A Symbiosis of Access: Proliferating STEM PhD Training in the U.S. from 1920–2010

Abstract

Over the course of the 20^th century, unprecedented growth in scientific discovery was fueled by broad growth in the number of university-based scientists. During this period the American undergraduate enrollment rate and number of universities with STEM graduate programs each doubled three times and the annual volume of new PhDs doubled six times. This generated the research capacity that allowed the United States to surpass early European-dominated science production and lead for the rest of the century. Here, we focus on origins in the organizational environment and institutional dynamics instead of conventional economic factors. We argue that three trends of such dynamics in the development of American higher education not often considered together—mass undergraduate education, decentralized founding of universities, and flexible mission charters for PhD training—form a process characterized by a term coined here: access symbiosis. Then using a 90-year data series on STEM PhD production and institutional development, we demonstrate the historical progression of these mutually beneficial trends. This access symbiosis in the U.S., and perhaps versions of it in other nations, is likely one critical component of the integration of higher education development with the growing global capacity for scientific discovery. These results are discussed in terms of the contributions of American universities to the Century of Science, recent international trends, and its future viability.

The research university played a central role in the global expansion of scientific capacity-building and discovery, particularly since the mid-20^th century (Meyer and Schofer 2007; Powell, Baker, & Fernandez, 2017). The modern research-intensive university primarily developed within a U.S. context and formed the foundation for the country’s leading contribution to global science, technology, engineering, mathematics, and health (STEM+) research (Fernandez & Baker, 2017). Realistically or not, the American research university serves as a model for university development in many countries (Ramirez & Tiplic 2013). However, past research has not fully explained the dynamics that drove the U.S. to achieving mass graduate education. In 1920 a handful of elite institutions annually trained just several hundred new STEM+ PhDs; by 2010 a wide array of universities graduated more than 25,000 new scientists. The expansion of mass STEM+ doctoral education represents large and sustained national scientific capacity.

The popular explanation is that a small core of wealthy, often private, research-intensive universities benefitted from U.S. economic growth and wartime spending to reach unprecedented levels of graduate training and research production. Reflecting this notion, past scholarship attributes U.S. science capacity to the forces of wealth accumulation, hegemonic geo-politics, and integration of capitalism and technology (e.g., Guston and Keniston 1994; Kennedy 2014; Leslie 1993; Smith 1990; Stephan 2012). Related too, other scholars focus on an informal hierarchy of universities topped by a core of research universities. For example, much is made of the fact that approximately five dozen members of the Association of American Universities (AAU) produced 46% of the nation’s PhDs in all fields (e.g. Labaree 2017). As shown below, in actuality a diverse, larger set of universities produces STEM+ PhDs, a telling feature less appreciated because past scholarship tends to overlook the role of the less research-intensive tiers of U.S. higher education in training and diversifying STEM+ human capital.

By focusing instead on the broader organizational environment and institutional dynamics of U.S. higher education, we argue that three trends—mass undergraduate education, decentralized founding of universities, and flexible mission charters for PhD training—formed a process characterized by a term coined here: access symbiosis. Each historical trend was characterized by access. Youth had access to mass undergraduate education; decentralized actors had access to found universities; and, universities had access to the market for doctoral education through flexible mission charters. In other words, our use of the word “access” underlines a relative openness that defined U.S. development of higher education.

We further argue that the three trends influenced one another symbiotically, expanding opportunities for advanced science training and increased diversity of graduate students. Here, “symbiosis” suggests connections over time, not in a narrow biological sense, but rather in a general sense of mutually beneficial relationships among these trends. Along with other resources central to science, this less-appreciated access symbiosis yielded significant growth in new scientists researching at a growing number of public and private research universities that were, and continue to be, major resources behind the nation’s dominance of science production over the 20^th century (Fernandez and Baker 2017).

