The National Science Foundation evaluates every proposal it accepts for review against two co-equal, National Science Board–approved merit review criteria: Intellectual Merit and Broader Impacts. Broader Impacts has a reputation among applicants as the vaguer, harder-to-satisfy half of that pair — and NSF’s own 2025 policy update changed what counts as an acceptable Broader Impacts plan in ways that are not yet reflected in a lot of older guidance still circulating online. This guide works entirely from NSF’s own Proposal & Award Policies & Procedures Guide (PAPPG), NSF’s published merit-review materials, and NSF’s April 2025 Statement of Priorities — not from claimed insider knowledge of what any specific panel discussed — because that documentation is itself specific enough to tell an applicant what a reviewer is instructed to look for.
Broader Impacts as one of NSF’s two merit review criteria
Since 1997, NSF has required every proposal it reviews to be evaluated against two criteria, applied together, not as alternatives. Per PAPPG Chapter III.A:
- Intellectual Merit — “the potential to advance knowledge” within a field or across fields.
- Broader Impacts — “the potential to benefit society and contribute to the achievement of specific, desired societal outcomes.”
These sit under three merit review principles the National Science Board has set for the whole system: NSF-funded projects should be of the highest quality and advance (if not transform) the frontiers of knowledge; NSF’s funded portfolio, in the aggregate, should contribute more broadly to societal goals; and assessment of broader impacts should use metrics appropriate to the scale of resources provided, rather than a one-size-fits-all evaluation. That last principle matters in practice: NSF does not expect a $50,000 planning grant to carry the same broader-impacts ambition as a $10 million center award, but it does expect the ambition to be proportionate and real, not decorative.
Both criteria are mandatory review elements — a proposal is not allowed to substitute strength in one for weakness in the other, and reviewers are specifically instructed to evaluate both. See CASRAI’s NSF Merit Review Panel entry for how the panel process that applies these criteria actually runs.
The five review elements NSF asks reviewers to apply
NSF does not leave “how to evaluate Broader Impacts” undefined. The PAPPG sets out five review elements that reviewers apply to both criteria — the only element that reads differently between the two is the first:
- Potential for the proposed activity to advance knowledge (Intellectual Merit) or benefit society or advance desired societal outcomes (Broader Impacts).
- Creativity and originality — the extent to which the proposed activities “suggest and explore creative, original, or potentially transformative concepts.”
- A well-reasoned plan — whether the plan for carrying out the proposed activities is “well-reasoned, well-organized, and based on a sound rationale,” including a mechanism to assess whether it succeeded.
- Qualifications — how well qualified the individual, team, or organization is to carry out the proposed activities.
- Adequate resources — whether the PI has access to adequate resources, at their home institution or through collaborations, to actually do what’s proposed.
Read together, this is the honest, primary-source answer to “what do reviewers score”: a Broader Impacts section is not evaluated on good intentions. It is evaluated on whether the proposed activity is specific and plausible, whether the plan for doing it is organized and includes a way to tell if it worked, whether the people doing it are actually positioned to do it, and whether the resources exist to carry it out. A statement that reads as an afterthought fails element 3 (no real plan, no assessment mechanism) even before a reviewer gets to judging the idea itself.
The categories NSF’s own guidance points to
NSF’s Broader Impacts guidance page lists examples of the kinds of societal outcomes a Broader Impacts plan can target — explicitly described as illustrative, not exhaustive or prescriptive:
- STEM education and educator development, at any level
- Public engagement with, and scientific literacy about, science and technology
- Societal well-being
- Development of a diverse, globally competitive STEM workforce
- Partnerships between academia, industry, and other sectors
- National security
- US economic competitiveness
- Research and education infrastructure
NSF is explicit that a project doesn’t need a bolted-on outreach activity to satisfy this criterion. Broader impacts can come from three different sources: the research activity itself (a new instrument or dataset that other labs can use), activities directly related to the research (training the graduate students and postdocs who do the work), or activities that are complementary to, but supported by, the research (a K-12 partnership that uses the project’s findings as teaching material). The strongest sections usually draw on more than one of these rather than treating Broader Impacts purely as a separate, unrelated add-on paragraph.
What changed in April 2025 — and why older guidance can now be wrong
On April 18, 2025, NSF Director Sethuraman Panchanathan issued a Statement of Priorities that substantively reframed what counts as an acceptable Broader Impacts activity, particularly around broadening participation. This is recent enough that a meaningful share of Broader Impacts advice still online — including guidance written before mid-2025 — describes an approach NSF has since moved away from. The substantive points, per NSF’s own statement:
- Applicants should “prioritize the first six broader impacts goals” defined by the America COMPETES Reauthorization Act of 2010 (things like a globally competitive STEM workforce, K-12 and undergraduate STEM education and teacher development, and public scientific literacy). Goal seven — expanding participation of women and individuals from underrepresented groups in STEM — is still a legitimate goal, but the statement narrows how it can now be pursued.
- Broadening-participation and outreach activities “must aim to create opportunities for all Americans everywhere” and, in the Director’s words, “should not preference some groups at the expense of others.” Activities framed as open and available to all Americans (for example, targeting by geography, socioeconomic status, or institution type) remain acceptable; activities that directly preference or exclude people based on a protected characteristic do not.
- NSF states it will not fund proposals whose “narrow impact [is] limited to subgroups of people based on protected class or characteristics,” and that Diversity, Equity, and Inclusion (DEI) frameworks are not an acceptable basis for project design.
- Research that focuses on a protected characteristic remains fundable when that characteristic is “intrinsic to the research question” (NSF’s own example: disability-focused assistive-technology research) and the work otherwise aligns with agency priorities — the restriction is on how broader-impacts activities are targeted, not a blanket ban on studying underrepresented populations.
