Chapter 17: NSF Grants — Where Discovery Meets Broader Impact

If the NIH (Chapter 16) is the giant of biomedical research, the National Science Foundation (NSF) is the giant of nearly everything else in fundamental science and engineering — mathematics, physics, chemistry, computing, geosciences, biology, the social and behavioral sciences, engineering, and STEM education. It funds a large share of the non-medical basic research done at American universities, and for researchers outside the biomedical world, the NSF is often the federal funder. It is also built on a distinctive idea that shapes everything about how you write for it: the NSF funds not just the advance of knowledge but the advance of knowledge in service of society, and it judges those two things as co-equal criteria.

That dual mandate is the heart of this chapter. The NSF's most famous, most distinctive, and most frequently mishandled feature is its two-criteria merit review: every proposal is judged on Intellectual Merit (does it advance knowledge?) and Broader Impacts (does it benefit society?), and — this is the part applicants resist — the two are both required and both weighted. An applicant who pours brilliance into the science and treats broader impacts as a paragraph of throat-clearing has misunderstood the NSF as deeply as the NIH applicant who picks the wrong mechanism. The universal craft of Part II still applies; so does the adaptation lesson of Chapter 16. What the NSF adds to your adaptation toolkit is the discipline of writing for two co-equal questions at once.

In this chapter we map the NSF's structure and its merit-review system, work through the two criteria and the proposal format that the PAPPG (the NSF rulebook) requires, walk through how panel review and the Excellent-to-Poor rating scale actually work, devote real attention to writing a Broader Impacts plan that earns its keep, and survey the NSF's mechanisms — including the early-career CAREER award that integrates research and education by design. Our anchor for the chapter is Dr. Daniel Cho, a composite early-career computer scientist pursuing a CAREER award, with Dr. Hernandez's NIH experience (Chapter 16) as a running contrast. As always: NSF policies and the PAPPG change regularly, so treat the specifics here as the durable shape of the system and verify current details at nsf.gov.

17.1 The NSF: Structure and Merit Review

The NSF is an independent federal agency that funds basic research and education in science and engineering, distributing on the order of \$9 billion a year — smaller than the NIH, but the dominant funder in its domains. Like the NIH, it is not a monolith: it is organized into directorates (broad areas such as Biological Sciences; Computer and Information Science and Engineering; Engineering; Geosciences; Mathematical and Physical Sciences; Social, Behavioral and Economic Sciences; STEM Education; and the newer Technology, Innovation and Partnerships directorate), each subdivided into divisions, each containing programs. The program is the unit that matters most to you: a specific program, run by one or more program officers (also called program directors), with its own scope, its own solicitations, and its own funding decisions.

🧩 Productive Struggle: Before reading on, predict: the NSF says it weighs Intellectual Merit and Broader Impacts equally, yet most applicants spend 95% of their effort on the science. If that's a mistake, why do so many strong scientists make it — and what would change if you took "co-equal" literally? Jot your guess. The answer, developed across this chapter, is that broader impacts feel like marketing to scientists trained to value only the science, so they under-invest — and the applicants who instead treat broader impacts as a genuine, planned, evaluated part of the work gain a real competitive edge precisely because so many competitors don't.

A crucial structural fact: at the NSF, program officers are powerful and central, and many are rotators — working scientists who serve at the NSF for a few years and then return to their institutions. Your program officer is the person who runs the panel, synthesizes its advice, and makes the funding recommendation. As Chapter 2 stressed and Chapter 16 reinforced for the NIH, contacting the program officer before you write — to confirm your project fits the program and to ask which program is the right home for it — is among the highest-value things you can do. At the NSF this matters even more than at the NIH, because the program officer has substantial discretion and there is no rigid percentile-and-payline formula forcing the decision.

Merit review is the NSF's term for its peer-review process, and the agency takes the phrase seriously: proposals are judged on merit, by experts, against the two published criteria. Understanding that the entire system is built to apply those two criteria — every reviewer instructed to address both, every panel summary structured around both — tells you how to write: address both, explicitly, where reviewers expect to find them.

