Tips for good structure and rhetoric in NSF GRFP applications

Leave a Comment / By /

This is a mild rant + framework for writing an NSF GRFP application. This is based on common issues that I’ve seen in GRFP applications. I’ve seen great presentations on how to write a good GRFP application, but keep seeing a common set of issues from students who attend those great presentations. This post provides tips and structure that can help create a good GRFP draft. It isn’t perfect; you may want to reject recommendations that conflict with other reputable resources.

Overview:

The purpose of this document is not to review all of the NSF guidelines for the GRFP. Check the NSF guidelines for more detail on IM and BI and why they are important.

In 2025, budget cuts to the NSF reduced the number of GRFs awarded to 1000. In contrast, over 3000 Honorable Mentions were awarded. In previous years, the Award:Honorable Mention split was roughly 2000:2000. This means that the award is, and will probably continue to be, much more competitive. Writing strong application statements is more important than ever.

This document focuses on how to structure a proposal. The goal of structure is to help create a coherent story that reviewers can follow easily. You want to make the reviewer’s life easy. Good structure also makes the application much easier to write.

I’ve included a few examples in the guidelines below and explanations of why each example is good. I rarely see examples provided and critiqued, despite almost everyone appreciating a good example when doing something they’ve never done before.

Below I include the following sections:

  • TL; DR
  • Expectation setting for the applicant
  • Making reviewers happy
  • Personal Statement
  • Research Statement
  • Writing efficiently
  • Rhetorical comments

Abbreviations:

  • NSF – National Science Foundation
  • GRFP – Graduate Research Fellowship Program
  • IM – intellectual merit
  • BI – Broader impacts

TL; DR

Don’t write like a creative writer. Write like a scientist. You may not yet know how to do that. If not, please read the below sections. Then write for a while. Then re-read the below sections to see if they make more sense.

Use consistent, predictable structure. This make reviewer’s lives easy. Get to the point. Find LOTS of examples of successful statements.

Use why-then-what writing structure: State why you did something, then what you did (then why it matters).

Don’t show that you want to be a badass scientist; provide evidence that you will become a badass scientist.

Start early. Get feedback early. Get to the point quickly in each paragraph.

Not all items below will be relevant or 100% accurate. Please use this as a supplement to recommendations that you get from others.

Expectation setting for the applicant

Especially if you’re an undergrad applicant and especially if you’ve never first-authored a scientific paper, you probably don’t know how to write for scientists. That’s fine. That’s also still the case for grad applicants, most of whom will submit a GRFP application early in their second year of grad school. This means that you need LOTS of feedback from your peers (see “Writing Efficiently”). This will help nail down the content (first) and then the style (second).

Plan to start writing a couple of months in advance. Aim for weekly feedback as a default. Busy mentors may take longer. A large and supportive lab may give you bits of feedback more rapidly. Expect many iterations.

You should probably have your CV completed before starting your Personal Statement. Your Personal Statement should point expand upon impressive things in your CV. CV and Personal Statement need to be consistent.

Making reviewers happy

This section is all about making reviewers’ lives easy. This section may make more sense after you read the Personal Statement and Research Statement sections.

Expect a good reviewer to read (1) the first sentence of each paragraph and (2) boldfaced text. In good scientific writing, the first sentence of each paragraph should tell the entire story if that’s all you read. Additional sentences in each paragraph support the first sentence for interested and/or skeptical readers. This writing style has a few implications.

  • If the first sentence of each paragraph is not compelling/interesting/clear, reviewers may skip the paragraph.
  • The first sentence of each paragraph should unambiguously tell the reviewer what you want them to say about you.
  • You should be able to read the first sentence of each paragraph and understand the story, especially for the Personal Statement.
  • You do not need to boldface the first sentence of a paragraph. It will be read either way.
  • Instead, boldface text should be strong, measurable support for the first sentence of the paragraph.

