Research Statement Example for a PhD Application

Research Statement Example for a PhD Application

A PhD applicant research statement reports research you have already done, project by project: the goal, your approach, the results and what they mean, in about 1,000 words at Penn and Iowa State, the two past-research prompts below. Only then does it point forward.

The examples that fill this search are mostly faculty-job research statements. They pitch a five-year research program, a funding plan and a lab. You are not in a position to propose that, and the applicant prompts do not ask for it. This page gives two complete constructed examples written to real prompts, annotated paragraph by paragraph, plus a rule for splitting material between the research statement and your other documents. If you are not sure which document you are writing, start with research statement vs SOP vs research proposal, which covers the vocabulary and the faculty-job trap in detail.

Every program page quoted below was checked on October 30, 2026. Limits and prompts can change between cycles; reread yours before you draft.

On this page: Which statement you are writing · Example 1: wet lab · Example 2: computer science · Splitting material · Results that did not work · Questions

Which research statement are you being asked for?

Four programs, four different jobs under the same or similar name:

If the prompt asks what you want to study rather than what you have done, see statement of research interests examples.

Applying without a master's? PhD straight from undergrad shows which programs take bachelor's applicants and what evidence they read.

Penn's prompt is the cleanest statement of the applicant genre: "The research statement should describe your meaningful research experiences, highlighting your findings." It lets you choose: "You may choose to describe a single research experience or several experiences".

Iowa State's list of "Common Pitfalls to Avoid" also covers writing tools. Quoted in full: "Generative AI Writing: Avoid the temptation to use these tools to write content for you. They often produce text that is bland, uninspired, and just awkward. Most importantly, they don’t convey your enthusiasm for your research and interest in our graduate program."

Our view: unless your program says otherwise, write the research statement as a report on work done, with one short forward paragraph. Penn and Iowa State, the two programs in the table that describe the report in detail, both put past research first. A statement that leads with a future agenda spends its space on the part a committee can verify least.

The countercase: Penn Weitzman's 500-word statement asks for the field and topic you propose and how you are prepared. UW iSchool asks for "one or two potential research projects". For those, write a short proposal grounded in your preparation, and use the PhD research proposal guide for the structure. The examples below fit the Penn and Iowa State shapes.

Annotated example 1: wet-lab project with partial results

Context (invented): a microbiology senior applying to a biomedical PhD program with a Penn-style prompt, around 1,000 words. She has one 18-month thesis project and one summer project. The names, lab and results are invented.

Constructed example, written by us, not a real applicant statement. The statement is 543 words, about half of Penn's limit, so there is room for a third project or more detail. Each paragraph is followed by a note on which part of the prompt it answers.

I have worked on two projects about how bacteria survive antibiotic stress: an 18-month honors thesis in Dr. A. Okafor's lab at Riverbend State University, and a ten-week computational internship. In both, the question was the same: which regulatory changes let a cell tolerate a drug before it becomes resistant?

Note: one sentence of orientation, one sentence of question. No childhood story. It tells the reader there are two projects and how they connect.

Thesis project: a candidate regulator of efflux. The lab studies drug tolerance in E. coli. Our goal was to test whether a putative transcription factor, which we called R7, controls the expression of an efflux pump during low-dose antibiotic exposure. My role was to build the deletion strain, design the exposure assay and run the expression measurements. Dr. Okafor proposed the target; I proposed the time-course design after the first single-timepoint assay gave inconsistent results.

Note: the goal of the project and the applicant's own role, separated from the lab's and the adviser's. This is the "goals of each project" element.

I made the R7 deletion strain using lambda Red recombination and confirmed it by PCR and sequencing. I then exposed wild-type and deletion cultures to sub-inhibitory drug concentrations and measured pump transcript levels by qPCR at 0, 30, 60 and 120 minutes. Each condition had three biological replicates. Because the first assays drifted between days, I added a reference strain to every plate and normalized to it.

Note: "approaches used". Methods are named with enough detail to show the applicant did them, and the troubleshooting step shows judgment.

In the deletion strain, pump induction at 60 minutes was roughly half of wild-type, and the difference disappeared by 120 minutes. Survival after four hours of exposure did not differ between strains. So R7 appears to speed up the pump's response without changing whether cells survive this dose. I presented the result as a poster at our department symposium; it is not published.

Note: "results obtained", including the part that did not support the hypothesis. The publication status is stated plainly.

The result matters for two reasons. For the project, it narrows R7's role from "required for tolerance" to "affects timing", which changes the next experiment: the lab is now testing whether timing matters at higher doses, where a slower response might be lethal. For the field, it is a small example of a broader point I find important: tolerance may depend on how fast a cell responds, not only on whether it can respond.