A few mechanics of how programs actually work will sharpen your strategy. Many NSF programs accept proposals on a published cycle — some with annual or semiannual target dates or deadlines, others with accepted-anytime windows — so the first practical question is when your program reviews, which the solicitation states. Because many program officers are rotators (working scientists on temporary assignment), the person running your program may change between submissions, and the program's emphasis can shift with new leadership and new agency priorities — another reason to read the current solicitation and talk to the current program officer rather than relying on what a colleague did three years ago. And because the program officer assembles a portfolio within a finite budget, your proposal is evaluated not only on its own merits but against the shape of what the program is trying to build — which is why a strong proposal that is a poor fit for a particular program can be redirected, and why asking "is this the right program for my work?" before submitting is so valuable. None of this replaces merit; it contextualizes it. The NSF funds excellent, well-fit work that addresses both criteria, submitted to the right program at the right time.

📜 How We Got Here: The two-criteria system wasn't always so explicit. For much of its history the NSF emphasized scientific merit, and "broader impacts" as a formal, named, co-equal criterion was strengthened over time — partly in response to congressional and public pressure that publicly funded science should demonstrably benefit the public that pays for it. The result is the modern merit-review framework in which Intellectual Merit and Broader Impacts are the two explicit criteria reviewers must address. Understanding this history explains why broader impacts can feel "bolted on" to scientists — it was, institutionally, added emphasis — and why the NSF is so insistent that it be taken seriously: the criterion exists precisely because public investment in science is justified by public benefit. Resenting it as politics misreads it; the NSF means it, reviewers score it, and the public mandate behind it is real.

17.2 The Two Criteria: Intellectual Merit and Broader Impacts

Everything about writing for the NSF flows from its two criteria, so let's define them precisely.

Intellectual Merit is the criterion you'd expect: the potential of the proposed work to advance knowledge and understanding within its field or across fields. It encompasses the importance of the problem, the soundness and creativity of the approach, the qualifications of the team, and the adequacy of resources — much of the universal significance-and-approach case you learned to make in Chapters 8 and 9, now framed as "does this advance knowledge?" For a researcher, this is familiar ground.

Broader Impacts is the criterion that makes the NSF distinctive: the potential of the proposed work to benefit society and contribute to desired societal outcomes. This is deliberately broad. It can include — and the NSF publishes lists like this — improving STEM education and educator development; broadening the participation of groups underrepresented in science; building research infrastructure; increasing public scientific literacy and engagement; improving national security, economic competitiveness, or quality of life; and more. Critically, broader impacts can be achieved through the research itself (e.g., research that directly informs policy or technology), through activities directly tied to the project (e.g., training students, developing curricula, public outreach), or through a combination. The NSF does not prescribe which broader impacts you pursue; it requires that you pursue some, genuinely, with a plan.

📊 From the Field: The NSF publishes lists of the kinds of broader impacts it values, and they are worth knowing because they expand the menu beyond what most applicants imagine. They include: full participation of women, persons with disabilities, and underrepresented minorities in STEM; improved STEM education and educator development at all levels; increased public scientific literacy and public engagement with science; improved well-being of individuals in society; development of a diverse, globally competitive STEM workforce; increased partnerships between academia, industry, and others; improved national security; increased economic competitiveness; and enhanced research and education infrastructure. The breadth is the point: you are not forced into "do outreach at a science fair." You can achieve broader impacts through the research itself (e.g., research that informs policy, builds open infrastructure, or trains a diverse workforce), through dedicated activities (curricula, mentoring, public engagement), or both. The strategic move is to choose broader impacts that genuinely fit your work and your strengths from this wide menu — which makes them credible and likely to happen — rather than bolting on a generic activity unrelated to who you are and what you study.

🚪 Threshold Concept: At NSF, Broader Impacts is co-equal with Intellectual Merit. The NSF funds the integration of discovery and societal benefit, not discovery alone. This is the threshold idea of the chapter, and it is genuinely hard for research-trained applicants to internalize, because their entire professional formation says the science is what matters and everything else is decoration. But at the NSF, broader impacts is one of two criteria, reviewers are required to address it, and a proposal with excellent science and a weak, generic, unplanned broader-impacts section is a proposal with a glaring weakness on one of its two graded dimensions. Once you cross this threshold — once you genuinely believe broader impacts is half the proposal and plan it as carefully as your methods — your NSF proposals change shape, and your competitiveness rises, because most of your competitors never cross it.

