SCIENCE & ENGINEERING INDICATORS

As researchers and policymakers increasingly emphasize the need for expanding S&E capabilities in the United States, demographic groups that are not as represented in S&E as they are in the general population may be seen as an under-utilized source of capacity building in S&E. This lower representation signals a lack of diversity in the workplace, which may negatively impact productivity and innovation. (See Hewlett, Marshall, and Sherbin 2013 and Ellison and Mullin 2014 for discussions on the impact of diversity on workplace productivity and innovation.) Historically, in the United States, S&E fields have had particularly low representation of women and members of several racial and ethnic minority groups (i.e., blacks, Hispanics, and American Indians or Alaska Natives), both relative to the concentrations of these groups in other occupational or degree areas and relative to their overall representation in the general population. More recently, however, women and racial and ethnic minorities increasingly have been choosing a wider range of degrees and occupations. See sidebar College-Educated Individuals with a Military Background for more information on veterans and workers with a military background.

The number of women in S&E occupations or with S&E degrees has doubled over the past two decades (Table 3-16). Despite these gains, the latest data show that women are underrepresented in general in S&E, with notable exceptions. In 2017, women constituted 29% of workers in S&E occupations—up from 23% in 1993—relative to over half (52%) of the college-educated workforce overall. Among S&E degree holders, women represented 40% of employed individuals—up from 34% in 1993—with a highest degree in S&E (Figure 3-18).

Women make up over 34% of all scientists (engineers excluded), although representation varies across the broad fields. Women account for approximately 48% and 59% of life scientists and social scientists, respectively, and nearly 30% of physical scientists and computer and mathematical scientists (Figure 3-19; Table S3-12). Notably, while 59% of social scientists are female, occupations within social sciences varied widely: women accounted for 21% of economists and 69% of psychologists. About 16% of engineers are women, ranging from about 7% of mechanical engineers to 25% of chemical engineers (Figure 3-19; Table S3-12).

In contrast to jobs in S&E occupations, a majority of jobs in S&E-related occupations (58%) are held by women (Table S3-12). Women comprise 71% of health-related occupations. Women in health occupations are employed primarily as nurse practitioners, pharmacists, registered nurses, dietitians, therapists, physician assistants, and health technologists and technicians; women represented the majority of workers in these particular health-related occupations. In contrast, among diagnosing and treating practitioners, women accounted for 44% of workers in these occupations.

Since the early 1990s, the representation of women has grown in most S&E occupations with some notable exceptions. For example, the percentage of women in engineering occupations nearly doubled during the period of 1993 to 2017 from 9% to 16%. In the case of computer and mathematical sciences, while the number of women tripled in this new, rapidly growing and changing field, they did not increase as much as the number of men. The result has been an overall decline in the proportion of women, from 31% in 1993 to 27% in 2017 (Figure 3-19). The declining proportion of women in computer and mathematical sciences occupations does not extend to doctorate-level workers: among those with a doctorate, the proportion of women increased, from 16% in 1993 to 31% in 2017.

Women are a larger share of S&E highest degree holders than of S&E occupations. In 2017, women constituted 40% of employed S&E highest degree holders—up from 34% in 1993 (Figure 3-18). The pattern of variation in the proportion of men and women among degree fields echoes the pattern of variation among occupations associated with those fields (Table S3-13).

The proportion of female S&E highest degree holders has risen at the bachelor’s, master’s, and doctoral degree levels over the past two decades (Figure 3-20). However, female S&E highest degree holders are underrepresented at all degree levels relative to the proportion of women in the college-educated population (52%).

The number of female doctorate holders employed in academia grew rapidly over time (more than doubling between 1997 and 2017) and faster than the number of men (16%) (Table S3-14). Women accounted for 38% of S&E doctorates employed in academia in 2017, up from 25% in 1997, and accounted for 32% of full-time senior faculty (including full professors and associate professors) in 2017, up from 17% in 1997. Gender differences in the doctoral academic workforce vary across disciplines (Table S3-14).

Underrepresentation of certain racial and ethnic groups has long been a concern of policymakers who are interested in the development and employment of diverse human capital to maintain the United States’ global competitiveness in S&E. Blacks, Hispanics, and American Indians or Alaska Natives together make up a greater share of the general population than they do of those receiving S&E degrees or working in S&E occupations. In contrast, whites and Asians tend to comprise about the same or greater portions of S&E degree holders and workers than their proportions of the general population would suggest.

