Every year, Morgan State reports its research and development (R&D) spending to the National Science Foundation. This framework turns "is this R&D?" from a judgment call into a repeatable, defensible answer — five levels of evidence, one consistent process.
Start here — before any of the mechanics, this is the situation the framework exists to fix.
Someone in the Office of Research Administration is looking at a sponsored award. Somewhere, a form or a spreadsheet needs a single answer: is this Research & Development, or isn't it? That answer feeds directly into what Morgan State reports to the National Science Foundation (NSF) every year through the Higher Education Research and Development (HERD) Survey.
Historically, that answer has come from judgment — whichever agency funded the award, or whatever was decided the year before. Not every dollar is obviously R&D. Some awards clearly are. Some are training, or service, or somewhere in between. If two different people classify the same award differently, the number Morgan State reports wobbles, and it becomes very hard to defend under audit.
This framework replaces "go interpret it yourself" with a repeatable process: tell it what evidence you have, and it tells you where that evidence lands — and how confident you should be in the answer.
The HERD Survey isn't paperwork. It's how Morgan State's research capacity gets counted.
NIH, NSF, and other federal agencies use HERD data in their allocation models. Higher reported R&D expenditures open access to funding programs — including some aimed at institutions still building research infrastructure.
HERD data feeds NSF's own R&D expenditure rankings and the Carnegie Classification® system. Morgan State's Doctoral University standing depends in part on accurately capturing the full scope of R&D activity.
Facilities and Administrative (F&A) cost rates are calculated in part from R&D expenditure. Misclassifying R&D shrinks that base — with direct financial consequences.
Because HERD data supports federal reporting obligations, misclassification is a risk in either direction. Pointing to a named document instead of a hunch is the strongest defense.
Where the HERD Survey comes from, and how classification practice has evolved.
The HERD Survey is run annually by NSF's National Center for Science and Engineering Statistics (NCSES) and is the primary national source of data on R&D spending at U.S. colleges and universities — collecting data from more than 900 institutions every year. Any institution spending at least $150,000 on separately budgeted R&D must report.
The definitions behind the survey trace back to the Frascati Manual, first published by the Organisation for Economic Co-operation and Development (OECD) in 1963 and revised many times since (most recently in 2015), as the international standard for measuring R&D. NSF's HERD Survey — and its predecessor surveys dating to the 1970s — adopted that definition to keep U.S. reporting comparable across institutions and, internationally, across countries.
What has not historically existed at most institutions, Morgan State included, is a documented, repeatable method for deciding whether an individual award meets that definition.
Classification by memory, sponsor name, or "however we did it last year" — undocumented, hard to defend when someone asks why.
A five-level, document-anchored process — every classification traceable to a specific, retrievable source.
The Frascati Manual (OECD, 2015) definition, as NSF applies it to higher education.
Creative and systematic work undertaken to increase the stock of knowledge, and to devise new applications of that knowledge. — Frascati Manual, OECD (2015)
Pursuing new knowledge with no specific application in mind.
e.g. theoretical physics, fundamental genetics, mathematical modeling
Investigation aimed at a specific practical objective.
e.g. health disparities, cybersecurity, climate adaptation
Using research knowledge to develop or improve a product or process.
e.g. diagnostic tool prototypes built from research findings
Key test: is the primary purpose of the award to generate new knowledge? If yes, it's R&D. If no, it isn't — regardless of which agency funded it.
The following are generally not R&D, even when funded by an agency known primarily for research: workforce development and training programs; service delivery contracts; equipment and infrastructure grants (unless the equipment is integral to a specific research project); educational curriculum development (unless the curriculum itself is the object of study); institutional capacity-building without a defined research objective; and conference or dissemination grants, unless the funding agency itself classifies the conference as R&D.
Many awards blend activities. When they do, the primary purpose stated in the Notice of Award and funding opportunity governs the classification.
What the five levels do for us — and why climbing toward Level 1 is the goal.
This framework is built around a hierarchy of evidence. The central principle is simple: the strongest evidence for whether a sponsored award constitutes R&D is the award document itself. When the award is clear, classification begins and ends there. When it isn't, the reviewer works systematically down the ladder until a supported determination can be made. This hierarchical approach has two practical advantages: it is efficient — reviewers stop at the first level that gives a conclusive answer — and it is defensible — the classification is always traceable to a specific, named source that can be produced under audit.
Office of Research Funds Accounting records, prior HERD listing, fund account type. Last resort — used only when nothing stronger is accessible.
Internal Routing Form (IRF), proposal or Statement of Work (SOW), Letter of Intent (LOI), or a written PI statement.
The federal program's listing on SAM.gov — program-level activity type, used when the award itself is silent.
The solicitation the award was made under — defines the program's research intent up front.
The award document itself — the strongest and most direct evidence of what the award funds.
A grant classified at Level 1 has the strongest evidence base. A grant classified at Level 5 has the weakest — not because the classification is wrong, but because the documentation behind it is more indirect. Tip: the ladder highlights live as you work through the classifier below.
Important: Federal Notices of Award are required under Uniform Guidance (2 CFR Part 200) to directly indicate whether an award is research. Before doing any inferential reading of the NOA's language, look first for that direct indication — NSF, NIH, and NASA notices all answer the "Is this R&D?" question explicitly. Keyword-reading the NOA is a fallback for when that direct answer isn't present, not the first move.
