A pilot study is a small-scale rehearsal of your main research, run to find problems before they become expensive. It tests the instruments, the recruitment, the procedures, and the analysis on a handful of cases, so that the main study runs on methods that have already been tried. Researchers who skip this step often discover too late that a question is ambiguous, a procedure is unworkable, or a sample cannot be reached, and by then the main data collection is already underway.
This guide explains how to run a pilot study properly: what to test, how many participants you need, how to design and run the pilot, how to interpret its results without over-claiming, and how to revise your methods afterwards. A pilot is not a small version of the main study done for its own sake. It is a diagnostic exercise with a clear list of questions, and its value lies in the changes it causes, not in the results it produces.
A pilot study is a small, preliminary version of a planned study, conducted to test feasibility rather than to answer the research question. It uses the same instruments, procedures, and analysis plans as the main study, but on a smaller scale and with the explicit aim of finding weaknesses. A pilot questionnaire, for example, is given to a few respondents to check whether the questions are understood, whether the answers can be analysed, and how long completion takes.
Pilot studies differ from the main study in purpose, not just in size. The main study aims to produce findings; the pilot aims to produce decisions about methods. That distinction shapes everything: pilot results are reported as lessons and changes, not as conclusions about the research question, and pilot participants are usually excluded from the main study if the methods changed between the two, so that the main sample remains clean.
Piloting saves time, money, and credibility. An ambiguous questionnaire item discovered in a pilot costs one revision; the same item discovered in the main study can invalidate hundreds of responses. A recruitment route that fails in a pilot can be replaced; a route that fails in the main study can leave the study underpowered. Procedures that confuse participants, equipment that malfunctions, and analysis plans that do not fit the data are all cheaper to fix before the main study begins.
Piloting also improves the study in ways that are hard to see in advance. Participants in a pilot often raise issues the researcher never considered, such as a question that feels intrusive, a task that takes twice as long as planned, or a setting that changes how people respond. These insights come only from trying the study for real, with real participants, under real conditions. A pilot is therefore not a delay to the main study but a protection for it.
Write a list of pilot questions before you begin, because a pilot without questions drifts into an informal trial that teaches little. Typical questions cover whether participants understand each item or instruction, whether the procedure runs in the time allowed, whether recruitment reaches the intended people, whether data can be recorded and cleaned as planned, and whether the planned analysis can actually be run on the kind of data collected.
Include practical questions too: how long does each stage take, what does it cost, what equipment or permissions are needed, and where do participants drop out or hesitate? Ask pilot participants directly for feedback, through a short debrief or a few open questions at the end, because their experience reveals problems that the data alone cannot show. A question that everyone answers the same way, or that everyone skips, is telling you something important even before any analysis.
There is no single correct number, because a pilot is not designed to estimate an effect. The size follows from what you need to test: enough people to reveal problems with the instruments and procedures, but not so many that the pilot becomes a second main study. Guidance in the methods literature often points to small numbers for feasibility purposes, with larger pilots only when the pilot must also estimate variability for planning the main sample.
The table below gives a practical starting point. Treat these numbers as rules of thumb to discuss with a supervisor, not as requirements, because the right size depends on the complexity of the instruments and the diversity of the target group.
| Pilot purpose | Typical size guidance | What it reveals |
|---|---|---|
| Test a questionnaire's wording | 5–15 respondents | Ambiguous items, timing, missing answers |
| Test an interview guide | 3–6 interviews | Flow, sensitive topics, depth of answers |
| Test a full procedure end to end | 10–30 cases | Recruitment, logistics, data handling |
| Estimate variability for sample planning | 25–50 cases | Rough spread of key measures |
Whatever size you choose, justify it in terms of the pilot questions, not by copying a number from another study. A pilot of five people that tests wording thoroughly is better than a pilot of fifty that tests nothing in particular. If your pilot must inform a sample size calculation, say so explicitly and treat the estimate as rough, because small samples give unstable estimates of variability.
Write a short pilot protocol before recruiting anyone. It should state the pilot questions, the participants, the instruments and procedures exactly as they will run in the main study, the data to be collected, and the criteria for deciding what counts as a problem. Running the pilot exactly as the main study will run is essential: a pilot that uses a simplified procedure tests the simplification, not the study.
Include ethics and consent in the protocol. Pilot participants are research participants, and they deserve the same information, consent, and confidentiality protections as main study participants, even when the pilot feels informal. If the pilot involves people or animals, it needs the appropriate approvals before it begins. Build that approval time into your schedule, and do not treat the pilot as a way to bypass review.