Note: "implications of the findings for the project and the field at large", in that order. The claim about the field is phrased as the applicant's interpretation, not as settled fact.

Summer project: reading tolerance from expression data. In summer 2025 I joined a computational group for ten weeks to ask whether public expression datasets show the same early-response pattern. My goal was narrower than the thesis: build a pipeline that could compare response timing across experiments. I wrote it in Python, processed four datasets and found that only two had enough early timepoints to compare. The answer was a methods lesson rather than a biological one: most datasets I found were sampled too late to see a timing effect.

Note: a second, shorter project with its own goal, approach and result. It shows a skill (computation) the first project did not, and it reports a limited result honestly.

What I took from both. The thesis taught me to change an assay when the data told me it was unreliable, rather than running more replicates of a flawed design. The internship taught me to check whether existing data can answer a question before planning an experiment. Both changed how I would approach the next project.

Note: what the applicant learned, stated as changed behavior, not as adjectives.

Where this points. I want to study how the speed of bacterial stress responses shapes survival, using both genetic tools and time-resolved measurements. I am not yet sure which system is right for that question, and rotations are how I want to find out. I would especially like to learn single-cell methods, because my bulk qPCR measurements could not tell whether all cells slowed down or only some did.

Note: one forward paragraph, tied to a gap in the applicant's own data. It does not propose a dissertation.

What example 1 shows

It follows Penn's four elements in order for each project and never claims a result it did not get. The forward paragraph grows out of a limitation in the applicant's own measurements, which is easier to believe than a five-year plan. If you are applying to a rotation-based life-sciences program, that same honesty about not yet knowing the right lab is what the forward paragraph is for.

Annotated example 2: two-part CS statement

Context (invented): a computer science senior applying to a PhD program that asks for an Iowa State-style two-part statement. One research project, one course project.

Constructed example, written by us, not a real applicant statement. The statement is 360 words, including the two part headings.

Part 1: Prior research experience

My research interest is making program-analysis tools usable on code that students write. For three semesters I worked in Dr. L. Moreau's software engineering group at Riverbend State University on a static checker for introductory Python assignments.

The group had a checker that flagged common errors but produced warnings that students ignored. My project was to find out why. I collected 1,200 anonymized submissions from two course sections, with the instructor's permission, and labelled each warning by whether the student fixed the flagged line in the next submission. Only about a third of warnings led to a fix. Warnings that named a line and suggested a change were fixed far more often than warnings that described a rule.

I then rewrote the twelve most frequent warning messages in that style and ran a one-semester comparison in one section. The fix rate rose, but the section also had a different instructor, so I cannot separate the two effects. I reported this limitation in my write-up for the group and proposed a within-section design for next year, which the lab has adopted.

In a separate course project I built a small type-inference prototype for a subset of Python. It handled only simple functions, but it taught me how much of the checker's difficulty comes from code that is valid but unusual.

Part 2: Research goals

I want to study how analysis tools should explain their results to people who are still learning to program. My first project as a PhD student would extend the warning study: test, with a proper control, whether explanation style changes fix rates, and whether the effect lasts beyond the course.

Longer term, I am interested in tools that adapt their explanations to what a student has already seen. That direction combines the program analysis I have started to learn with human-subjects methods I have used only once, and I would need training in both.

I am applying to work with faculty whose groups study program analysis and computing education. The combination matters because the question is not only how to detect an error but how to make a learner act on it.

Annotation:

• Part 1 reports, Part 2 plans. Every claim in Part 2 grows out of something in Part 1: the warning study, its uncontrolled comparison, the prototype's limits.
• The applicant's role is clear. "I collected", "I labelled", "I rewrote", "I proposed". The group's existing checker is credited to the group.
• The weak result is the strongest paragraph. The confounded comparison is reported with its cause and the fix the applicant proposed. That is evidence of research judgment, which a perfect result would not show.
• No program-specific padding. The final paragraph names the kind of faculty, not a list of names. In your own statement, name specific people; Iowa State's guidance says "Mention specific faculty members and their work that interests you."

Splitting material between the SOP and the research statement

Penn Biomedical Graduate Studies asks for three documents in one application: a personal statement "around 500 words in length", the research statement "around 1000 words", and a community statement of "150-200 words". The same split works for any program that asks for more than one.

If your program also asks for a paper, check the grad school writing sample rules before you excerpt one.

