How to Write a PhD Research Proposal: Structure, Examples, and Clear Figures
A practical PhD research proposal guide covering structure, research questions, methods, timelines, common mistakes, and professional figures.
An applicant can have a promising question, strong grades, and relevant research experience—and still submit a proposal that feels impossible to evaluate. The problem is often not the idea. It is that the reader cannot quickly see the chain from research gap to question, method, evidence, and contribution.
A professional figure can make that chain visible. It cannot rescue a vague project, and it should never be added just to make the document look expensive. Used well, however, a research framework, method workflow, or timeline lets a supervisor or admissions panel inspect your reasoning without decoding five dense pages first.
This guide applies broadly to doctoral applications: university and research-institute admissions, application-and-assessment routes, structured doctoral programmes, and applications built around a self-proposed project. Requirements vary sharply, so use the structure below as a thinking tool—not as a substitute for the instructions of the programme you are applying to.
Thousands of researchers use SciDraw AI to make publication-ready figures for papers, grants, and journal submissions — in minutes, with no design skills.
A 2026 doctoral admissions outline from the Institute of Zoology, Chinese Academy of Sciences asks for a research plan of about 3,000 Chinese characters, including rationale, key scientific questions, methods, innovation, and an annual plan.
A 2026 application-and-assessment route at Renmin University of China allows up to 10,000 Chinese characters and does not prescribe one fixed template.
Charité's 2026 PhD/MD-PhD exposé format is capped at three pages and permits an optional appendix with up to two figures.
Before drafting, record five constraints: required sections, word or page limit, citation style, whether figures count toward the limit, and whether you must name or contact a prospective supervisor.
“AI in medical education” is a field. “How does AI-assisted feedback affect diagnostic calibration among first-year radiology trainees?” is a question that can drive a study.
A proposal is not a miniature textbook. A useful literature section moves through three steps: what is established, what remains uncertain, and why that uncertainty matters.
“Interviews, surveys, and machine learning” is a shopping list. Explain who or what will be studied, how access or sampling works, what will be measured, how the evidence will be analysed, and why the method answers the question.
You are not expected to know the findings in advance. State the evidence you expect to generate, the outcomes that would support or challenge the hypothesis, and what you will do if a critical step fails.
Panels assess not only whether the project is interesting, but whether the institution has the expertise, data access, equipment, archives, participants, or collaborations needed to support it. Cambridge's guidance explicitly frames value, achievability, and supervisory fit as central questions.
A colourful diagram with vague arrows consumes space and creates new questions. A figure deserves inclusion only if it shows a relationship, sequence, comparison, or decision more clearly than prose.
The research question, literature judgement, citations, and methodological choices must remain yours. The University of Manchester warns that proposals relying heavily on descriptive AI-produced text and showing limited independent thought are unlikely to succeed.
In one compact paragraph, state the problem, gap, main question, proposed approach, and expected contribution. Write this after the rest of the proposal even if it appears first.
Use one central question and a small number of supporting aims. Every aim should produce evidence that helps answer the central question. If an aim could disappear without affecting the main argument, it may be a separate project.
access, ethics, validity, or reproducibility considerations;
a realistic alternative if the preferred route fails.
For a visual first draft, turn this sequence into a method workflow. Keep the diagram at the level of research decisions; a laboratory protocol or software command list is usually too detailed for the main proposal.
Show that you understand what the project requires. Briefly connect your previous research or training to the proposed work, identify skills you still need to acquire, and explain why the prospective supervisor, department, laboratory, archive, dataset, or facility is relevant.
A timeline should show decisions and deliverables, not just dates. Depending on programme length and discipline, milestones might include ethics approval, pilot work, data access, fieldwork, analysis, chapter drafts, papers, prototypes, or validation studies.
Use the Research Roadmap Maker to draft a phase-based plan, then edit it to match the programme's actual duration. Do not recycle a three-year template for a four- or five-year programme without changing the logic.
Address consent, privacy, vulnerable participants, data security, dual-use concerns, fieldwork safety, cultural permissions, or conflicts of interest when relevant. Pair important risks with specific alternatives rather than writing “no major risks are anticipated.”