💡 Key Insight: The most common NSF mistake is treating broader impacts as an afterthought — a vague promise to "disseminate findings" or "mentor a graduate student" tossed in at the end. Reviewers see this constantly, and they read it for exactly what it is: a checkbox. A strong broader-impacts plan, by contrast, is specific, genuinely connected to the project and the investigator's strengths, resourced in the budget, integrated into the timeline, and evaluated — you say what you'll do, who will benefit, how you'll know it worked, and why you (with your particular partnerships and skills) are positioned to do it. The gap between a checkbox and a plan is one of the largest sources of competitive advantage available to an NSF applicant, precisely because so many applicants leave it on the table.

Consider our anchor. Dr. Daniel Cho is an assistant professor of computer science proposing fundamental research on more reliable machine-learning systems — strong Intellectual Merit. For Broader Impacts, he doesn't write "results will be disseminated at conferences." He builds a genuine plan: he will partner with a youth coding program (modeled on the kind of community STEM program we've followed elsewhere in this book — 90 students across three sites) to develop and test accessible computing modules, broadening participation among students underrepresented in computing; he will integrate undergraduates into his research; and he will release open educational materials. Each element is specific, tied to his actual strengths and partnerships, costed in his budget, scheduled, and paired with a plan to evaluate whether it worked. That is a broader-impacts plan that a panel will reward — and it is the difference between Cho's proposal and the many technically-comparable proposals whose broader impacts are an afterthought.

🔄 Check Your Understanding: An applicant writes a superb 14-page Project Description on the science and a three-sentence broader-impacts paragraph promising to "share results widely and train a student." A reviewer rates the Intellectual Merit "Excellent." Why might the proposal still not be funded?

Answer

Because Broader Impacts is a co-equal criterion, not a formality. A three-sentence, generic broader-impacts plan is a clear weakness on one of the two dimensions reviewers must address — and in a competitive program where many proposals have excellent science, the broader-impacts plan is often where proposals are differentiated. Reviewers and the program officer can and do decline proposals with strong science and weak, unplanned, unevaluated broader impacts. The fix is to plan broader impacts as carefully as the science: specific activities, genuine connection to the work and the investigator, resources, timeline, and evaluation.

17.3 The PAPPG and the NSF Proposal Format

The NSF's rulebook is the Proposal & Award Policies & Procedures Guide (PAPPG) — the detailed, authoritative specification of what an NSF proposal must contain and how it must be formatted. It is the NSF equivalent of taking the funder's instructions literally (Chapter 15): the PAPPG governs page limits, fonts, required sections, and compliance, and proposals that violate it can be returned without review. Read the current PAPPG, and the specific solicitation for your program, before you write — the general PAPPG sets defaults that a particular solicitation can override.

Most NSF proposals are submitted through Research.gov (the NSF's system; some opportunities also use Grants.gov), and a typical research proposal includes these components:

• Project Summary (usually one page): a self-contained overview with three explicitly labeled parts — an Overview, a statement of Intellectual Merit, and a statement of Broader Impacts. The labeling is not optional stylistic advice; the NSF requires the two criteria to be addressed here under those headings, and a Project Summary that omits one can be returned. This is the NSF's analog to the abstract/specific-aims page (Chapters 5–6), and like them it is the most-read part of your proposal.
• Project Description (commonly up to 15 pages): the heart of the proposal — your significance, prior work, approach, and a section that must explicitly address broader impacts. This is where the universal approach craft of Chapter 9 lives, framed to NSF's criteria.
• References Cited, Biographical Sketches (in the NSF-specified format, often via SciENcv), Budget and Budget Justification (Chapters 11–12), Current and Pending (Other) Support, Facilities, Equipment and Other Resources, a Data Management Plan (Chapter 14), and, where applicable, a Mentoring Plan for postdocs and other required documents. Solicitations add or modify requirements, so build your checklist from the current solicitation.

To make the labeled Project Summary concrete, here is the shape of Cho's (paraphrased and condensed; a real one fills the page and is specific):

Overview. Modern machine-learning systems fail in ways that are hard to predict, limiting their safe use in high-stakes settings. This project develops methods to characterize and bound those failures, combining [the specific technical approach] with [the specific validation strategy]. The research will produce [specific advances], advancing the reliability of deployed ML systems.