There are, of course, important past, separate lines of research on all three trends that provide the foundation for the integrating argument here. For example, research thoroughly documents increasing global demand for higher education and the rise of mass postsecondary enrollment and the implications of “mass-ness” for universities’ changing societal role and the early appearance of these changes in the United States (Baker 2014; Meyer and Schofer 2007; Schofer and Meyer 2005; Trow 2007). Other research charts the rise of mass doctoral education as a product of both growing national and international enrollments, as well as a form of competition around national science outputs (e.g., Berelson 1960; Bowen and Rudenstine 1992; Taylor and Cantwell 2015). Additionally, historical accounts detail the development of a large decentralized set of public and private research universities from a variety of societal origins with shifting missions (e.g., local communities, states, religious groups, industrialists, etc.) (Geiger 1993, 2016). Plus, recent insightful appraisals foreshadow the access symbiosis argument, such as studies of the consequences of greater social inclusion and lack of a mechanism for central higher education planning (Brint 2019; Labaree 2017).

Although the constituent trends in our access symbiosis argument have been noted, lacking is a systematic quantification of the historical patterns over a long historical period—particularly one that focuses on the development of STEM+ human capital for U.S. dominance in the Century of Science. This is provided here in several steps. First, to put what American higher education did in developing U.S. science capacity in historical perspective, we present the trajectories of the global growth in STEM+ research production, the U.S.’s contribution to this growth, and the share done by American university-based scientists. Then, combining known historiography with a national dataset on the training of PhDs in STEM+ fields from 1920 to 2010, we present four multivariate graphs plus a timeline of STEM+ PhD production illustrating the access symbiosis over this period. Lastly, we discuss the implications of the results for a broader understanding of the contributions of American universities to global science, along with an assessment of the viability of the access symbiosis in the face of recent challenges.

The Historical Context: A Century of Science

From approximately the end of the 19^th century until the present, what is referred to as the “Century of Science” is marked by three phases of unprecedented expansion in the world’s volume of scientific discovery. As displayed in Figure 1 plotted against the right y-axis, growth has increased from about 10,000 annually published STEM+ journal papers in 1900 to more than one million by 2010. Up until about 1950, “Small Science” was notable by novel increases of scientific journal papers and scientific topics; starting about 1950 “Big Science” emerged with the take-off of exponential growth in science publications (de Solla Price 1963). Then, instead of a much predicted leveling off, from about 1980 a globally expansive “Mega-Science” produced papers at an exponential annual growth rate of 3.5% (Powell et al. 2017). American universities and scientists played pivotal roles in generating science capacity in all three periods.

Figure 1.

Estimated Total Worldwide STEM+Health Journal Research Publications, Proportion of Total Publications with at least one U.S.-based Scientist, and Proportion of U.S. Publications with at least one U.S. University-based Scientist, 1900 – 2010.*

*Calculated from the Science Productivity, Higher Education Development and the Knowledge Society database (see AUTHOR for details). STEM+ journals publications are original research papers on topics in science, technology, engineering, mathematics, plus health.

In addition to increasing STEM+ research production, there are two other trends motivating the analysis. First, as plotted against the left y-axis, U.S.-based scientists have contributed the largest share to science production as reflect by annual research articles. By 1900 the volume of annual papers from the U.S. was growing relative to European nations and at the height of Big Science, U.S.-based scientists were responsible for 40 to 50% of the world’s publications (see the dotted line in Figure 1). Even in the Mega-Science period, with renewed research capacity from European nations and the new large producers of China, Japan, and South Korea, plus an increasing number of other nations, U.S.-based scientists consistently publish the largest share of scientific publications: about 25% of the over one million annual papers in 2010—the next largest is China at about 14% (Powell et al. 2017).