You probably have a vague idea of what you want reviewers to say about you. Tell them what you want them to write in their review of your application. Do not make the reviewers infer why you’re great. If left to interpret your qualifications, reviewers will misinterpret your accomplishments, think you’re a poor communicator, and might confuse your application with someone else’s (this has happened).

For example, if you want reviewers to say “X shows clear potential as an independent and productive researcher, evidenced by their presentations at X conference and their upcoming first-authored manuscript,” you should start a paragraph with “My productivity on an independent research project supports my potential to create and disseminate impactful research.” In the same paragraph, you describe how you led an independent project and include a statement in the paragraph like “I presented this work at X conference and am preparing a first-authored manuscript.” This statement should be boldfaced. This will let the reviewer construct their description of you by skimming the first sentence of the paragraph and the boldfaced sentence, which will be the next thing their eyes go to.

Telling what the reviewer what to write is easier in the Personal Statement. For the Research Statement, you want to match reviewers’ expectations. Generally, the most important things are to use consistent language throughout the statement and to use consistent structure across paragraphs.

An example of consistent language is using the same terms in a paragraph describing the proposal scope (i.e., goals and projects) and the title of each project. For example, if you say “Project 1 will uncover mechanisms driving individual differences in X.” The Project 1 title should be “Uncovering mechanisms driving individual differences in X,” not “Discovering drivers of inter-individual differences in X.“

An example of consistent structure could be if you propose 2 projects and Project 1 is structured: Gap > research question > Hypothesis/predictions > population > intervention > outcomes and comparisons > potential interpretation / metrics for success, use that same structure in Project 2. There are cases where the projects are so different that you’ll need to change the structure. However, the more consistent you can be, the easier it will be to read.

Personal Statement

The Personal Statement should tell reviewers why you are likely to be an intellectually and societally impactful researcher and/or educator. There is not a single way to structure the Personal Statement. Often, the statement should clearly convey the following topics: (1) Overview and motivation, (2) Intellectual Merit, (3) Broader Impacts, (4) Why a GRF would be impactful. I will use this structure below. The principles described below should generalize to the other structures. I encourage you to review other structures.

Before writing paragraphs, I recommend first writing a list of takeaways that you want the reviewer to take away from reading your statement. Get feedback on those. Then those statements become the topic sentences of different paragraphs. Then the paragraphs are written to provide measurable evidence of the main takeaway.

Important Supporting evidence statements are ideally measurable and should at least contain a metric for success. Don’t just say what you did, describe how successful it was. Better yet, first say why you did it, then what you did, then how it was impactful. For example, if you led an outreach event, say “do [goal of the event], I [what you did], then discuss impact. For impact, mention how many kids attended, age ranges, and anecdotes about their experiences. If you have a measurement of success (e.g., start vs end or the event tests), include the results.

Writing sequentially may not be compatible with telling a compelling story with main takeaways. Sometimes it’s valuable. See Rhetorical comments point #4 for more details.

Below, I outline a potential structure for the Personal Statement. There exist other good resources with information on what characteristics you want to convey to reviewers. The following is a clear structure for a Personal Statement. This version does not include a paragraph with a compelling personal story. See Rhetorical comments point #1 a mild rant of personal stories.

Paragraph 1: Career goals and summary of subsequent sections

This paragraph is super important: it frames the entire statement.

  • State your long-term career goal. Frame it in terms of intellectual merit and broader impacts. Ex: “My long-term career goal is to lead academic research to discover the next generation of robotic exoskeleton controllers to improve mobility in [population X].” Notice that the example hints at your potential long-term IM (exoskeleton controllers) and BI (improving mobility). You can also talk about teaching or your vision for broad societal impacts achieved through research/teaching/service.
  • Provide more detail if needed.
  • One-sentence personal motivation, if relevant.
  • Frame your decision to pursue a PhD as a stepping stone to achieve your career goal.
  • Summarize the characteristics that support your potential for success in research. These should match, in order, the topic sentences of the subsequent paragraphs.