Why a document hierarchy is, in effect, a measurement instrument — and what makes one trustworthy.
In research methodology, a good measurement has two separate properties, and it's easy to have one without the other:
Tightly clustered — but off-target. Consistent answers that are consistently wrong.
Tightly clustered on target. This is what the five-level ladder is built to produce.
Would two different people, looking at the same award, reach the same classification? A reliable process doesn't depend on who happens to be reviewing it.
Is the classification actually true to what NSF and the Frascati Manual mean by "R&D"? A process can be perfectly consistent and still be consistently wrong — reliable without being valid.
The five-level ladder improves both at once. It improves reliability because every reviewer starts at the same place (the NOA) and works down the same sequence in the same order — the outcome depends on the evidence, not on individual habit or guesswork. It improves validity because each level is anchored to a document authored by the funding agency itself — closer to the federal government's own definition of the award — rather than institutional memory, a sponsor-name shortcut, or an assumption carried forward from a prior year.
In short: the ladder isn't paperwork for its own sake. It's what turns "is this R&D?" from a judgment call into something closer to a scientific measurement — one where the answer holds up no matter who is asking.
A classification that depends on who happens to be reviewing it isn't a classification — it's a guess with paperwork.
Tell it what you've got. It tells you where the classification lands, and how strong the evidence is.
Do you have the Notice of Award (NOA) — the award document from the sponsor?
Does the NOA directly state whether this is research? Federal NOAs are required under Uniform Guidance to answer this — check for an explicit "Is this R&D?" or activity-type field before anything else.
Look at the purpose language in the NOA itself. Does it describe research, study, investigation, hypothesis-driven work, or new-knowledge generation — or service delivery, training, outreach, technical assistance, or infrastructure?
Do you have the Funding Opportunity Announcement (FOA / RFA) the award was made under?
Does the solicitation describe the program as funding research or knowledge generation — or does the agency legislate the whole program as R&D, as NSF generally does?
Look up the ALN / CFDA number at sam.gov/fal. Which best describes the program?
Check the Internal Routing Form (IRF) or proposal/Statement of Work. Is "Research & Development" checked (vs. Instruction/Training), with a hypothesis, data collection and analysis, or a stated plan to publish?
Use the institutional record — Office of Research Funds Accounting records, prior HERD listing, or fund account type designation — as a last resort.
Every acronym used above, in one place. Search or filter by type.
Higher Education Research and Development Survey — NSF's annual survey measuring R&D expenditures at U.S. colleges and universities.
National Science Foundation — administers the HERD Survey and funds research across science and engineering.
National Center for Science and Engineering Statistics — the statistical unit within NSF that runs the HERD Survey.
Organisation for Economic Co-operation and Development — publisher of the Frascati Manual, the international standard defining R&D.
Notice of Award — the official document a sponsor issues to formally grant an award, including its purpose, amount, and terms.
Funding Opportunity Announcement / Request for Applications — the public solicitation under which an award is made (NIH uses RFA/RFP; NSF and DOE use FOA).
Also abbreviated "RFA" — a different meaning from "Request for Applications" above. The internal office whose records form the Level 5 evidence source when nothing stronger is available.
Assistance Listings Number / Catalog of Federal Domestic Assistance — a five-digit federal program identifier (XX.XXX). Renamed CFDA→ALN in 2021 under the DATA Act. Looked up at sam.gov/fal.
Internal Routing Form — the internal form a PI completes when submitting a proposal, including an activity-type declaration.
Statement of Work — the section of an award or proposal describing the specific work to be performed.
Letter of Intent — an early document sent to a sponsor signaling intent to apply, sometimes describing project purpose.
Facilities and Administrative costs — indirect costs recovered on sponsored awards; the rate is calculated in part from R&D expenditure.
Department of Energy — funds scientific research, largely through its Office of Science.
Department of Defense — funds both basic/applied research and non-research service contracts.
Department of Health and Human Services — parent agency of NIH; also funds many service- and training-oriented programs.
National Institutes of Health — federal biomedical research agency within DHHS; funds a mix of research and training programs.
Historically Black College or University — an institution established before 1964 with the principal mission of educating Black Americans.
Minority-Serving Institution — a federal designation for institutions enrolling significant proportions of minority students.
2 CFR Part 200 — the federal regulations governing grants administration, including the requirement that federal NOAs indicate R&D status directly.
The framework used to categorize U.S. colleges and universities by research activity, among other factors.
This isn't a gap unique to any one type of institution — it's common across research-active universities generally.
Most research-intensive institutions do not have procedural documentation like this for HERD classification — the practice is informal almost everywhere. Common shortcuts include classifying by sponsor name alone (NSF or NIH = automatically R&D), relying on PI self-classification with no independent check, or simply carrying forward the prior year's listing without review. Each approach is fast, and each is prone to error in both directions.
Morgan State's five-level framework — every determination tied to a named, retrievable document — goes beyond what most peer institutions have formally published. Organizations including NCURA (National Council of University Research Administrators), AURA (Association of University Research Administrators), and SRAI (Society of Research Administrators International) have all published guidance emphasizing award-level documentation over sponsor-based shortcuts; this framework is a concrete, adoptable methodology in that spirit.
Habits that keep the framework working year over year.