To accomplish large-scale research production, the U.S. had to develop the world’s largest sustainable science capacity in terms of new STEM+ PhDs and university positions for scientists to do research. The dashed line, plotted against the left-axis, showing the percentage of U.S. generated STEM papers that included at least one university-based author. In 1900 about an estimated quarter of all papers from the U.S. included a university-based author and this steadily increased to just over 80% by 2010.ⁱ American universities and their faculty led the way in a global trend of the greater integration of academia and science. Compared to governmental institutes and private firms, university-based scientists worldwide played a substantial role in increasing their share of the total volume of papers, particularly in the Mega Science period, where now an estimated 80% of all annual papers worldwide include at least one university-based author.

The Access Symbiosis in the Unique American Context

Unlike other nations, the U.S. federal government did not open a national university, nor did it directly control the number of enrollments at the undergraduate and graduate levels including PhD training (Clark 1995). At historical periods for certain segments of youth, the federal government subsidized tuition, but the founding and chartering of universities was never tightly controlled. American higher education expanded without federal restriction on numbers of new universities, research intensity, or missions; the supply of institutions and their organizational diversity were never centrally mandated nor bureaucratically maintained (Labaree 2017).

The same is true on the demand side. The U.S. was among the earliest of nations to strive toward compulsory primary education for children, and it achieved near-universal secondary education by the 1960s. Unlike many European countries, the U.S. never relied on streaming high proportions of youth into vocational, non-university training (e.g., Powell and Solga 2011). Therefore, near inelastic demand for postsecondary education accrued from this unique and comparatively early expansion of secondary education.

At the beginning of the 20^th century, only a small percentage of young adults enrolled in postsecondary institutions. As secondary education expanded after World War II, Americans pursued higher education in increasing rates at a variety of organization forms, including community colleges, liberal arts colleges, public regional colleges, land grant universities, and private universities (Baker 2014). Over a span of a few decades, higher education went from something for the privileged few to something that was approaching mass access—that is now attended by more than two-thirds of a national cohort of young adults (National Center for Education Statistics 2005; Snyder 1993).

Early in the century, government policy did not necessarily seek to increase the number of PhD researchers by growing the overall capacity of the higher education system. Yet when enrollments grew at primary schools and then secondary schools, they spilled into and built upon an early-century elite system of higher education (where the founding of new universities was already unrestricted). This process sustained increased demand for training of PhDs in science and related fields. As the higher education system expanded, universities and departments sought greater numbers of PhD-holding faculty members to supply instruction to growing cohorts of students (Berelson 1960).

Along with the massification of higher education, the 1960s witnessed an “academic revolution” whereby universities internally began to set more demanding standards for faculty to produce greater amounts of research and obtain external funding to subsidize the research process, including the hiring of graduate research assistants (Jencks and Riesman 1968). Federal funding support for basic scientific research hit its zenith and began to decline in the 1970s, yet STEM+ faculty faced consistent pressure to produce scholarship and compete for scarce funding (Geiger 1993).

In retrospect, the emergence of mass graduate education in the U.S. seems like a foregone conclusion, but things could have turned out differently. At each period of the expansion of U.S. higher education, a concurrent alternative “elitism” was reflected in public debate to limit graduate training and science production to the presumed “best and brightest” among individuals and institutions (Ben-David 1977). For example, as early as the middle of the 19^th Century a self-referent coalition of preeminent scientists known as the Scientific Lazzaroni lobbied for the establishment of the National Academy of Sciences and pushed for a centralized European model of both university and state-funded institutes of science (Hunter 1988). Also, there was clear political resistance to the expansion of the university by the 1960s that echoed earlier positions and has since periodically returned in American society (e.g., Schofer et al. 2019). And certainly, it was in spite of nativist resistance, bigotry, and racism, that the U.S. managed to incorporate significant proportions of successive generations of young people into elementary schooling up through postsecondary education.

