A public health research proposal should connect a defined population, a useful question and a feasible study. This guide shows how to explain that connection using an annotated fictional example about appointment access.
Fictional teaching example: the project, district and proposed clinics below are invented. No recruitment, dataset access, ethics approval or findings exist. The research gap is a working proposition that a real applicant must verify through a literature review. Numbers are planning illustrations, not a sample size recommendation.
Before you write: narrow the question
“Improving healthcare” is a goal, not a research question. Specify whose experience you will study, which service matters and what you can measure. A cross-sectional survey can describe patterns and associations; it cannot establish that a barrier caused missed appointments.
Annotated public health proposal example
1. Working title
Reported barriers to attending scheduled primary-care appointments among adults in fictional Arunika District.
Why this works: it identifies the population, service and setting without promising an intervention’s effectiveness.
2. Background and proposed gap
Appointment attendance is a practical issue for service planning. This proposed study would examine patients’ accounts of transport, scheduling and information barriers alongside reported attendance. The preliminary question is whether locally collected information could help participating clinics distinguish the types of difficulty patients describe.
Before submission, I would review research on appointment access and attendance in comparable settings. I would identify what those studies measured, which populations they included and whether the proposed local study adds a useful perspective. I would revise or abandon the project if the literature and available local evidence already answer the question adequately.
Annotation: this paragraph establishes a problem and a verification task. It does not invent a claim that “no previous studies exist.” In a real proposal, replace it with a sourced argument and explain the study’s contribution beyond collecting another dataset.
3. Question and objectives
Main question: among eligible adult patients with a scheduled appointment during the preceding three months, what barriers are reported, and how are those reports associated with self-reported non-attendance?
- Describe reported barriers and the proportion reporting at least one missed appointment.
- Explore associations between selected barriers and non-attendance, accounting for a small set of prespecified relevant variables.
- Describe patients’ suggestions for making appointment information easier to use.
Annotation: define the recall window and outcome before collecting data. “At least one missed appointment” is a patient-level outcome, not the clinic’s administrative non-attendance rate.
4. Design, eligibility and recruitment
I propose a cross-sectional questionnaire at two hypothetical primary-care clinics. Eligibility would require age 18 or above and at least one scheduled appointment in the preceding three months. Recruitment would use a documented consecutive approach during prespecified sessions, subject to site permission and ethical review. Participation would be voluntary and would not affect care.
A planning target of 200 respondents would be assessed against the precision needed for the primary descriptive estimate, the likely number of non-attendance outcomes, recruitment capacity and nonresponse. I would provide a justified sample size calculation before the protocol is finalised. If there are too few outcomes, I would simplify or omit adjusted modelling rather than fit an unstable model.
Annotation: recruiting people who are currently at a clinic may miss those facing the greatest access barriers. The proposal must acknowledge this selection problem. An alternative recruitment route would require separate feasibility and privacy assessment.
5. Measures and analysis
The questionnaire would distinguish appointment history, transport time, work or caring conflicts, clarity of appointment information and demographic variables needed for the question. I would review suitable existing measures, check permissions where relevant and pilot comprehension before recruitment. A locally written question is not automatically a validated scale.
I would report denominators, missing values and descriptive estimates with uncertainty. Prespecified exploratory analyses could use logistic regression for the binary patient-level outcome, with attention to sparse data and clinic differences. With only two clinics, I would avoid claiming reliable cluster-level effects. A short optional open-text question would be summarised using a documented descriptive coding process.
Analysis would use R or jamovi, with a data dictionary and a reproducible record of cleaning decisions. Results would be described as associations, with recall and selection limitations stated explicitly.
6. Ethics, access and data management
I would seek the appropriate institutional ethics review and clinic permissions before starting. Information sheets would explain voluntary participation, data use and withdrawal arrangements. The survey would avoid unnecessary identifiers; any consent records would be stored separately under institutional rules. Public outputs would contain aggregate results, and small groups would be handled carefully to reduce identification risk.
Annotation: do not write “ethical approval has been obtained” in an application unless that is true. Institutional review determines the approved procedure.
7. Feasibility, timeline and contribution
An illustrative 12-month plan allocates months 1–2 to the literature and protocol, 3–4 to review and permissions, 5 to piloting, 6–8 to recruitment, 9–10 to analysis and 11–12 to writing. Recruitment dates remain conditional on approvals. A budget would itemise travel, printing or secure survey tools, transcription if needed and dissemination, using verified local costs.
The intended output is a dissertation and an accessible summary of reported barriers for participating services. The study would not prove that changing appointment procedures improves attendance. If access or recruitment fails, a revised secondary-data or literature-based question would need supervisor agreement and any relevant approval.
Common proposal mistakes to avoid
- Calling an association an intervention effect.
- Using a convenient sample size without a justification.
- Promising access to patient records before permission.
- Listing statistical tests without defining outcomes and variables.
- Writing expected findings as if they were results.
Adapt this example for your application
- Read the degree and scholarship instructions first, including headings, word limits and supervisor-contact rules.
- Replace the illustrative gap with a conclusion supported by your own literature review and accurate references.
- Confirm access to data, equipment or sources. Do not describe permission, supervision or funding as secured unless it is.
- Match each objective to a method, an output and a realistic timeline.
- Remove the teaching annotations before submitting, and follow the institution’s rules on editing and AI assistance.
This is a teaching example, not an approved protocol or a successful scholarship proposal. A Master’s project may focus on one manageable question. A PhD proposal needs a defensible original contribution and a plan appropriate to the program’s duration.
Official proposal-writing guidance
Check Oxford’s research proposal guidance and Edinburgh’s proposal-writing guidance, then consult your target course. These are general writing resources, not validation of this fictional project’s methods. Links checked 10 October 2026.
Continue with the research methods guide, or return to Academic Writing.