While blacks, Hispanics, American Indians or Alaska Natives, Asians, and Native Hawaiians or Other Pacific Islanders are minorities relative to whites in the general population, they are not necessarily “underrepresented minorities” in S&E (Table 3-17). For example, Asians are over-represented in S&E relative to their proportion of college-degree holders and the general population. At 20% and 16% of the S&E occupations and degree holders, respectively, Asians comprise a much larger proportion of these groups than their proportion of the general population (6%). The proportions of white S&E workers and degree holders are similar to their proportion in the general population—they comprise a little over two-thirds of each of these groups (Table 3-17). The proportion of blacks and Hispanics in both science and engineering groups is less than 10% each relative to their 12% and 16% of the general population, respectively. Between 1995 and 2017, the representation of Asians and Hispanics in S&E has increased considerably while the shares of white scientists and engineers has declined; the representation of black scientists and engineers has risen slightly (Table 3-18).

Representation of workers in certain S&E occupations varied by race and ethnicity (Table S3-15, Table S3-16). Asians had a large presence in computer and engineering occupations, accounting for about a third of computer software engineers, software developers, computer hardware engineers, computer and information research scientists, and postsecondary teachers in engineering. Hispanics had a relatively large presence among psychologists (15%), political scientists (33%), postsecondary teachers in computer science (13%), and industrial engineers (17%). Blacks had high representation rates among computer systems analysts (13%), computer support specialists (14%), and network and computer systems administrators (14%) relative to their representation in S&E occupations overall.

Among S&E highest degree holders, the shares of racial and ethnic groups vary similarly across degree fields, as they do in occupations (Table 3-19, Table S3-20). Compared to most other broad S&E fields, Asians have higher representation rates among those with degrees in engineering and in computer and mathematical sciences; blacks have higher representation rates among those with degrees in computer and mathematical sciences and in social sciences; Hispanics have lower representation rates among those with degrees in computer and mathematical sciences. Whites represent smaller segments of degree holders in engineering and computer and mathematical sciences than in life, physical, and social sciences. In the academic workforce, underrepresented minorities (blacks, Hispanics, and American Indians or Alaskan Natives) constituted 9.3% of total academic doctoral employment and 9.0% of full-time faculty positions in 2017, up from about 6% of both these positions in 1997 (Table S3-18).

The percentage of underrepresented minorities in S&E has grown at all degree levels since 1993 (Figure 3-21). The number of underrepresented minorities with S&E highest degrees at the bachelor’s, master’s, and doctoral degree levels quadrupled in recent decades.^{​Minority-serving institutions (MSIs) play a prominent role in training black and Hispanic scientists and engineers. In 2016, approximately 15% and 46% of S&E bachelor’s degrees earned by blacks and Hispanics, respectively, are from historically black colleges and universities and high Hispanic enrollment institutions (NCSES WMPD 2019: Table 5-8 and 5-9). Between 2013 and 2017, about 37% of Hispanic doctoral graduates earned their bachelor’s degree from high-Hispanic enrollment institutions, and a quarter of blacks earned their bachelor’s degree from historically black colleges and universities.}

Women and racial and ethnic minorities generally receive less pay than their male and white or Asian counterparts (Table 3-20; Table S3-19, Table S3-20). Differences in average age, work experience, academic training, sector and occupation of employment, and other characteristics can make direct comparison of salary statistics misleading. Degree areas with lower salaries generally have higher concentrations of women and racial and ethnic minorities. Disproportionately larger shares of degree holders in life and social sciences work in occupations not categorized as S&E, and the salaries for these occupations are generally lower than for other S&E occupations. Salaries also differ across employment sectors. Academic and nonprofit employers typically pay less for similar skills than employers in the private sector, and government compensation generally falls somewhere between these two groups. These differences are important for understanding salary variations by sex, race, and ethnicity because men, Asians, and whites are more highly concentrated in the private, for-profit sector. Salaries also vary by indicators of experience, such as age and years since degree completion. Because of the rapid increase of the number of females in S&E fields in recent years, women with S&E degrees who are employed full time generally have fewer years of labor market experience than their male counterparts.