Paragraph 2 (optional/flexible): Elaboration or compelling personal story that uniquely qualifies you for your proposed research trajectory

  • Opening statement of main takeaway (e.g., “Through my experience with X, I have a unique/rare perspective on Y that will help me innovate in my proposed career.”)
  • Provide support for the main takeaway.

Paragraph 3-X: Intellectual Merit

You can include an optional 1 paragraph summary of IM, but that probably isn’t needed. IM does not need to go sequentially unless the sequence strengthens the story. See Rhetorical Comments point 4. Each IM paragraph should have the following structure:

  • Main takeaway emphasizing an intellectual characteristic that supports your potential for success in research.
    • Supporting evidence for the main takeaway. For each example of supporting evidence, include:
      • Brief overview of the work (e.g., a project description)
      • What you did: These are your specific contributions, not the team’s achievements. Can be boldfaced if impressive.
      • What you found: Measurable outcomes like presentations, publications, and results of your specific contributions.
      • What it means (if relevant): What did science or your team gain from your efforts?

Note: There is flexibility in how you structure this and how much you describe a project. You don’t want to describe 3 projects in a single paragraph. Instead, it’s probably better to show how each project supports a different takeaway.

Paragraph (X+1)-Y: Broader Impacts

You can include an optional 1 paragraph summary of BI, but that probably isn’t needed. Similar to IM, BI does not need to go sequentially unless the sequence strengthens the story. See Rhetorical Comments point 4. Each BI paragraph should have the following structure:

  • Main takeaway emphasizing and societal impact that supports your potential for future contributions to society.
    • Supporting evidence for the main takeaway. For each example of supporting evidence, include:
      • Brief overview of the work (e.g., an outreach activity that you created/led)
      • What you did: These are your specific contributions, not the group’s or event’s achievements. Can be boldfaced if impressive.
      • What you found: Measurable outcomes like numbers of people impacted, demographics, measurable outcomes of the activity (e.g., pre- vs post-test), or feedback that you got from the activity.
      • What it means (if relevant and not redundant with “What you found”): What did society gain from your efforts?

Paragraphs (Y+1)-Z: Why a GRF would be impactful

These are like summary paragraph(s). However, you also want to convey clear thinking about how your choice of grad program and advisor (grad applicants) / target programs-advisors (undergrad applicants). Convey that you’ve thought critically about how a university/ lab/ advisor/ collaborators can help you meet your training goals and move you closer to your career goal upon graduation. Invoke the same IM and BI language that you used in the first paragraph of the statement.

Paragraph Y+1:

  • Main takeaway about your rationale for universities/labs/advisors where you are or are applying.
    • Supporting evidence for why this is a good choice.
      • Summary of the lab’s work / advisors expertise (connect these back to the main takeaway and possibly your research statement)
      • Summary of collaborators’ expertise
      • Summary of research resources provided by the university/lab/collaborators.

Note: For undergrads, highlight multiple schools/advisors to which you plan to apply and discuss broadly why they are appropriate for your proposed work and career goals.

Paragraph Z (may be able to merge with paragraph Y+1):

  • Main takeaway about why the NSF GRF is ideal for your career goals.
    • (optional) Statement showing that you understand the generic benefits of the GRF: identifying you as a potentially top researcher and provide decent financial support.
    • Supporting evidence for how the GRF would specifically support your work.
      • You can write something like “Without the GRF, I will not be able to pursue this important research work because….” This is appropriate if your work is important but hard to fund with grants or not entirely within existing grants that your PI has. You could also emphasize how it will help build a collaboration “The GRF will provide me with support to form a new collaboration between X and Y. Combining X and Y’s skillsets would enable me to do Z (research statement thing) that would have Q impact.“

Note: You want to be specific. If you don’t have a specific need beyond “it looks good on a CV and it’s stable funding,” you can lump this in as the last sentence of Paragraph Y+1.