Use a focused bibliography that supports the decisions in the proposal. Verify every reference against the original source. Do not include papers simply to make the list look long.
Use a Conceptual Framework Maker for relationships among concepts, variables, mechanisms, or research aims. Label arrows with their meaning; an unlabeled arrow can hide the very claim the figure is meant to clarify.
Use a Research Roadmap Maker for phases, milestones, dependencies, and outputs. A roadmap explains research logic; a Gantt chart mainly explains scheduling. Use the one the instructions and your project actually need.
When the project depends on a biological mechanism, material process, system architecture, device, intervention, or spatial relationship, a Scientific Figure Maker can make the object of study concrete.
Methods explain how evidence will be produced; a timeline shows when decisions and milestones occur.
The following are not diagrams about diagram-making. They are hypothetical case figures of the kind an applicant could place in the proposal itself. The scientific claims, variables, and methods are examples only: replace them with your own project and verify every relationship before submission.
A mechanism figure turns a broad topic into testable links and makes the intervention point explicit.
This belongs after the background or hypotheses. Replace the exposure, biological compartments, mediators, outcome, and intervention with the elements of your study. Each arrow must be supported by literature or clearly marked as a proposed hypothesis; the T-bar must mean inhibition consistently. A first draft can be prepared with the Scientific Figure Maker, but the applicant remains responsible for biological accuracy.
The figure separates the main pathway, a moderator, a direct effect, and control variables.
Place this beside the conceptual framework or research hypotheses. Replace the constructs and H-labels, then define every variable in the text. Do not draw a causal arrow if an observational design can establish only association, and do not add a mediator or moderator merely to make the model look sophisticated. The Conceptual Framework Maker is useful for testing alternative structures before choosing one.
A methods figure should reveal data provenance, processing, model development, validation boundaries, and planned reporting without inventing results.
This fits in the methods section. Replace the data sources, quality checks, features, model, validation sets, and metrics with your actual design. Keep training, internal testing, and external validation visually separate. A proposal may name planned metrics, but it must not contain fabricated performance plots or scores. Use the Workflow Diagram Generator to make the sequence auditable.
short noun phrases instead of paragraphs inside boxes;
consistent fonts, arrow styles, and colour meanings;
enough contrast for grayscale printing;
a caption that states what the reader should learn;
editable output for supervisor feedback.
The NIH figure guidance recommends self-explanatory, relevant visuals, consistent typography and colour, and a grayscale check. Those principles transfer well to doctoral proposals even when the application is not a grant.
A figure is professional when it reduces the reader's work, not when it contains more effects.
Make a beautiful illustration for my PhD proposal.
This provides no scientific content, relationship, or acceptance criterion. The result will usually be generic decoration.
Better prompt
Create a clean research framework diagram for a doctoral application.Top: research gap — [one sentence].Center: central research question — [one sentence].Below: three aims arranged in [parallel / sequential] order.Under each aim, show the required data, core method, and expected evidence.Bottom: the original contribution.Use one top-to-bottom reading direction, short noun-phrase labels,a white background, deep navy and cyan with one coral accent,and no university logos, decorative icons, or fabricated results.
Generate a first draft, then verify every arrow and label yourself. If you need to keep revising in PowerPoint or Illustrator, convert the selected figure to an editable vector rather than rebuilding it from a flattened screenshot.
No. Some programmes require a detailed self-proposed project, while others recruit into an advertised project or ask only for research interests. Follow the current programme page.
Only when permitted and useful. A framework, workflow, or timeline is worth the space when it makes a relationship or sequence easier to evaluate. Decoration is not.
No. A proposal argues for the value and feasibility of a project. A personal statement explains your preparation, motivation, and fit. Some applications ask for both in one file, but the functions remain different.
Often not. Doctoral projects normally evolve after deeper literature review, training, data access, and supervisor feedback. The application proposal still needs to be coherent and feasible at the time you submit it.
Davie Chen is a researcher at the Faculty of Animation and Intermedia, University of Arts in Poznan, studying generative AI for scientific figure creation, patent illustration, and manuscript drafting. SciDraw AI is one of the research-to-product tools built from this work.
