1 In this report, S&E fields include agricultural sciences; biological and biomedical sciences; health sciences; physical sciences; geosciences, atmospheric sciences, and ocean sciences; mathematics and statistics; computer and information sciences; psychology; social sciences; and engineering.

2 Comparisons over more than 1 year are made in inflation-adjusted constant 2012 dollars using gross domestic product (GDP)-implicit price deflators based on the calendar year. GDP deflators come from the U.S. Bureau of Economic Analysis and are available at https://www.bea.gov/national, accessed September 2022. The totals presented from the HERD Survey differ from similar totals reported in NCSES’s National Patterns of R&D Resources data and in the forthcoming Indicators 2024 report “Research and Development: U.S. Trends and International Comparisons.” In the HERD Survey data, spending from pass-through funds (funding of subawards from one university to another) is reported as spending by the universities initially receiving the money and as spending by the universities to which the funds are passed. The other sources remove such pass-throughs to avoid double-counting, such that academic R&D expenditures reported in National Patterns in 2021 are about $84 billion in current dollars, compared with about $90 billion reported in the HERD Survey data (NCSES NP 2021: Table 2; NCSES HERD 2021: Table 3). For more details, see the Methodology notes for the National Patterns of R&D Resources data, available at https://ncses.nsf.gov/pubs/nsf23321#technical-notes_methodology. These other sources also present calendar year approximations based on fiscal year data, rather than the academic year data reported in the HERD Survey.

3 For definitions of R&D and these categories, see OECD (2015).

4 “In the March 2023 edition of the Main Science and Technology Indicators (MSTI), the OECD decided to put under review data for several R&D indicators for the People’s Republic of China (hereafter ‘China’) for the period 2019 to 2021, suppressing the publication of several headline indicators until questions on the coherence of R&D expenditure and personnel data have been effectively addressed” (OECD 2023c:1).

5 In 2022, in response to Russia’s invasion of Ukraine, OECD ended the collection of statistical data from Russia. Therefore, 2020 is the latest year with R&D spending reports for Russia. See “Box 1: Main coverage changes and major revisions to R&D data in the MSTI March 2023 edition” in OECD (2023b:3).

6 Unless otherwise noted, all of the data and indicators in this section are based on the NCSES HERD Survey of spending in 2021 that was published in 2022. For the HERD Survey, NSF asks institutions of higher education that spent at least $150,000 in annual R&D to report the amounts they spent for separately budgeted R&D activities as well as the sources of those funds. Financial data are in constant 2012 dollars and are reported for the respondents’ most recent fiscal years.

7 Spending funded from “institutional sources” as reported in the HERD Survey includes institutionally financed research, cost sharing other than unrecovered indirect costs, and unrecovered indirect costs. Institutional sources do not include funding from general state appropriations spent on R&D (which are accounted as state and local government support) or philanthropic grants designated for research (accounted as support from nonprofit organizations). This amount does include funds spent from an institution’s own affiliated foundation directed at research activities, encompassing start-up packages, bridge funding, and seed funding for faculty; competitively awarded internal research grants; other departmental grants designated for research; and tuition assistance for students working on organized research. Spending on research administration and similar support services is considered an indirect cost and therefore is reported as part of federal funding support.

8 The HERD Survey asks institutions for data on separately budgeted R&D expenditures, in part to avoid the conceptual and measurement challenges in compiling data on expenditures for research activities carried out as part of the ordinary expectations of faculty members. Separately budgeted R&D generally includes sponsored R&D, the costs of which are paid for by gifts, grants, and contracts, as well as institutions’ own support of R&D through budgeted accounts set aside for that purpose.

9 The total number of degree-granting institutions is reported by the National Center for Education Statistics (see De Brey, Zhang, and Duffy [2022:Table 317.40]). The HERD Survey is distributed to two populations: institutions with more than $1 million in R&D expenditures per year and those with at least $150,000 but less than $1 million in R&D expenditures per year. The number of institutions in each population is found in NCSES (HERD 2021: Table A-3).

10 The Carnegie Classification of Institutions of Higher Education (https://carnegieclassifications.acenet.edu/) is widely used to characterize differences in academic institutions. The Basic Classification categorizes academic institutions primarily based on highest degree conferred, level of degree production, and research activity. This report identifies the 131 doctoral institutions with very high research activity based on the Carnegie Classification as of 2018, as reported in the 2018 Data File of Carnegie Classifications for each institution, accessed from the American Council on Education (https://carnegieclassifications.acenet.edu/wp-content/uploads/2023/03/CCIHE2018-PublicDataFile.xlsx). This categorization does not include some academic institutions that are top R&D performers but whose training programs are exclusively focused on a small number of fields (e.g., institutions focused exclusively on biomedical research and training). The relative share of academic R&D expenditures represented by these 131 institutions is calculated by summing the institutional responses reported in NCSES (HERD 2021: Table 21).

11 In this report, S&E fields include the health fields.

12 For the list of institutions qualifying as MSIs, by designation, see the NASA Minority Serving Institutions Exchange (https://msiexchange.nasa.gov/institutions).