Data, Indicators, Analytic Approach

In this section, we analyze data from the Survey of Earned Doctorates (SED), including data from the Doctorate Records File. Although it is referred to as a survey, the SED is the National Science Foundation’s census of all doctorate recipients who graduated from American universities since 1920 (National Science Foundation, 2006). We analyzed SED restricted micro-data for all STEM+ PhDs and calculated aggregate statistics for PhD-granting universities and new doctorates. Historical aggregates of new STEM+ PhDs include gender (male, female), race-ethnicity (white, minority), country of origin (native born, foreign born or, hereafter, international), institutional trajectory of undergraduate education (starting at two-year institutions, four-year only), and parents’ educational attainment. To examine changes in PhD-granting universities, we merged the annual SED with data on universities funding sector (public or private) from (IPEDS) and national data on undergraduate enrollments and national gross postsecondary enrollment rates of 18–24-year-olds (NCES 2005, 2010; Snyder 1993).

We also included data from the Carnegie Foundation’s classification system of a university’s levels of research activity (Carnegie Foundation for the Advancement of Teaching 2012). Because there are no Carnegie classifications for early in the century, current institutions’ classifications are applied back to earlier periods. While some likely changed their research level over time, these categories are relatively stable and approximate different types of universities contributing to STEM+ graduate training. Four research activity classifications are used: 1) universities with the highest levels of research activity (R1); 2) universities with higher research activity (R2); 3) universities with moderate research activity (R3); and 4) a subset of sixteen R1’s (hereafter, Elite-16), including: University of California (Berkeley), University of Chicago, California Institute of Technology, Columbia University, Cornell University, Harvard University, University of Illinois (Urbana-Champaign), The Johns Hopkins University, Massachusetts Institute of Technology, University of Michigan (Ann Arbor), University of Minnesota (Twin Cities), University of Pennsylvania, Princeton University, Stanford University, University of Wisconsin (Madison), and Yale University (Geiger 1997, 2016). As the earliest leaders in both PhD training and scientific publications, the Elite-16 is analyzed separately because they represented what could have been the foundation for an alternative, exclusive elite system of science production. Ultimately, though, with a mix of some of the oldest and most elite private universities with five newer public universities, the Elite-16 group foreshadowed the expansion and institutional diversity to come (Geiger 2016).

Historical Trends

Proliferation and Diversity of STEM+ PhD Training

Figure 2 graphs the historical production of STEM+ PhDs between 1920 and 2010 streamed by race, gender, and nationality. From the perspective of the access symbiosis, there are three notable trends. First is rising growth in PhD’s, second is its concurrence with expansion of undergraduate enrollment, and third is a later intensifying diversity among PhD’s that likely staved off a significant decline in science capacity. A few decades into the Small Science period, universities were training fewer than 300 new STEM+ PhDs annually, but this volume started to increase at an exponential rate of 8.5% doubling approximately every 8 years. By the 1950s—the beginning of the Big Science period—the rate of growth in STEM+ PhD production slowed to 4.2% and the doubling time increased to approximately a decade and one half, by this point, however, the national output was sizable, and universities produced over 4,000 new scientists every year. And by 1970, annual STEM+ PhD production grew to over 14,000. This growth in the number of scientists was a crucial component in enabling U.S.-based scientists to go from producing about 20% of the world’s scientific papers at the beginning of the century to over nearly half of all papers 60 years later (see Figure 1).

Figure 2.

Production and Diversification of New PhD’s in STEM+ and National Gross Enrollment Rate of 18–24-year-olds in Postsecondary Education, 1920–2010.

As undergraduate education grew, so did graduate training programs.ⁱⁱ As early as the 1840s, formal education was emerging as a central determinant of occupational and social prestige; this irreversible trend encouraged educational expansion at each sector (Hogan 1996; Hout 1984). As displayed in the Figure 2 trend line, gross enrollment rates among 18–24 year-olds swelled over the century and increased at an overall growth rate of 2.4%. Postsecondary enrollment rates tripled from 5% of the young adult cohort in 1920 to 14% in 1950. In the Big Science period, enrollment rates expanded to 40% and then climbed to 70% through the Mega-Science period. In total, U.S. postsecondary enrollment increased from fewer than 600,000 in 1920 to over 21 million in 2010. The undergraduate enrollment part of the access symbiosis was made possible because the state never capped university attendance or prohibited university charters in such a way that blocked universities from using undergraduate training to support research activity and graduate training (Clark 1995; Trow 2010).