Intellectual Merit. The project advances knowledge by [the specific contribution to the field] — addressing a gap that current methods leave open, with an approach whose feasibility is established by the PI's preliminary results [cited]. If successful, it changes how the field [specific intellectual payoff].

Broader Impacts. The project broadens participation in computing by partnering with a youth coding program serving students underrepresented in computing across three sites, co-developing and evaluating accessible computing modules; integrates undergraduates into the research; and releases open educational materials. An external evaluator will assess participation, learning gains, and materials adoption.

Notice three things. First, the three blocks are explicitly labeled — exactly as the PAPPG requires. Second, the Broader Impacts block is as specific and concrete as the Intellectual Merit block — named activities, named beneficiaries, an evaluation plan — not a vague promise. Third, all three are self-contained: a reviewer who reads only the summary understands the science, why it advances knowledge, and how it benefits society. That is what a compliant, competitive NSF Project Summary looks like.

📋 Template — NSF proposal checklist (verify against the current PAPPG/solicitation): Project Summary with labeled Overview / Intellectual Merit / Broader Impacts; Project Description (within page limit) addressing both criteria, with an explicit broader-impacts treatment; References Cited; NSF-format Biographical Sketches; Budget + Justification; Current and Pending Support; Facilities, Equipment and Other Resources; Data Management Plan; Mentoring Plan (if postdocs); Collaborators and Other Affiliations (COA) information; any solicitation-specific documents. Confirm format (fonts, margins, page limits) against the current PAPPG — violations can mean return without review.

⚠️ Common Pitfall: Forgetting that the Project Summary must explicitly label and address both Intellectual Merit and Broader Impacts. Applicants who write a flowing one-paragraph summary in the style of a journal abstract — with no labeled criteria sections — risk having the proposal returned without review, a pure compliance failure that wastes the whole effort. The fix is mechanical but essential: structure the Project Summary exactly as the PAPPG specifies, with the criteria addressed under their headings. This is the NSF version of Chapter 15's lesson that a noncompliant proposal is never read.

17.4 How NSF Review Works

Once submitted, your proposal is assigned to a program and reviewed under the merit-review process. The mechanics differ from the NIH in ways that change how you write.

The NSF uses external peer review, typically through a panel of experts convened to read and discuss a batch of proposals (some proposals also or instead receive ad hoc mail reviews from individual experts). Reviewers evaluate each proposal against both criteria and provide written comments and a rating. The NSF's rating scale is not the NIH's 1–9: reviewers rate proposals on a descriptive scale — Excellent, Very Good, Good, Fair, Poor — and the panel produces a panel summary capturing the collective assessment of strengths and weaknesses on both criteria. The panel may also sort proposals into broad categories (for example, "highly competitive," "competitive," "not competitive"), but it generally advises rather than dictates.

The decisive difference from the NIH: the program officer makes the funding recommendation, exercising real judgment. There is no rigid percentile-and-payline machine. The program officer weighs the reviews and panel summary, but also considers the program's portfolio — balance across subfields, institutions, geography, risk, and the agency's strategic priorities — and recommends which proposals to fund within the available budget. This means two things for you: first, the panel's enthusiasm matters enormously but is filtered through a person with discretion; second, fit with the program and its portfolio is a real factor, which is again why talking to the program officer beforehand is so valuable.

📊 From the Field: NSF funding rates vary widely by program but have often hovered in the rough range of one in four or one in five proposals — competitive, though the exact rate depends heavily on the program and year. As with the NIH, a decline is common and not a verdict on your worth as a scientist. But the NSF resubmission culture differs: a declined NSF proposal is generally not handled as a formal "A1" with a required response-to-reviewers (as at the NIH). You typically revise and submit again, often treated as a new proposal, sometimes with an optional "Response to Previous Reviews" depending on the program. The constant across both agencies: read your reviews carefully, strengthen the real weaknesses, and try again — persistence and responsiveness are rewarded everywhere in federal science funding.

