A pilot study tries out parts of a planned study to find out whether the procedures are workable and what needs changing. Decide what uncertainty you need to resolve, how you will check it, and what decision the check will inform. A small set of responses does not automatically prove that the main study's hypothesis is true or that an instrument is valid.
Your supervisor and institutional requirements determine whether a pilot is appropriate and what permissions it needs. Do not recruit participants first and describe the exercise as an informal test to avoid review.
Keep the purpose, problems and changes from your pilot connected in the written project. UniDraft helps organise the procedure and outcomes you confirm into an editable Word document. Start your project documentation with your real records and supervisor instructions, then review the draft against what happened. See the project workflow; you conduct the pilot and remain responsible for permissions and interpretation.
A pilot is broader than checking a questionnaire
You may need to test instructions, a recruitment route, a data-recording procedure or whether the intended analysis can use the collected record. The questionnaire validity and reliability guide already covers item review and appropriate scale checks. This guide focuses on the wider collection process, not a second Cronbach's alpha tutorial.
Eldridge and colleagues' framework discusses feasibility and pilot work in preparation for randomised trials. Its terminology is useful background, but not a compulsory design for every undergraduate project. Name and justify the actual work your project requires rather than copying trial procedures into an unrelated study.
Start with a decision you need to make
Fictional planning example: imagine a proposed student survey about access to a departmental resource. No survey, respondent, pilot finding or approved protocol is being reported here. These original examples show how to plan a check.
| Practical uncertainty | Possible permitted pilot check | Decision it could inform |
|---|---|---|
| Do instructions explain the eligible time period? | Ask pilot participants to explain the period in their own words | Revise ambiguous instructions |
| Can the approved recruitment route reach eligible people? | Document the permitted invitations and eligibility checks | Review access or recruitment arrangements |
| Does the form handle a person who has never used the resource? | Try the relevant route through the questions | Add an appropriate branch or response option |
| Can the exported file be interpreted? | Check variable labels, category codes and missing entries | Fix the codebook or export process |
| Is the procedure practical within the available session? | Record actual duration and interruptions during the approved trial | Adjust the procedure and review the change |
Do not select arbitrary thresholds because another article used them. Agree relevant decision criteria in advance where the design calls for them. If a problem appears that you did not anticipate, record it honestly instead of changing the definition of success after seeing the result.
Write a short pilot plan
Use this original blank worksheet before the exercise:
| Field | Your actual plan |
|---|---|
| Question about feasibility | What uncertainty are you trying to resolve? |
| Procedure to try | Which part of the study will be rehearsed? |
| People or materials | Why are they appropriate for this check? |
| Required permission | What review, access and consent arrangements apply? |
| Evidence to retain | What record will show what happened? |
| Decision rule | How will you decide whether to continue, revise or seek advice? |
| Limits | Which parts of the main study are not being tested? |
For an interview-based project, use the interview-guide workflow to test clarity and follow-up, not just whether a recorder works. For observation, test whether an item can actually be judged and whether the recording sheet distinguishes an absent event from missing information.
Choose the size for the purpose
There is no single pilot size that fits every method and uncertainty. A wording check, a trial of an extraction procedure and a formal feasibility study have different needs. Explain the rationale with your supervisor and use appropriate methods literature for the actual design.
The main-study sample-size guide addresses a different decision. Do not take a percentage of that sample automatically or claim that a handful of convenient responses establishes statistical reliability.
NCCIH's pilot-study guidance, written for intervention research, cautions against treating a small feasibility pilot as a test of effectiveness or using unstable pilot effects to determine a later trial's sample size. The relevant lesson here is to keep the claims tied to the pilot's purpose; this is not a blanket prohibition on every statistic in every preliminary study.
Record problems and changes, not just success
Keep a dated change log alongside the approved pilot records:
| Version and item | What the pilot showed | Change proposed | Why it matters | Review before main collection |
|---|---|---|---|---|
| Fill in from your real record | State what actually happened | Revised wording, route or procedure | Link to the feasibility question | Record the applicable review decision |
Preserve unsuccessful checks too. A pilot that exposes a confusing question has produced useful information. Hiding the problem and claiming the instrument was already perfect removes the reason for the pilot.
If a change alters the study materially, obtain the required review before continuing. Do not imply that an earlier approval automatically covers a new participant group, data source or procedure.
Keep pilot records identifiable as pilot work
Use clear file and dataset labels. Whether pilot material may enter the main analysis depends on the design, changes and applicable approvals; make that decision explicitly. Do not quietly merge different instrument versions and describe all responses as collected under the final procedure.
Protect personal data under the agreed arrangements. A small pilot can still contain identifiable or confidential information. Do not upload it to a tool without permission for that processing or attach identifying consent forms to a public appendix.
Report it honestly in Chapter Three
In the methodology chapter, explain the pilot's purpose, who or what was involved, the procedure, what you learned and the changes made. State limitations: a rehearsal of one part of collection does not prove the whole design will work.
Original reporting scaffold, not a completed student account: “The pilot examined [specific practical uncertainty] using [actual permitted procedure]. The recorded issue was [real observation]. The guide or procedure changed from [earlier version] to [revised version] because [reason]. [State the actual review and remaining limitation].” Replace every bracket with true information. If the pilot has not happened, use future tense in the proposal and do not insert invented outcomes.
Before your defence, practise explaining one change that the pilot justified and one question it did not answer. If the panel asks whether the pilot proves the hypothesis, return to the purpose and limits of the evidence.
UniDraft can help organise the methodology and confirmed findings you provide. It does not conduct the pilot, recruit participants or turn this planning example into research evidence. Explore the project workflow.