Run the pilot under realistic conditions, with the same setting, timing, and staff roles as the main study will use. Observe carefully and record everything: how long each stage takes, where participants hesitate, which questions are asked back, what goes wrong with equipment or forms, and how the data look when they come in. A researcher who only collects the pilot data without watching the process misses half the value of the exercise.
Debrief every pilot participant briefly. Ask which questions were unclear, which tasks felt difficult or uncomfortable, and what they thought the study was about, because a mismatch between their understanding and your intention signals a design problem. Also debrief anyone helping to run the study, since research assistants and fieldworkers notice practical problems that the lead researcher may miss. Keep a log of every issue, however small, because small issues in a pilot become large ones at full scale.
Analyse pilot data to answer the pilot questions, not the research question. Check whether items behave as expected, whether response patterns make sense, whether missing data cluster on particular questions, and whether the planned analysis runs on the data collected. Descriptive summaries are usually enough; formal statistical tests on pilot data are rarely meaningful, because the sample is small and the study was not designed to test the hypothesis.
Resist the temptation to report pilot findings as if they were main study results. A difference observed in ten pilot cases may be noise, and presenting it as a finding misleads readers and embarrasses the researcher when the main study finds otherwise. If pilot results are mentioned in the main paper at all, they belong in the methods section as a brief account of what was tested and changed, not in the results section as evidence for the conclusion.
Turn the issue log into decisions. For each problem, decide whether to revise the instrument, change the procedure, adjust the recruitment, or accept the issue as a known limitation. Revise generously: rewording a question, reordering a section, shortening a task, and clarifying an instruction are all cheap at this stage. After major revisions, test the revised version again on a few new participants, because a fix that has not been tried is only a guess.
Document every change and its reason, and keep the original and revised versions of all instruments. This record matters for three reasons: it shows reviewers that the main methods were tested, it lets you explain any difference between pilot and main procedures, and it protects you if a later reader asks why a question was worded a particular way. If revisions were substantial, pilot participants should not join the main study, since they have already experienced an earlier version of the methods.
When you write up the main study, report the pilot briefly and honestly in the methods section: how many people took part, what was tested, what problems were found, and what changed as a result. This account strengthens the paper, because it shows that the methods were not merely asserted but tried. Reviewers generally respond well to a pilot that led to visible improvements, and poorly to methods that appear untested.
Do not hide a pilot that went badly, and do not inflate a pilot that went smoothly. If the pilot revealed that recruitment was difficult, say so and explain how the main study addressed it. If the pilot changed nothing, say that too, since a clean pilot is itself useful information. Transparency about methods, including their testing, is one of the clearest signals of a careful researcher.
A pilot is not always necessary. Studies using well-established instruments in familiar settings, replications of a previous design, and analyses of existing datasets may need only a small check rather than a formal pilot. Very small qualitative studies sometimes pilot through the first interview or two, revising the guide as they go, which is a legitimate approach when it is described honestly.
The decision should be deliberate, not a shortcut. If you skip a formal pilot, write down why: the instruments are validated, the setting is familiar, or the design has been used successfully before. If none of those reasons applies, run the pilot. The time it costs is almost always less than the time a flawed main study wastes, and the confidence it gives is worth more than the delay it causes. For the wider design choices a pilot protects, see our guide on how to choose a research methodology for your study and our guide on how to design a research questionnaire.
Ready to publish your study? Submit your manuscript to Global Academic Publishers and follow the journal's author guidelines for the Global Journal of Veterinary and Animal Sciences.
Only if the methods did not change between the pilot and the main study. If instruments or procedures were revised, pilot participants should be excluded so the main sample experiences one consistent version.
They overlap. A pilot study is a small version of the planned study run to test its methods, while a feasibility study is broader and may also examine recruitment, resources, and whether the main study is possible at all.
Yes, when the pilot involves people or animals. Pilot participants deserve the same consent and confidentiality protections as main study participants, and approval should be obtained before the pilot begins.
Sometimes, if the pilot itself makes a useful methodological contribution. Usually, though, pilot results belong in the methods section of the main paper as an account of what was tested and changed.
That is the pilot doing its job. Revise the design, test the revision, and only then run the main study. A failed pilot is far cheaper than a failed main study.
It depends on the design, but a pilot should be quick relative to the main study: long enough to test every stage realistically, short enough that it does not consume the time and participants the main study needs.