1. 1 The standard definition of the term labor force is a subset of the population that includes both those who are employed and those who are not working but seeking work (unemployed); other individuals are not considered to be in the labor force. Unless otherwise noted, when data refer only to employed persons, the term workforce is used. For data on unemployment rates by occupation, calculations assume that unemployed individuals are seeking further employment in their most recent occupation.

  2. 2 NCSES has developed an STW initiative to measure and understand the STW. One of the activities of the initiative is to develop a new federal survey called the National Training, Education, and Workforce Survey, which is currently being planned with a target data collection start in 2022. For more details on the initiative, visit

  3. 3 These five S&E broad occupational categories are divided into finer occupational categories (Table SLBR-1). Examples of workers in S&E occupations are software engineers, agricultural and food scientists, physicists, economists, psychologists, and mechanical engineers. These occupations are identified in the NCSES taxonomy of occupations (see NSCG 2017: Technical Table A-1 for taxonomy) and can also be identified by the BLS standard occupation codes (Table SLBR-1).

  4. 4 A primary difference in the NCSES taxonomy and the identification of S&E and S&E-related occupations used in this report is that the ACS and CPS data used in this report do not allow for the separation of teachers by level: primary, secondary, or postsecondary. The NCSES taxonomy places postsecondary teachers of S&E fields of study in S&E occupations. When using the ACS or CPS data, all teachers are included in non-STEM occupations regardless of subject matter and level of teaching.

  5. 5 Estimates of the size of the S&E workforce may vary across the different surveys because of differences in the scope of the data collection (the NSCG collects data from individuals with at least a bachelor’s degree); because of the type of survey respondent (the NSCG collects data from individuals, the OES survey collects data from employers, and the ACS collects data from households); or because of the level of detail collected on an occupation, which aids in classifying a reported occupation into a standard occupational category. For example, the NSCG estimate of the number of workers in S&E occupations includes postsecondary teachers of S&E fields; however, postsecondary teachers in ACS are grouped under a single occupation code regardless of field and are, therefore, not included in the ACS estimate of the number of workers in S&E occupations.

  6. 6 The methodology used to identify STEM middle-skill occupations has not changed since the Indicators report Science and Engineering Labor Force 2019, and details can be found in the section “Skilled Technical Workforce Data” of the Technical Appendix.

  7. 7 Other federal agencies have an established STEM taxonomy.
    -U.S. Census Bureau: Women Making Gains in STEM Occupations but Still Underrepresented (
    -Bureau of Labor Statistics options for defining STEM under the Standard Occupational Classification (SOC) system:
    -Department of Homeland Security list of eligible STEM fields to study and work in the United States:

  8. 8 The SOC Policy Committee recommendation to the Office of Management and Budget, which includes options for defining STEM occupations under the 2018 SOC system can be found here: Accessed 18 October 2020. ONET STEM occupations can be found here: Accessed 18 October 2020. These recommendations do not include middle-skill occupations but are referred to here for purposes of comparing their recommendations to our expanded definition that does include middle-skill occupations.

  9. 9 A new federal survey called the National Training, Education, and Workforce Survey (see endnote 2) will include questions about educational attainment, credential attainment, and training of workers without a bachelor’s degree.

  10. 10 CPS data for a longer period (1990–2019) confirm this long-term trend for workers with a bachelor’s degree or higher in S&E occupations and for workers at all educational levels in S&E technician and computer programming occupations (Table SLBR-6). The information in Table SLBR-6 is organized to be consistent with data from previous cycles for comparability over time.

  11. 11 The CPS is the source of the official U.S. unemployment rate.

  12. 12 At the doctorate level, data for 2015, 2017, and 2019 differed from earlier years shown here. SESTAT incorporated data from the SDR, which correspond to the doctor’s research and scholarship degree level and are research-focused degrees. The NSCG data used for 2015, 2017, and 2019 cover all doctorates regardless of type.

  13. 13 Beside the unemployment rate, the IPT rate is the most widely available measure of labor underutilization (Bell and Blanchflower 2021). Along with the IPT rate (U-6 measure) and the unemployment rate (U-3 measure), BLS publishes four other measures of labor underutilization: U-1, the percentage of civilian labor force unemployed 15 weeks or longer; U-2, the percentage of civilian labor force that are job losers and persons who completed temporary jobs; U-4, the percentage of civilian labor force and discouraged workers that are unemployed and discouraged; and U-5, the percentage of civilian labor force and marginally attached workers that are unemployed, discouraged, and marginally attached. See for more details.

  14. 14 All earnings reported are in current dollars unless otherwise noted as constant dollars.

  15. 15 Health sciences is included in S&E fields of study because these data correspond to the doctor’s research or scholarship degree level, which are research-focused degrees.

  16. 16 Although the formal job title is often postdoc fellowship or research associate, titles vary among organizations. This report generally uses the shorter, more commonly used, and best-understood name: postdoc. A postdoc is generally considered a temporary position that individuals take primarily for additional training—a period of advanced professional apprenticeship—after completion of a doctorate.