Trace Cho's review to see it concretely (composite). His CAREER proposal goes to the relevant computing program. A panel of computing researchers reviews it against both criteria; most rate the Intellectual Merit "Very Good" to "Excellent," and — because he built a genuine, specific, evaluated plan — they rate the Broader Impacts strongly too, noting the youth-coding partnership and the integration of education with research as authentic rather than decorative. The panel summary flags one weakness: an ambitious timeline. The program officer, weighing strong reviews on both criteria, the program's interest in broadening participation in computing, and the available budget, recommends Cho's proposal for funding — with a conversation about scope. Notice what carried him: not science alone, but science and a broader-impacts plan the panel believed, evaluated by a program officer with discretion who valued the fit.

🔄 Check Your Understanding: The NSF rating scale runs Excellent–Very Good–Good–Fair–Poor, and a panel can sort proposals as "highly competitive," "competitive," or "not competitive." Yet a proposal rated highly by the panel is not guaranteed funding. Why not — and what does that imply for how you write?

Answer

Because the panel advises; the program officer makes the funding recommendation, weighing the reviews and the program's portfolio and budget. A highly competitive rating makes you fundable, but the program officer chooses among many fundable proposals to build a balanced portfolio within finite money. The implication: write not only to earn strong ratings on both criteria, but to fit the program — read the solicitation, talk to the program officer, and make clear how your work advances what the program is trying to build. Merit gets you into the fundable set; fit and the program officer's judgment often decide who in that set is funded.

🪞 Learning Check-In: Notice whether, reading Cho's story, part of you thinks the broader-impacts work is "not real research" or a distraction from the science. That reaction is the exact thing standing between many strong scientists and NSF funding. The NSF has decided — as a matter of public mandate — that societal benefit is co-equal with discovery. You don't have to love that to succeed within it; you have to take it seriously. The investigators who win NSF funding are often not those with the best science in some abstract sense, but those who pair strong science with a broader-impacts plan they actually believe in and can execute. Treat the criterion as real, and it becomes an advantage rather than a chore.

17.5 Broader Impacts Done Well

Because broader impacts is where proposals are so often won or lost, it deserves direct treatment. A strong broader-impacts plan has recognizable features.

It is specific and activity-based. Not "we will broaden participation," but "we will partner with [a named kind of program] to develop and deliver [specific activities] reaching [a defined group], with [named] partners." Specificity signals that you've actually thought it through and intend to do it.

It is genuinely connected to the project and to you. The best broader impacts grow out of the research itself and the investigator's real strengths and relationships — a computing researcher building computing-education modules, a field scientist engaging the community where they do fieldwork. Connection makes the plan credible and likely to happen; a bolt-on activity unrelated to your work or skills reads as exactly that.

It is resourced and scheduled. Real activities cost money and time. A broader-impacts plan that appears in the narrative but nowhere in the budget or timeline is not a plan; it's a wish. Build the activities into both (Chapters 11–12 and 9).

It is evaluated. Just as your research has methods and your project has an evaluation plan (Chapter 10), a strong broader-impacts plan says how you'll know it worked — participation numbers, learning gains, materials produced and adopted, partnerships sustained. Evaluation turns a promise into a commitment.

It often addresses broadening participation. The NSF cares deeply about increasing the participation of groups underrepresented in science and engineering, and a thoughtful, genuine plan to do so — not tokenism, but real engagement with real partners — is among the most valued broader impacts.

🎓 Going Deeper — broadening participation without tokenism: Broadening participation is the broader-impacts category applicants most want to claim and most often handle badly, so it rewards care. The failure mode is tokenism: a vague promise to "recruit diverse students" with no partner, no pipeline, no plan, and no evaluation — which reads to reviewers as exactly the gesture it is. The strong version has three features. First, real partners: you work with people and programs already connected to the communities you hope to reach (a minority-serving institution, a community program, a teachers' network), rather than assuming you can reach them alone. Second, genuine activities with a theory of why they work: not "we will inspire students," but specific, evidence-informed activities (mentored research experiences, co-developed curricula, sustained engagement rather than one-off events) with a rationale for why they actually broaden participation. Third, evaluation: you measure participation, experience, and outcomes, and you report honestly. Notice that this is the same rigor you'd apply to your research and to a project evaluation (Chapter 10) — which is the deeper point. The NSF treats broadening participation as serious work deserving serious design, and the applicants who approach it that way, rather than as a diversity statement to be performed, are the ones reviewers believe. If you cannot do it genuinely, choose a different broader-impacts category you can do genuinely; an authentic infrastructure or public-engagement impact beats a hollow participation promise.