The comparatively more open access of American higher education is also demonstrated in how a looming crisis in science capacity was unintentionally avoided through a change in demographic and nationality characteristics of STEM+ PhDs. As shown in Figure 2, starting in 1970 PhD numbers began to drop, and were it not for the greater diversification of STEM+ training it is probable that the U.S. decline in relative share of science discovery would have been far more extreme. During the early proliferation of PhD production, virtually all the growth was among enrollment of native-born white males in STEM+ graduate programs; in 1950 only 5% of new PhDs were from outside this part of the population. Setting aside the 1965 to 1975 spike from the baby-boom population and educational deferments from the Vietnam war, if PhD training had been limited mostly to native-born white males, it is doubtful that the U.S. could have maintained its level of contributions to Mega-Science.

By the 1970s, the proportion of native-born males dropped as a combination of native-born white females and native-born minorities, and international students, mostly with baccalaureates from non-American universities, made up about 40% of all new STEM+ PhDs. And this trend toward greater diversity intensified through the rest of the century, so that by 2010, only 28% of new PhDs were native-born white males. Broadened access to PhD training in 2010 included 20% native-born white females, 12% native-born minorities, and 40% international students. As the percentage of international students earning PhDs in STEM+ fields increased, so did the percentage of international PhD earners who undertook research in the U.S. In 1995, half of the international students who completed STEM+ PhDs intended to stay in the U.S., while by 2015, approximately 75% of international PhD earners intend to stay and research in the U.S. (National Science Foundation 2017; Woodrow Wilson National Fellowship Foundation 2005).

Greater variety among educational paths to graduate study also likely helped support growing capacity for PhD production. By 1960, 40% of new PhDs started their undergraduate training at public, less prestigious universities, and the proportion rose to just over one half by 2010 (not shown in Figure 2). Also, by 2010, 18% of all STEM+ PhDs had started their undergraduate training at a two-year college. A similar pattern occurred for students of less-educated parents (e.g. neither parent earned a baccalaureate, hereafter “first-generation”). At the height of the Big Science period, about 60% of all STEM+ PhDs were first-generation students. As the expansion of higher education for the entire population continued, this proportion would drop, but even by the beginning of the Mega-Science period, 45% of new PhDs were first-generation students.

Figure 3 shows the timing of some key milestones in the demographic and educational diversification of American PhD training. First, we consider gender diversity. In 1950, for example, 5% of new PhDs were U.S.-born women, a share that would double by 1976, and double again by 2009. At the same time, by 1966, female students accounted for 40% of all undergraduate enrollments—a percentage which continued to increase over the remainder of the century (Bowen and Rudenstine 1992). Not long after this, an ideology emerged that increasing educational opportunities for women was good for economic growth and promoting equal opportunities to access all levels of education was generally accepted as a human rights issue underpinning both American and the global expansion of higher education (e.g., Berkovitch and Bradley 1999; Schofer and Meyer 2005).

Figure 3.

Timeline of Milestones in STEM+Health PhD Production.

Next, in considering the contributions of public versus private university sectors, there are significant milestones in how the relative distribution of PhD production shifted between 1920 and 2010. STEM+ training was already undertaken in public universities by 1930, and soon these publics would also send large shares of undergraduates on to graduate science training. In 1945 public universities graduated half of the total supply of American undergraduates who later earned STEM+ doctorates. Public universities were also an essential pathway to becoming a scientist for students whose parents had not attended university; by 1990 more than two-thirds of first-generation undergraduates who went on to earn the PhD originated from public universities.