  17. 17 Three NSF surveys—the SDR, the Survey of Graduate Students and Postdoctorates in Science and Engineering (GSS), and the Early Career Doctorates Survey (ECDS)—include data related to the number of postdocs in the United States. The three surveys overlap in some populations (such as U.S.-trained doctorate holders and those working in academia) but differ in others. For instance, the SDR covers U.S.-trained postdocs in all sectors—academia, industry, and government—whereas the GSS and the ECDS cover both U.S.- and foreign-trained doctorate holders in academia and federally funded research and development centers, which may be run by for-profit or nonprofit businesses, and the National Institutes of Health (NIH) Intramural Research Program, which is in the government sector. Therefore, postdocs in the business and government sectors may be missed. In addition, the titles of postdoc researchers vary across organizations and often change as individuals advance through their postdoc appointments; both factors further complicate the data collection process (NIH 2012). Because the SDR covers only U.S.-trained individuals, it substantially undercounts postdoctoral researchers, most of whom were trained outside the United States. To present more complete counts of postdoctoral researchers, this report uses counts from the GSS, which include foreign-trained postdocs employed in U.S. higher education institutions.

  18. 18 According to NCSES’s (2021e) 2019 GSS, the majority of academic postdocs (60%) in 2019 were supported by research grants; the rest were supported by fellowships, traineeships, or other mechanisms (NCSES GSS 2019: Table 3-6).

  19. 19 Those with faculty rank (e.g., research faculty, scientist, associate, fellow, postdoc) may conduct research and hold an administrative position (e.g., president, provost, chancellor) or teach (e.g., teaching and adjunct faculty).

  20. 20 The other 10 activities are used to define four additional broad categories of primary or secondary work activities: teaching; management and administration; computer applications; and professional services, production workers, or other work activities not specified.

  21. 21 Information on workers in S&E-related occupations reported in the NSCG began to be collected in 2003.

  22. 22 According to NCSES, women, persons with disabilities, and some racial or ethnic groups—Blacks or African Americans, Hispanics or Latinos, and American Indians or Alaska Natives—are underrepresented in S&E. That is, their representation in S&E education and S&E employment is smaller than their representation in the U.S. population. (see NCSES WMPD 2021 for more details.) Although some of these groups may have reached parity with their representation in the U.S. population in some occupations within STEM, they are still defined as underrepresented following the NCSES definition.

  23. 23 The choice of price deflator is nontrivial, affecting the magnitude of the constant dollar median salaries. The U.S. Census Bureau uses the BLS Consumer Price Index for all Urban Consumers Research Series to deflate income estimates (Census Bureau 2020b). This price index is known to be an upper-bound estimate of the cost of living (Gordon 2006). The Federal Reserve Board uses the Bureau of Economic Analysis’ personal consumption expenditure (PCE) price index to measure price changes (Board of Governors of the Federal Reserve System 2016). The PCE price index is a subset of prices in the GDP price index, and it captures prices paid by rural and urban households as well as payments made on behalf of households by third parties (like health insurance payments). While the different price indexes produce constant dollar median salaries that differ in magnitude, the trends in constant dollar measures are similar over time for the analysis in this report. Hence, in this report, the constant dollar median salaries based on the PCE price index are presented, and discussion of the constant dollar median salaries only pertain to changes over time.

  24. 24 Kimberlé Crenshaw (1989) introduced the term intersectionality in critiquing how traditional feminist ideas and antiracist policies exclude black women because they face overlapping discrimination unique to them.

  25. 25 Another pathway to employment for noncitizens is the Optional Practical Training, which allows graduates who were on student visas (i.e., F-1) to stay in the United States for up to an additional 3 years to work full time on their student visa (Zwetsloot et al. 2019). Optional Practical Training approvals (257,000) outpaced initial H-1B approvals (115,000) between 2014 and 2016, and more than half (53%) of the Optional Practical Training participants were STEM graduates (Ruiz and Budiman 2018).

  26. 26 The H-1B program allows companies in the United States to temporarily employ foreign workers in occupations that require the theoretical and practical application of a body of highly specialized knowledge and a bachelor’s degree or higher (or its equivalent) in the specific specialty. H-1B specialty occupations may include such fields as science, engineering, and information technology (DOL 2020a). High-skill labor can also enter the United States for employment through the J-1 (exchange) and L-1 (intracompany transferee programs).

  27. 27 However, precise counts of H-1B visas issued to individuals in these occupations cannot be obtained because U.S. Citizenship and Immigration Services does not classify occupations with the same taxonomy used by NSF.

  28. 28 Stay rates refer to the proportion of U.S. S&E doctorate recipients on temporary visas at graduation that remain in the United States after graduation. Among temporary visa holders who received their S&E doctoral degrees approximately 5 years and 10 years before 2017, 71%–72% remained in the United States in 2017 (NSB 2019b). Updated 5- and 10-year stay rates were not available for this report.