- Legally mandated programs with broadening-participation elements, and partnerships and collaborations across different institution types, continue to be supported.
The practical implication for anyone drafting a Broader Impacts section in 2026: a plan built around, for example, “a workshop series exclusively for [a specific demographic group]” needs to be reframed — either opened to a broader population while still reaching the intended audience through outreach and recruitment, or restructured around a non-protected-characteristic axis (institution type, geography, first-generation status, socioeconomic background) that NSF’s statement treats as still acceptable. Verify current wording directly against NSF’s Statement of Priorities and the specific solicitation before finalizing a plan, since this is an active, still-evolving area of NSF policy rather than a settled one.
What actually distinguishes a strong Broader Impacts section from a weak one
Combining the five review elements above with NSF’s own framing, a Broader Impacts section that clears review consistently does the following; one that struggles consistently fails one or more of them:
- Specific, not aspirational. “Will inspire the next generation of scientists” is not evaluable against element 3 (a well-reasoned plan). A named activity, a defined audience, and a stated scale (“a two-week summer research module for 15 community-college students, run in years 2 and 4”) is.
- Has an assessment mechanism. PAPPG explicitly asks whether the plan includes “a mechanism to assess success.” A strong section states what will be measured, how, and by whom — even briefly — rather than asserting an outcome with no way to check it.
- Connected to the research, not adjacent to it. A generic outreach activity unconnected to the actual science is weaker, under NSF’s own three-pathway framing above, than a plan that extends the specific research (using the project’s own methods, data, or infrastructure) to benefit a defined community.
- Matched to actual resources and qualifications. Elements 4 and 5 ask whether the team is positioned to do what it proposes and has the resources to do it. A solo PI with no education background proposing a large K-12 curriculum-development effort, with no named partner and no budget line for it, reads as unrealistic against these elements regardless of how good the idea sounds.
- Compliant with the current Statement of Priorities. As of 2026, a plan built around excluding or exclusively targeting a group by protected characteristic is a real risk factor, independent of how well-designed the activity otherwise is — see the section above.
- Scaled to the award. Per the third merit review principle, reviewers are told to weigh “the likely correlation between the effect of broader impacts and the resources provided” — an outsized broader-impacts promise attached to a small award can read as implausible rather than impressive.
Where Broader Impacts fits in an NSF proposal
Broader Impacts isn’t a single free-standing document. NSF requires it to appear in more than one place in a standard proposal: a dedicated statement in the one-page Project Summary (alongside a separate Intellectual Merit statement), and a substantive discussion within the Project Description itself, where the plan actually gets explained in enough detail for a reviewer to evaluate it against the five elements above. See CASRAI’s How to Apply for an NSF Grant guide for where these pieces sit in the full submission process, and NSF Grants: An Overview for how Broader Impacts fits alongside NSF’s other proposal components. Some mechanisms tighten this further: the NSF CAREER award, for example, requires broader impacts to be integrated with a dedicated education-activities component across the full award period, not treated as a separable add-on.
Broader Impacts vs. Broadening Participation vs. a diversity statement
These three terms get conflated often enough in proposal-development offices that it’s worth stating the distinctions plainly. Broader Impacts is the merit review criterion covered in this guide — a required element of every NSF proposal, covering any of the eight categories above. Broadening Participation is NSF’s own, narrower name for its coordinated portfolio of programs and goals specifically aimed at increasing STEM participation among underrepresented individuals, institutions, and regions — one possible way to satisfy the Broader Impacts criterion, not a synonym for it; see CASRAI’s NSF Broadening Participation entry for the full distinction. A diversity statement, by contrast, is a different document entirely, most often requested in faculty hiring rather than grant review; see CASRAI’s Diversity Statements for Researchers guide for how that document differs in purpose and audience even though the underlying skill — specific past activity and feasible future plans, not general statements of value — overlaps.
Frequently asked questions
Is Broader Impacts weighted equally with Intellectual Merit?
Yes. NSF’s own language describes them as co-equal criteria, both mandatory, both evaluated against the same five review elements. A proposal cannot compensate for a weak Broader Impacts section with an unusually strong Intellectual Merit case, or vice versa — panel discussion and individual reviews are expected to address both.
Does my research need a separate outreach activity to satisfy Broader Impacts?
No. NSF explicitly allows broader impacts to be achieved through the research activity itself, through activities directly related to it (such as training the students and postdocs who do the work), or through complementary activities the research supports. A separate, disconnected outreach program is only one of three legitimate paths, and not automatically the strongest one.
What actually changed with NSF’s April 2025 Statement of Priorities?
NSF reframed how broadening-participation and outreach activities can be targeted: they must be open to all Americans rather than preferencing or excluding people by protected characteristic, and DEI frameworks are no longer an acceptable basis for project design. Research where a protected characteristic is intrinsic to the research question itself remains fundable. This is a live policy area — verify current wording against NSF’s own Statement of Priorities and the specific program solicitation before finalizing a plan.
Does every NSF proposal need a Broader Impacts section?
Every proposal NSF accepts for merit review is evaluated against both criteria per PAPPG Chapter III.A, so in practice yes for standard research proposals; consult the specific program solicitation for any mechanism-specific variation in how the section should be structured or scaled.
How specific does the assessment plan need to be?
PAPPG asks whether the plan includes “a mechanism to assess success” — it does not mandate a particular research-design rigor (a randomized controlled trial is not required for an outreach activity, for example). What matters is that the applicant states what will be measured, roughly how, and by whom, rather than asserting an outcome with no way to check it.