Research Statement

The Research Statement needs to convey clear and self-consistent rationale, IM, BI, hypothesis, research questions, and methodology for each project.  You’ll be reviewed by engineers or neuroscientists, or… depending on which directorate you specify. However, reviewers will not probably be experts in your exact field. This means:

  • No jargon. Be sure to define any field-specific terms. Get feedback from people outside your field to help identify jargon.
  • Tell a compelling and self-consistent story. The relationship between each rationale point and your central hypothesis, research question(s), methods, IM, and BI needs to be super clear. Odds are it’ll be fine if you don’t cite an important paper in the field. The reviewers won’t always be able to assess the novelty and impact of your work, only that it’s novel and impactful given the information you provided. You should still propose novel research.

There is a lot of flexibility in the Research Statement structure. Some people include BI before getting into the research. Others include BI at the end. I will present one possible structure, but I encourage you to look for other examples.

If you are doing health-related research, focus on mechanistic discovery, NOT on (e.g.) evaluating a specific intervention (NIH domain). For example, here are two titles for the same proposal:

  • Uncovering modifiable mechanisms underlying individual differences in muscle strength
  • Evaluating the effects of foam rolling on isometric muscle strength in adolescents.

In the above examples (1) is appropriate for NSF: it focuses on a mechanism that could generalize across populations (= broader impact). Option (2) is an NIH study. It suggests that you know the mechanism and just need a little more data to validate it in a new population. Important, but less potential for broad impact across populations and applications.

If you use patient populations, you want to describe how the population represents an appropriate and necessary model system to answer the mechanistic question that you are asking.

Below is an example structure for a research statement:

Title: If health-related, focus on mechanisms

Rationale and proposed study ——————————————————————————

Most people title this section “Introduction.” Call it “Rationale” or “Motivation” and provide only information that supports the need and scientific rationale for your proposed research.

Paragraph 1: Grand challenge and broader state-of-the-art.

If doing health research, you may not want to invoke a specific patient population here.

  • Highlight a grand challenge that you will contribute to. (e.g., Mobility deficits impact X people in the US, but remain challenging to mitigate.)
  • (multiple sentences) Describe a more specific problem that you plan to address (e.g. Exoskeletons are useful for X, but exoskeleton controllers that robustly improve mobility across individuals have not been developed). Review the state-of-the-art in how the challenge has been addressed.
  • Highlight a fundamental gap in knowledge.

Paragraph 2 (optional): Specific detail on intellectual or methodological gaps

This is relevant if you are proposing both a gain in scientific knowledge and methodological innovation.

  • Repeat for each gap that you will address (1-2 sentences each):
    • Describe the state-of-the-art.
    • Describe the gap.

Paragraph 3: Proposed study

  • Describe the central (ideally mechanistic) hypothesis for the proposed study.
    • This hypothesis should span your proposed projects/aims.
    • Describe how you will test the hypothesis: “To test this hypothesis, I will…”
      • This should be one sentence that spans projects/aims.
      • For grad students, note here or in the prior paragraph(s) what you have accomplished towards the proposed work. You want to show that you are already productive.

Paragraph 4 (optional; recommended for grad students): Resources

This paragraph can also go at the bottom of the Statement and may make more sense if placed after the projects/aims as “Logistics” or “Resources”. The goal is to show that you have all of the resources that you need.

  • Describe the resources and expertise of your PI and collaborators
  • Describe additional university resources that could support your work.

For undergrads, you can describe how the schools/labs where you are applying have the resources that you need.

Note: Paragraphs 1-4 are lumped into a single paragraph.

Science  ——————————————————————————

Paragraph [5+(X-1)]-Y: Project/Aim X

Here, you want to convey your research clearly and concisely. You should have already provided bread rationale in the prior paragraphs. If needed, you can include ~1 sentence describing a specific gap, but you may not have space.