🗣️ From the Review Panel: On the panel, broader impacts is where I see the clearest separation between proposals. The science is often comparably strong across the top proposals — these are good scientists. What differs is whether the broader-impacts plan is real. I can tell in a paragraph: a real plan names activities, partners, beneficiaries, resources, and how the applicant will know it worked; a checkbox plan says "results will be disseminated and a graduate student trained." When two proposals are scientifically close, the one with the genuine, evaluated broader-impacts plan wins, every time. Applicants who think we don't read it, or don't weigh it, are leaving the easiest points on the table.

🔍 Why Does This Work?: Why does a specific, evaluated broader-impacts plan beat a generic one so reliably? Because reviewers and program officers are assessing likelihood of real benefit, and specificity is the signal of intent. Anyone can promise to "broaden participation"; a plan that names the partner program, the activities, the number of students, and the evaluation method demonstrates that the applicant has done the work of planning and is likely to do the work of executing. It also lets the program officer defend the funding decision — they can point to concrete, evaluable societal benefits. A vague promise gives them nothing to point to. So specificity isn't just stylistic polish; it's the evidence on which the broader-impacts judgment actually rests.

📐 Project Checkpoint — If NSF-bound, build your two-criteria spine: If your project targets the NSF, (1) draft your Project Summary with the three labeled blocks — Overview, Intellectual Merit, Broader Impacts — and make sure each is genuinely substantive. (2) Write a real Broader Impacts plan: list specific activities, the beneficiaries, the partners, the budget lines and timeline that resource them, and how you'll evaluate whether they worked. (3) Confirm your Project Description explicitly addresses broader impacts (not only in the summary). (4) Identify and contact the program officer for the right program. Save it all in your "My Proposal" document. If your project is not NSF-bound, do step 2 anyway as a thought exercise — many funders increasingly ask about societal benefit, and the discipline of planning and evaluating impact transfers everywhere.

17.6 NSF Mechanisms and the CAREER Award

The NSF offers several proposal types and programs. The most common is the standard or continuing grant for a research project, submitted to a disciplinary program. Beyond that, several special mechanisms are worth knowing:

• CAREER (Faculty Early Career Development Program): the NSF's prestigious early-career award, for tenure-track (pre-tenure) faculty. Its defining feature embodies the chapter's theme: a CAREER proposal must integrate research and education, building a unified plan in which the research and the educational/broader-impacts activities reinforce each other. It is a multi-year award and a career-shaping one — the NSF analog, roughly, to the launching role the early R01 plays at the NIH. (Cho's mechanism.)
• EAGER (Early-concept Grants for Exploratory Research): for high-risk, high-reward exploratory work, akin in spirit to the NIH's exploratory R21.
• RAPID: for research with an urgent need for rapid response (e.g., studying a transient event), with expedited review.
• Conference/workshop grants: to support scientific meetings.
• Graduate Research Fellowship Program (GRFP): a prestigious fellowship for graduate students — the NSF's student-support analog (compare Sam's NIH F31 in Chapter 16), awarded to the student and portable across institutions.
• Postdoctoral fellowships in various directorates, and a range of center, infrastructure, and education programs (often large, multi-investigator).

🎓 Going Deeper — the CAREER award and integration: The CAREER award rewards genuine integration of research and education, and the proposals that win it treat that integration as a single coherent vision rather than two stapled-together plans. A weak CAREER proposal has excellent research and, separately, a generic education plan ("I'll develop a new course and mentor students"). A strong one shows how the research and the education feed each other: the research generates material and questions for new curricula or outreach; the educational activities recruit and train the next generation into the research area; the broadening-participation work draws new people into the field the research advances. For an early-career scientist, learning to articulate that integrated vision is the central CAREER skill — and it is, again, the chapter's threshold concept in its most concentrated form: at the NSF, the science and its broader benefit are not separable, and the CAREER award is the mechanism that says so most explicitly. (Verify current CAREER eligibility and rules in the active solicitation; they have specific limits on attempts and career stage.)