Statistical models can estimate the size of the salary difference between men and women, or the salary differences between racial and ethnic groups, when various salary-related factors are taken into account. The analyses presented in this section show that statistical models used to control for effects of education, experience, and other factors on salaries tend to reduce these differences. The models used here estimate salary differences between men and women among individuals who are otherwise similar in age, work experience, field of highest degree, occupational field and sector, number of children, and other relevant characteristics that are likely to influence salaries. Also included are data related to salary differences between Asians and whites, and individuals in the remaining race and ethnic categories (American Indians or Alaska Natives, blacks, Hispanics, Native Hawaiians or Other Pacific Islanders, and those reporting more than one race).^{​Salary differences represent estimated percentage differences in women’s reported full-time annual salary relative to men’s reported full-time annual salary as of February 2017 and in whites and Asians relative to all other races and ethnicities, respectively. Coefficients are estimated in an ordinary least squares regression model using a natural log of full-time annual salary as the dependent variable. This estimated percentage difference in earnings differs slightly from the observed difference in median earnings by gender and race and ethnicity because the former addresses differences in mean earnings rather than median earnings.}

Controlling for the effects of differences in field of highest degree, degree-granting institution, field of occupation, employment sector, and experience,^{​Included are 20 NCSES-classified field-of-degree categories (out of 21 S&E fields), 38 NCSES-classified occupational categories (out of 39 categories), 6 NCSES-classified employment sector categories (out of 7 categories), years since highest degree, and years since highest degree squared.} the estimated salary difference between men and women narrows by more than half relative to the total difference in full-time salary (Figure 3-22). However, after controlling for these effects, a salary differential remains. Women earn 9% less than men among S&E highest degree holders at the bachelor’s or doctoral level and 10% less at the master’s level. Compared with whites and Asians (controlling for education and employment), S&E highest degree holders in other racial and ethnic groups working full time earn 10% and 5% less at the bachelor’s and master’s degree levels, respectively (Figure 3-23).^{​The estimates for doctoral degrees in the “controlling for education and employment” category and for doctoral degrees and master’s degrees in the “plus demographics and other characteristics” category are not statistically significant at the 90% confidence level}

Salaries vary by factors beyond education, occupation, and experience.^{​For example, marital status, the presence of children, parental education, and other personal characteristics are often associated with salary differences. These differences reflect a wide range of issues, including (but not limited to) factors affecting individual career- and education-related decisions, differences in how individuals balance family obligations and career aspirations, and productivity and human capital differences among workers that surveys do not measure.} Salaries also differ across regions, partly reflecting differences in the cost of living across geographic areas. However, adding such measures as well as personal and family characteristics to education, occupation, and experience results in marginal changes in the estimated salary differences between men and women, and among racial and ethnic groups,^{​In addition to the education- and employment-related variables, the following indicators are included in wage regression models: nativity and citizenship, marital status, disability, number of children living in the household, geographic region (classified into nine Census Bureau divisions), and whether either parent holds a bachelor’s or higher-level degree. The gender regression controls for racial and ethnic minority status, and the race and ethnicity regression controls for gender.} compared with estimates that account for education, occupation, and experience alone. Women’s adjusted salary differentials remain at all degree levels (Figure 3-22), while the adjusted salary difference among racial and ethnic groups remain at the bachelor’s degree level (Figure 3-23).^{​The estimates for doctoral degrees in the “controlling for education and employment” category and for doctoral degrees and master’s degrees in the “plus demographics and other characteristics” category are not statistically significant at the 90% confidence level.}

The analysis of salary differences suggests that attributes related to human capital (i.e., fields of education and occupation, employment sector, and experience) rather than socioeconomic and demographic attributes have a greater influence in explaining the salary differences observed among S&E highest degree holders by sex and across racial and ethnic groups. Nonetheless, the analysis also shows that measurable differences in human capital do not entirely explain salary differences between demographic groups.

Readers should keep in mind that the interaction between demographic attributes and those related to human capital are complicated and may impact labor market outcomes. The regression analysis addresses major factors that affect differences in earnings but does not attempt to cover all possible sources of difference (for more detailed discussions, see Blau and Kahn 2017; Ceci and Williams 2011; Mincer 1974; Polachek 2008; and Xie and Shauman 2003).