  • Title: Be specific and directional. Include the following information:
    • Population/mechanism (you will usually focus on an underlying mechanism, rather than a specific population).
    • Intervention (or experimental condition)
    • Outcome (what did you measure)
    • Comparisons (what comparison did you make)
    • The title should be directional, if relevant:
      • For example, a title could be “Determining if stretching-induced decreases in gamma drive increases isometric muscle strength.”
        • Your population is secondary to the mechanism, so you describe that later.
        • Intervention = stretching
        • Outcome  = gamma drive and strength
        • Comparison (implied) = gamma vs strength
        • Directionality: less gamma drive -> more strength.
    • Provide a prediction.
      • Central hypotheses are broader, mechanistic, and can be tested multiple ways. Predictions are specific and have an obvious test to refute the prediction. For example, “I predict that gamma drive will be negatively associated with muscle strength.” That’s clearly going to be a regression analysis.
      • Grad students: If you have supporting data, include that in a figure and refer to it.
      • Undergrads: You can create and refer to conceptual figures of what the “ideal” result would look like.
      • In the above example, the figure would be a regression plot of gamma vs strength.
    • Describe what you will do. See Rhetorical comments point 5 for examples of using “Why-then-what” structure.
    • Describe what you will measure and how you will measure it.
    • Describe your comparisons and how you will evaluate success. Evaluation of success should be related to your hypothesis. For example, “Associations between decreased altered gamma drive and increased strength would support gamma drive as a modifiable mechanism of muscle strength.“
      • Important A really cool study is one where you learn something regardless of the outcome. In that case, you could have a second sentence saying “A lack of association/opposite association would…“
      • You can also highlight contingencies in this section. What if your prediction is not supported? How does this affect subsequent projects? What next steps will it motivate? etc…
    • Also highlight experimental contingencies. For example, what if you can’t reliably measure gamma drive? How would you modify the experimental design?

Note: Ideally, every methodological decision is justified in the rationale section. That isn’t always possible in 2 pages. Focus on justifying methodological decisions that hinge on specific assumptions/hypotheses and could invalidate the results if those assumptions/hypotheses are bogus.

Paragraph Y+1: Intellectual Merit

Connect this back to your scientific rationale, research question(s), and central hypothesis. Don’t make up new stuff; just answer the challenges that you raised earlier.

  • How does providing evidence in support of/against your central hypothesis advance the field’s knowledge?
  • Specifically, how will your projects/aims advance knowledge (1 sentence each). This should answer the gaps that you identified in the rationale.
  • Are there any other secondary gains in knowledge?

Paragraph Y+2: Broader impacts

Connect this back to your grand challenge. How are you improving the world?

  • How does the completion of this work help society?
    • What are potential applications of the knowledge gained (clinical, military, industrial, recreational,…)?
    • What populations could benefit from the work?
    • How will this project help you directly impact society?
      • Will you spin it into an outreach event?
      • Will you mentor undergrads or high school students?
      • Will you use it to meet unmet needs in your community?
      • Etc…

Note: For BI, you want to be specific. Don’t just say “I will do outreach.” Say “I will create an outreach event in collaboration with [organization], which will use [aspect of your research] to teach [population] about [topic].“

References: Use shorthand; see example Research Statements.

Writing efficiently

This section assumes that you have a support group that can give you feedback. If not find people who are willing to review your documents!