A brief word on the student and early-career path, since it parallels and contrasts with the NIH's. The NSF's Graduate Research Fellowship Program (GRFP) supports promising graduate students directly, and — unlike the NIH's F31, which funds a specific project under a specific mentor (Chapter 16) — the GRFP is awarded to the student and is portable: it follows the fellow across institutions and projects, funding the person rather than a defined research plan, and is judged heavily on the applicant's demonstrated potential and (true to NSF form) their broader-impacts trajectory. The contrast is instructive: the NIH funds a training project, the NSF a promising person. Both agencies also support postdoctoral researchers through fellowship programs in various areas. For an early-career trajectory, the rough NSF arc runs from GRFP (as a student) toward a first standard grant and the CAREER award (as faculty) — the NSF analog to the fellowship-to-R01 ascent of Chapter 16, with the same long-game logic: build a track record, fit the program, persist through declines, and treat each proposal as a move in a multi-year campaign rather than a single shot.

17.7 Strategy: Writing for Two Questions at Once

Pull the threads together into NSF strategy. Succeeding at the NSF means doing three things well. First, fit the program: identify the right program, read its solicitation, and talk to its program officer — fit and program-officer judgment matter even more here than at the NIH. Second, make the Intellectual Merit strong using the universal craft of Part II, framed as advancing knowledge. Third — and this is what separates NSF winners — make the Broader Impacts genuine, specific, resourced, and evaluated, treating it as co-equal with the science because the NSF does. The investigators who internalize that broader impacts is half the proposal, and who plan it as carefully as their methods, win disproportionately, because most of their competitors never cross that threshold.

✅ Best Practice: Contact the NSF program officer the right way. A program officer is busy and fields many inquiries, so make yours easy to answer: a short, professional email with a one-paragraph description of your project, the program you think it fits, and one or two specific questions — most usefully, "Is this a good fit for your program, and is there a better-suited program?" Attaching a one-page summary (project, intellectual merit, broader impacts) lets them give a real answer. The goal is not to pitch or to extract a promise — they cannot pre-approve you — but to confirm fit, learn the program's current priorities, and surface any mismatch before you spend weeks writing. Program officers genuinely want well-fit proposals in their programs, so a focused inquiry is welcome. Do this early, take their guidance seriously (including a redirect to another program), and you convert the program officer's central role and discretion from a black box into a source of strategic intelligence. This is the NSF expression of the program-officer relationship Chapter 2 urged and Chapter 16 reinforced — and at the NSF, with its judgment-driven decisions, it pays off most of all.

It helps to hold the NSF and NIH side by side, since many researchers will encounter both:

The deepest lesson is the one Part III keeps teaching: the universal proposal craft is your engine, and each funder is a different chassis. The NSF's chassis is defined by its two co-equal criteria. Master writing for both at once — discovery and societal benefit, in one coherent proposal — and you've learned the NSF. And because the broader-impacts discipline is increasingly mirrored by other funders asking what public good their dollars buy, the skill you build writing for the NSF's second criterion will repay you far beyond the NSF itself.

🔄 Check Your Understanding: In one sentence each, contrast (a) how the funding decision is made at the NIH versus the NSF, and (b) the single most distinctive thing an NSF proposal must do that an NIH proposal need not.

Answer

(a) At the NIH, the institute funds largely down a ranked list to a payline (with some program-officer discretion); at the NSF, the program officer makes the recommendation with real discretion, weighing reviews and the program's portfolio, with no rigid payline. (b) An NSF proposal must explicitly address Broader Impacts as a co-equal criterion — planned, resourced, and evaluated — whereas an NIH proposal, while it has its own components, does not center a co-equal societal-benefit criterion in the same way.

Spaced Review

Retrieve these from earlier chapters without looking back.

1. (From Chapter 16) How does the NIH separate review (study section) from funding (institute), and how is the NSF's decision structure different?
2. (From Chapter 9) How does the pitfalls-and-alternatives strategy apply to an NSF Project Description, given that reviewers assess the soundness of the approach under Intellectual Merit?
3. (From Chapter 14) How does a Broader Impacts plan relate to the dissemination and societal-benefit ideas you met in the sustainability and dissemination chapter?

