the order flows, whether prompts are needed, and how long a session runs.
- Experimental and observational protocols. Instructions, stimuli, timing, equipment, randomisation and the data-recording sheet.
- Procedures around the instrument. Consent forms, participant information sheets, recruitment messages and the sampling approach you describe in your sampling methods.
- The data pipeline. How responses export, how you will code and clean them, and whether your planned analysis runs on real (if tiny) data.
If your study runs a survey, it is worth piloting alongside the operational checks you would make when planning how to conduct surveys, so that distribution, reminders and response capture are all rehearsed together.
How big should a pilot sample be?
There is no universal figure, and a pilot is emphatically not powered to test hypotheses, so do not run formal significance tests or draw conclusions from it. Sensible rules of thumb include:
- A common guideline is to pilot on roughly 10% of the intended main sample, up to a practical ceiling. If your main study targets 200 respondents, a pilot of about 20 is reasonable.
- For small dissertations, even 5 to 10 participants who resemble your target group will surface most wording and logistics problems.
- Choose people similar to your target population but who will not be in the main sample, because they cannot then take part again without contaminating the real data.
- Bigger is not always better: the aim is to detect problems, not to estimate effects. Once new piloting stops revealing new issues, you have enough.
Because the pilot is underpowered by design, treat any numbers it produces as diagnostic only. Save the real estimation, comparisons and inference for the main study, where your sample is large enough to support them.
The steps of a pilot study
A pilot study follows a clear, procedural sequence. Work through it in order:
- Plan. Finalise a near-final version of your instrument and protocol, decide who will take part, how many, and exactly what you want the pilot to test (clarity, timing, logistics, analysis). Secure ethics approval if your institution requires it for piloting.
- Run. Administer the questionnaire, interview or experiment to your small pilot group under conditions as close to the real study as possible. Observe carefully and note anything that goes wrong.
- Gather feedback. Ask pilot participants directly: which questions were confusing, how long it felt, whether instructions made sense, and whether anything was missing or repetitive. This debrief is often more valuable than the data itself.
- Analyse. Examine the pilot responses for items with no variation, high missing rates, or patterns that suggest misunderstanding. Run your intended analysis on the tiny dataset to check it works mechanically.
- Refine. Make the changes: reword items, drop or add questions, reorder sections, adjust scales, tighten instructions, fix the platform, and update your timing estimates and recruitment plan.
- Proceed. Launch the main study with the improved instrument and procedure. If the pilot revealed major problems, loop back, revise and (occasionally) pilot again before going live.
Worked example: piloting a questionnaire
What she found. Average completion time was 22 minutes — far longer than the “10 minutes” she had advertised. Two items had high missing rates: Q9 (“How satisfied are you with your manager and your workload?”) was skipped by 6 of 15 respondents (40%), and Q14, which used an undefined term (“asynchronous culture”), was skipped by 5 (33%). A 7-point agreement scale was inconsistently labelled, and three respondents said the “Other” gender option was missing.
What she changed. (1) She split the double-barrelled Q9 into two separate items — manager satisfaction and workload satisfaction — so each could be answered cleanly. (2) She replaced the jargon in Q14 with plain wording and added a one-line definition. (3) She trimmed the questionnaire from 28 to 22 items, cutting completion time to about 12 minutes in a quick re-check. (4) She fixed the scale labels and added the missing response option. The pilot produced no “results” about remote working — and Aisha reported none — but it turned a confusing, over-long survey into a clean instrument and almost certainly lifted her main-study response rate.
How to report a pilot study
Describe your pilot briefly but transparently in the methodology chapter, usually just after you introduce the instrument. A clear report covers:
- Who the pilot participants were and how many (and that they were excluded from the main sample).
- What you tested (clarity, timing, logistics, the analysis plan).
- What problems emerged.
- What you changed as a result — this is the part examiners most want to see.
One or two well-written paragraphs are usually enough, and you should weave them into the methodology rather than tucking them into an appendix where examiners may miss them. The key message is that you piloted, learned something specific, and acted on it. Reporting the concrete changes you made — the items you reworded, the questions you dropped, the timing you corrected — demonstrates methodological care and reflexivity; reporting a pilot and then ignoring its findings does precisely the opposite and invites criticism.
Strengths and limitations
Piloting is one of the highest-return, lowest-cost things you can do in a dissertation, but it has boundaries.
Strengths
- Catches instrument and logistics problems while they are still cheap to fix.
- Improves data quality and likely boosts response and completion rates.
- Builds your confidence and competence in running the procedure.
- Gives realistic timing and recruitment estimates for the main study.
- Signals methodological rigour to supervisors and examiners.
Limitations
- Too small to test hypotheses or estimate effects reliably.
- Pilot findings about the topic itself can mislead if treated as real results.
- A non-representative pilot group may miss problems your real sample would hit.
- It costs time and a slice of your available participants.
- It cannot fix a fundamentally flawed research question or design.
Common mistakes to avoid
Most pilot-study errors come down to either skipping the step or misusing what it produces.
- Treating pilot data as findings. The cardinal sin. A pilot is underpowered; reporting its numbers as results, or running significance tests on them, is invalid.
- Ignoring what the pilot tells you. Running a pilot, spotting problems, and then launching the main study unchanged wastes the whole exercise.
- Piloting on the wrong people. Using participants who do not resemble your target population hides the problems that matter.
- Reusing pilot participants in the main sample. This contaminates your real data and inflates familiarity effects.
- Piloting too late. If you pilot only after launching, you cannot act on what you learn.
- Not actually testing the analysis. Skipping the dummy analysis means you discover broken variables only when the full dataset arrives.
Get your methodology chapter right the first time
Our experts help you design, pilot and write up rigorous primary research for your dissertation.
Conclusion
A pilot study is the rehearsal that protects your main study. By trialling your instrument, timing, logistics and analysis on a small, similar group, you find and fix the problems that would otherwise ruin your real data — and you do it while change is still easy. Keep it small, keep it close to the real conditions, never treat its output as results, and always act on what it teaches you. A few hours of piloting routinely saves a dissertation from weeks of unrecoverable error.