13 The HERD Survey collects data from two populations. Institutions with at least $1 million in annual R&D spending respond to the standard-form HERD Survey, whereas those with at least $150,000 but less than $1 million in annual R&D spending respond to the short-form HERD Survey. The totals presented here include spending by HBCUs in the standard-form and short-form populations. Because of the $150,000 spending threshold, some HBCUs reported R&D spending for only a subset of the years from 2012 through 2021.

14 HHE institutions are defined by the U.S. Department of Education as nonprofit public and private institutions of higher education whose FTE enrollment of undergraduate students is at least 25% Hispanic, according to data that institutions reported in the Integrated Postsecondary Education Data System, conducted by the National Center for Education Statistics. For more detail on HHE institutions, see the Indicators 2022 report “Higher Education in Science and Engineering.”

15 The list of institutions currently designated as TCUs can be found on the NASA Minority Serving Institutions Exchange (https://msiexchange.nasa.gov/institutions?i=75&qo[0]=is&qp[0]=categories__name&qv[0]=Tribal%20College%20or%20University). Those reporting over $150,000 but less than $1 million in R&D expenditures (responding to the HERD Survey short form) can be identified from NCSES HERD 2021: Table 82. Those with R&D expenditures over $1 million in 2021 can be identified from NCSES HERD 2021: Table 5.

16 For a contemporary view of the nature and importance of external research infrastructures, see National Science and Technology Council (2021) and the NSB statement in its Vision 2030 report (NSB 2020).

17 Data releases and the accompanying InfoBriefs for the previous cycles of the Survey of Science and Engineering Research Facilities are available at https://www.nsf.gov/statistics/srvyfacilities/.

18 Data on human resources in academic R&D are drawn primarily from three sources: the NCSES Survey of Graduate Students and Postdoctorates in Science and Engineering (GSS), the NCSES Survey of Earned Doctorates (SED), and the NCSES Survey of Doctorate Recipients (SDR).

19 Data on the academic workforce are drawn primarily from the NCSES SDR, which uses a sampling method to measure trends in doctorate holders in science, engineering, and health fields. In this report, these fields are noted collectively as S&E.

20 Data in this section on human resources are based on statistical samples of relevant populations, whereas the R&D spending data in previous sections are based on a census of the full population of relevant academic institutions.

21 Data are from a special tabulation of the 2021 SDR. For longer time trends, see Figure LBR-18 in the Indicators 2022 report “The STEM Labor Force of Today: Scientists, Engineers, and Skilled Technical Workers.”

22 For longer time trends, see the Indicators 2022 report “The STEM Labor Force of Today: Scientists, Engineers, and Skilled Technical Workers.”

23 Fields shown in Figure URD-19 are those in which the differences observed were statistically significant.

24 The GSS is an annual census of all U.S. academic institutions granting research-based master’s degrees or doctorates in science, engineering, and health fields as of the fall of the survey year. In this report, science, engineering, and health are noted collectively as S&E.

25 Comprehensive data are not available on the number of undergraduate students engaged in research or on the sources or amounts of financial support for those so involved.

26 Additional aspects of graduate study, including degrees by field and debt, are available in the Indicators 2020 report “Higher Education in Science and Engineering.” Likewise, other aspects of the postdoctoral labor force, including salaries and demographics, can be found in the Indicators 2022 report “The STEM Labor Force of Today: Scientists, Engineers, and Skilled Technical Workers.”

27 In 2021, these were the most common funding mechanisms for S&E doctoral students: 40% received RAs, 24% TAs, 16% fellowships, 9% self-support, 4% traineeships, and 7% other types of support (NCSES GSS 2021: Table 3-5). In 2021, large majorities (73% and above) of doctorate recipients in most S&E fields reported holding no debt related to their graduate education (NCSES SED 2021: Figure 17). In social sciences, health sciences, and in non-S&E fields, the proportion of doctorate recipients with no debt ranged between 53% and 63%; in psychology, it was 48%.

28 This report discusses sources and mechanisms of graduate student funding. Funding sources include federal, institutional, and personal or self-support, among others. Personal sources include loans (including federal loans) or personal or family financial contributions. Funding mechanisms include assistantships, fellowships, and traineeships, among others.

29 Data are from a special tabulation of NCSES (GSS 2021).

30 The data tables from 2009 do not separate nonfederal sources of support for postdocs, so a comparison cannot be provided.

31 These calculations exclude appointments in natural resources and conservation and in multidisciplinary and interdisciplinary studies.

32 The GSS began to separate the data by level (master’s or doctorate) in 2017.

33 Data are from a special tabulation of NCSES (GSS 2021).

34 Data are from a special tabulation of NCSES (GSS 2021).

35 For GSS, “institutional” support includes support from academic institutions and from state and local government.

36 Data are from a special tabulation of NCSES (GSS 2021).

37 Data are from a special tabulation of NCSES (GSS 2021).