A comparable trend is apparent for native-born racial and ethnic minority students. To be clear, compared to their relative share of total population, racial and ethnic minority groups are still under-represented as undergraduates in elite universities and also in subsequent STEM+ PhD training (Ashkenas, Park and Pearce 2017; Posselt, Jaquette, Bielby and Bastedo 2012). Nevertheless, while access moved more slowly for a number of reasons, there has been some diversification that has added to the supply of new scientists. Between the 1970s and early 2000s, the percentages of high school graduates who enrolled in some form of higher education increased across racial and ethnic groups. For example, fewer than 50% of black high school graduates enrolled in higher education in 1972, but almost 75% did by 2004 (Posselt et al. 2012). And, between 1974 and 1995, the percentage of STEM+ doctorates earned by native-born people of color increased from 13% to 18%. Yet, the minority share decreased by 2010 as the relative share of STEM+ PhDs increased more quickly among international students.

Undergraduate Growth and the Supply of STEM+ Training Universities

The access symbiosis is also evident in the historical arc of the founding of universities and upgrading of their research intensity. As shown in Figure 4, over the century as the higher education system enrolled an increasingly volume of baccalaureate students, an array of non-governmental and provincial (U.S. states) actors, with open access, founded new universities (Berelson 1960; Jencks and Reisman 1968). In such an institutional environment increased demand can more directly influence supply in unrestricted, significant growth in the number of universities, which in turn provided a larger platform for the training of new scientists.

Figure 4.

Total Number of Universities training STEM+ PhDs by Carnegie Classification and National Gross Enrollment Rate of 18–24-year-olds in Postsecondary Education, 1920 – 2010.*

*Elite-16 is a subset of R1’s.

Nevertheless, it took some time for this process to fully unfold. As historical accounts have shown, the Elite-16 universities generated the bulk of university-based science from the middle of the 19^th century through to the end of the Small Science period—at which point they accounted for one-half of all university research expenditures in the U.S. (Geiger 2016; U.S. National Resources Committee 1938). Not surprisingly, by 1920 these universities represented 41% of the STEM+ PhD granting institutions, but even at this early point, the American pattern of expanding supply of graduate training opportunity had emerged. In addition to the Elite-16 there were 21 other universities contributing to STEM+ training, most of which were less research active than the Elite-16. Some universities increased STEM+ PhD production during the Small Science period; as for example the Ohio State University increased production from 2 PhDs in 1920 to more than 50 by 1930. Additionally, institutions such as Boston College, which eventually became a research university, was founded by the non-government organization of the Jesuit Order of the Catholic church in 1863, and Michigan State University, founded in 1865, did not start awarding STEM+ PhDs until the late 1920s.

By the beginning of the Big Science era, the number of universities with active STEM+ doctoral programs had more than doubled to 95. By the advent of Mega-Science, the number of STEM+ doctoral granting universities had grown to 253 and expanded to 323 by 2010. From 1920 to mid-century, the U.S. supply of graduate training universities grew at an exponential annual rate of 3%. Similarly, as worldwide science publications shifted into a sustained exponential climb in about 1950, the supply of universities awarding STEM+ PhDs grew by 3.3%. Building on a bigger base, the rate of growth of universities awarding STEM+ doctorates declined to 0.82% during the Mega-Science period. Of course, not all universities contributed new PhDs in equal volume and breadth of STEM+ disciplines, but Figure 4 represents significant expansion over the century in the supply of opportunities to create new scientists.

Flexible Mission Charters and STEM+ Graduate Training

Within the access symbiosis U.S., universities’ access to determine their own development also played a key role in the growth of STEM+ PhDs. Higher education institutions across the century had access to relatively unfettered autonomous development. Without central control, mission charters of universities were relatively flexible, meaning that higher education institutions were, and still are, not restricted in their pursuit of status as a university (e.g. a teaching college transforms itself into a research university), nor in the extent of their graduate training programs, nor in the intensity of their research mission. In other words, without rigidly controlled mission charters, U.S. institutions could self-determine their status as a university and their level of research activity. By contrast, in England, Parliament did not grant a select few universities the authority to grant PhDs until 1917, and in Germany there is still strict government control of the training and research missions of postsecondary institutions (Dusdal et al. 2020; Simpson 1983).