Answers

1. At the NIH, CSR assigns a study section (which scores) and an institute (which funds), and the institute largely funds to a payline. At the NSF, a program's panel advises and the program officer recommends funding with real discretion, weighing the reviews and the program's portfolio — a more judgment-driven, less formulaic decision. 2. Naming your real risks and contingencies strengthens the Intellectual Merit case by showing reviewers a sound, feasible, self-aware approach (disclosed weakness beats discovered weakness); a Project Description that ignores obvious risks reads as naïve and lowers the merit rating. 3. Broader Impacts is, in part, dissemination-and-societal-benefit made into a graded criterion: the active knowledge translation, broadening of participation, and public benefit you met in Chapter 14 become, at the NSF, half of what you are formally judged on — so they must be planned and evaluated, not promised.

Chapter Summary

Key Takeaways

• The NSF (~\$9B/year) is the dominant funder of non-medical basic science, engineering, and STEM education, organized into directorates → divisions → programs, with powerful, often-rotating program officers running programs and making funding recommendations.
• NSF merit review judges every proposal on two co-equal criteria: Intellectual Merit (advancing knowledge) and Broader Impacts (benefiting society). Threshold concept: Broader Impacts is co-equal with the science — NSF funds discovery and societal benefit together, not discovery alone.
• The PAPPG governs format and required components. The Project Summary must explicitly label and address both criteria (omitting one can mean return without review); the Project Description (~15 pp) must also address broader impacts; submission is usually via Research.gov.
• Review is by panel (and/or ad hoc), rated Excellent–Poor with a panel summary; the program officer then recommends funding with real discretion, weighing reviews and the program's portfolio — no rigid payline.
• A strong Broader Impacts plan is specific, genuinely connected to the work and the investigator, resourced in the budget and timeline, and evaluated — and it is where competitive proposals are most often differentiated.
• The CAREER award is the early-career flagship, defined by the integration of research and education; other mechanisms include EAGER, RAPID, conference grants, GRFP (students), and postdoctoral fellowships.

Action Items

1. Identify the right program for your work and read its solicitation and the current PAPPG.
2. Contact the program officer to confirm fit before you invest in writing.
3. Draft a Project Summary with the three labeled blocks, and a Project Description that addresses both criteria.
4. Build a genuine Broader Impacts plan — specific activities, beneficiaries, partners, budget, timeline, evaluation.
5. If early-career, consider whether a CAREER award fits, and design an integrated research-and-education vision.

Common Mistakes

• Treating Broader Impacts as an afterthought — the single most common and most costly NSF error.
• Omitting the labeled criteria in the Project Summary (a compliance failure that can mean return without review).
• Letting broader impacts appear in the narrative but not in the budget or timeline — a wish, not a plan.
• Assuming a payline governs the decision; forgetting the program officer's discretion and the program's portfolio.
• Writing a CAREER proposal as two stapled plans (research + generic education) instead of an integrated vision.

Decision Framework — "Is the NSF my funder, and am I ready?"

1. Is my work fundamental science/engineering/STEM-education in an NSF domain (not biomedical, which is the NIH)? → If unsure, the program officer can tell you.
2. Which program fits, and what does its solicitation require? → Read it; build your checklist from it.
3. Is my Intellectual Merit strong (Part II craft, framed as advancing knowledge)? → If not, fix the science first.
4. Is my Broader Impacts plan genuine — specific, connected, resourced, evaluated? → If it's a checkbox, rebuild it; it's half the proposal.
5. Have I talked to the program officer and addressed both criteria explicitly where reviewers expect them? → If not, do so before submitting.

🔁 Carry this forward: You've now seen the two federal giants — the NIH's mechanism-and-payline machinery (Chapter 16) and the NSF's two-criteria merit review (this chapter). Both teach the same meta-skill: fit your universal proposal to the funder's specific machinery. Next we turn to foundations (Chapter 18), where the machinery is looser, the relationships warmer, and the "rules" often unwritten — a very different adaptation, and one where the cultivation lessons of Chapter 2 come fully into their own.