Two rules keep the set from overlapping. First, if a sentence describes a method, it belongs in the research statement. Second, each document may mention the others' material only in a clause that points to it. Iowa State's guidance says it in one line: "Do not simply rehash your Statement of Purpose (SOP)." For the community statement, see our diversity and community statement examples.

Writing up results that did not work

Both examples above report a result that did not support the original idea. That is deliberate. A research statement is the one document where a committee can see how you think when an experiment fails.

Three rules:

1. State the result before the explanation. "Survival did not differ between strains" comes before any reason why.
2. Say what the result changed. The next experiment, the design, or your interpretation.
3. Do not apologize or hide it. A null result written up as evidence of judgment reads better than a vague claim of success.

For more on this, see explaining negative research results in a PhD SOP and describing research still in progress. If you are turning a thesis abstract into the research paragraph of your SOP instead, see turning a thesis abstract into evidence.

Questions about PhD research statements

How long should a research statement be if the program gives no limit?

Our recommendation: about 1,000 words. Penn BGS asks for "around 1000 words" and Iowa State caps it at 1,000 words or two pages. UW iSchool's research-focused SOP runs 1,000 to 2,000 words, and Penn Weitzman's forward-looking statement is capped at 500, so check which shape your program wants.

Can I include figures or citations in a research statement?

None of the four program pages above mentions figures. Iowa State recommends "informative section headings, bullet points, an easily readable font size, and reasonably sized margins". Our view: cite a paper or poster briefly in the text when you mention it, list it fully on your CV, and leave figures out unless the program invites them.

Should a research statement be written in the first person?

Yes. The reader needs to know what you did, and "I designed the assay" is clearer than "the assay was designed". Use "we" only for work the group did together, and say who did what.

Which parts of example 1 can I reuse for another program?

The project paragraphs carry over. Rewrite the forward paragraph and any fit sentences for each program, and recut to its limit: around 1,000 words at Penn BGS, at most 1,000 at Iowa State. Penn Weitzman's 500-word statement asks for a proposed field and topic, so write that one fresh. Whether master's applicants need a research statement at all is covered in research statement vs SOP vs research proposal. If your application starts with an expression of interest to an Australian supervisor, write that one-page pitch before the research statement.

For a biology program that asks for one statement, see biology PhD statement of purpose examples for rotation and advisor-admit programs.

Writing for an NIH fellowship instead of admission? The F31 has its own four-heading candidate section; see what replaced the F31 Applicant's Background and Goals.

If your next document is an NIH fellowship, the aims page works differently; see this annotated F31 Specific Aims example.

Fellowship statements follow their own rules: the DOE CSGF Personal Statement has to argue for a required Program of Study.

Once you are in a PhD, fellowship proposals reuse the same skills; the Google PhD Fellowship research proposal is capped at 3 pages and needs a university nomination.

If you wrote a Goldwater research essay as an undergraduate, its first-person account of one project is a usable starting point for this statement.

Results that arrive after the deadline follow different rules; see how to update a submitted grad application with a new paper.

Applying to a practice doctorate in education instead of a research PhD? See EdD statement of purpose examples.

Mathematics departments read the statement differently; see the math PhD statement of purpose guide for describing an REU result so a mathematician can check it.

Get a free review of your research statement

Get a free Quick Review of your research statement against GradPilot's PhD SOP Research Dossier rubric. It is a PhD SOP rubric built for statements that programs read as a research record, the closest fit because GradPilot has no separate research-statement rubric. It reads for how your past research shows what you can do, your own role in team work, and whether the work connects to a problem you want to explore next. Applying to a rotation-based life-sciences program? Use the life sciences PhD SOP rubric. For the statement of purpose that sits beside it, use the PhD statement of purpose rubric or start from the PhD SOP review.

More guides for the same application:

• PhD statement of purpose examples: structure and paragraph budgets for the SOP itself.
• CV for grad school applications: list every project in this statement with the same dates and role.
• All PhD essay guides.

Sources

Retrieved October 30, 2026 (UTC), except where noted.

• University of Pennsylvania Biomedical Graduate Studies, PhD Application Information (Fall 2027): research, personal and community statement prompts and lengths.
• Iowa State University Department of Computer Science, How to Write a Research Statement (no year stated): two-part structure, 1,000-word cap, formatting advice, the SOP and generative-AI lines.
• Penn Weitzman School of Design, How to Apply (fall 2027): 500-word research statement for PhD and MS in Architecture applicants.
• University of Washington Information School, Ph.D. Application Materials (Autumn 2027): 1,000–2,000-word research statement elements.