  • Before writing, get feedback on the broad messages. Presenting your proposal as a slideshow or bulleted list of main ideas will save you a bunch of time. For the Personal Statement, you can also write the first 1-2 paragraphs and get feedback on that. Those paragraphs should contain a summary of the main points that you’re going to talk about. For the research statement, a presentation highlighting the main rationale points supporting the research,
    • In the Personal Statement, I highly recommend first writing a list of takeaways that you want the reviewer to…take away from reading your statement. I included this point in the Personal Statement section because it’s important. Those takeaway statements become the topic sentences of different paragraphs. Then the paragraphs are written to provide measurable evidence of the topic sentence.  List takeaways, get feedback on takeaways, provide supporting evidence for takeaways, and turn it into a paragraph.
      • Topic sentence example: “My productivity on an independent research project supports my potential to create and disseminate impactful research.”
      • Supporting evidence example: “I presented this work at X conference and am preparing a first-authored manuscript.” This statement should be boldfaced.
    • For the Research Statement, draft the title and rationale bullets, question, and project/aims first. Then make a figure representing the approach and expected outcomes. Get feedback on that. Once the rationale clearly supports the question and people understand what you’re doing from a brief description of the figure, you’re ready to write.

Rhetorical comments

These are some of the most common stylistic mistakes that I see when reviewing NSF GRFP applications. The items are in no particular order:

  • (almost) Never start the Personal Statement with a paragraph-long personal story. Do this ONLY if your personal story supports your (relatively) unique potential to succeed in your proposed research and career track. For example, having an amputation would be strong support for a proposed research career studying a specific challenge related to your imputation. Otherwise, your motivation/passion for a project is irrelevant. If you really want to description of your motivation, make it a single concise sentence in the first paragraph.
    • Avoid phrases like “I had the opportunity to…;” “X enabled me to…;” etc…   Stop being humble. Own your successes. Even if you feel like someone handed something to you, just say “I did X.” Humble descriptions waste reviewers’ time and make it sound like you had no active role in your success.
    • Get to the point. This issue is common in creative writing, where a paragraph builds up to a climactic conclusion. Human brains typically reasons in this order: build evidence, then synthesize the evidence to draw a conclusion. Scientific writing does the opposite. Get to the point first, then provide support. A relatively quick to identify the main point is to write a paragraph as it feels most natural, then find the sentence that conveys the most important takeaway from the paragraph. Make that the first sentence of the paragraph. Usually, the important takeaway is the last sentence of the paragraph. This is called “upside-down writing.” Move that sentence to the top of the paragraph and re-read.
    • In the personal statement, write sequentially ONLY when it supports your main takeaways. Don’t try to write sequentially if doing so isn’t critical for your main takeaways. It often doesn’t matter that you transitioned from lab X to lab Y or project X to project Y. If Lab X and Project Y best support Takeaway #1, lump them together. If the sequential nature of labs/projects/etc… supports a main takeaway, sequential writing is fine. For example, if you transitioned from Lab X to Lab Y to achieve a training goal, or if going from Project X to Project Y was a big step towards independence in research, keep the text sequential and connect to the takeaway. 
      • For example, you could write a topic sentence like “My focus on specific training goals has supported my rapid advancement toward independence as a researcher.“
      • The supporting evidence would be written as “To build skill Z, I switched from Lab X to Lab Y. In Lab Y, I quickly built independence in research, switching from my role of supporting experiments on Project X to proposing and beginning my own Project Y in only 6 months.“
    • Important Using “Why-then-what” sentence structure is valuable in both statements.  This structure makes it easy to follow rationale. Typically, such a sentence reads “To [accomplish/achieve/learn/…] X, I [did Y].” This will make it seem like you’ve carefully and proactively thought through training and/or research decisions. Compared to saying “I did X, which accomplished Y,”  the why-then-what structure is easier for the reviewer to understand.
    • Do NOT say “I aim to…” or “My goal was to…,” etc… Just say what you did. You can convey the same information more concisely using “why-then-what” sentence structure.

For example, instead of “My goal was to help toddlers learn calculus. To do that, I created a daycare activity where we estimated derivatives using Tonka trucks.” (Tonka trucks were toys that kids used to play with before they got iPads) try “To help toddlers learn calculus, I created a daycare activity where we estimated derivatives using Tonka trucks.” 

Scroll to Top