One indicator of the consequence of flexible mission charters is reflected in the shaded groupings of the research activity level of universities (i.e. Carnegie Classifications) in Figure 4. Two processes are reflected here. First, over the Small Science period, 57 new R1 universities with graduate training in STEM+ were founded, adding to the Elite-16. Non-state individual and collective actors played a recurring role in the establishment of many public universities that were to become research active in a relatively a short time (e.g. Williams 2018). Thus, the R1 category expanded as older universities increased their research mission (e.g. University of Texas) and as future R1s were founded (e.g. University of Houston in 1927). Second, over the Big Science period and into Mega-Science, STEM+ PhD training was also supplied by a growing number of R2 and R3 universities, so that by 2010 approximately two-thirds of all STEM+ PhD granting institutions were less research-intensive universities.

Although some countries tried, and continue, to focus their national policy attention and resources on developing a small core of elite universities, the decentralized U.S. system resulted in numerous universities offering doctoral programs and preparing large numbers of PhDs to do scientific research (e.g. Fu, Baker, Zhang 2018). For example, by 1900 Germany already had 30 universities whose faculty were training new scientists and publishing a significant share of the world’s STEM+ papers, while the 25% larger U.S. would not have over 30 research active universities until the 1920s. But while Germany maintained a highly stratified secondary system throughout the century, a slower rate of expansion of postsecondary, and a culture of smaller elite science, the U.S. developed approximately three times the research universities by the global Mega-Science period (Dusdal 2018; Fernandez and Baker 2017).

Panel A of Figure 5 illustrates the implications of having greater diversity in research activities among universities that contributed to the expansion of STEM+ PhD production. While the Elite-16 made up less than one half of all STEM+ doctorate-granting universities in 1920, the Elite-16 awarded almost 80% of the country’s STEM+ PhDs. Yet over the Small Science period, the contribution of the Elite-16 began a long decline. Newer R1s awarded about one-half of STEM+ PhDs from 1950 on, and the R2s and R3s increased their combined share of PhDs from less than 1% in 1920 to approximately over a fifth by 2010 (14% and 8% respectively). It is true that the sixty-five, most research-intensive R1’s including the Elite-16 that make up the self-referent Association of American Universities, contribute the largest share of PhD’s, earlier and contemporarily (e.g. Brint 2019; Labaree 2017). But this misses the point, steadily over the century significantly more universities—some to become very research-intensive at a later point, others to remain less so—have increasingly made major contributions to STEM+ training.

Figure 5.

Production of New PhD’s in STEM+ by Carnegie Classification, Public-Private Sector, and National Gross Enrollment Rate of 18–24-year-olds in Postsecondary Education, 1920–2010

Panel B of Figure 5 shows the same trend, but from the relative contributions of public and private universities to STEM+ PhD production. In 1920 64% of all new STEM+ PhDs were awarded by private universities, including many in the Elite-16, but this dominance began to drop through the remainder of the Small Science period. By the beginning of Big Science, new STEM+ PhD production was nearly evenly split with public universities eclipsing privates. In the Mega-Science era, the balance had completely reversed, and public universities awarded more than 70% of STEM+ PhDs. This empirical pattern illustrates that without federal control through caps on undergraduate enrollment, public universities grew enough to undertake the costly endeavor of large-scale doctoral education in a wide array of scientific topics. Certainly, the greater allocation of high education funding from federal and state administrations after World War II at least up to about 1970 benefited this expansion and diversification, but so did the organizational and institutional dynamics behind the growth of undergraduate education at large public research universities, and as federal and state funding leveled out and declined, these dynamics have continued (Berelson 1960).

Discussion

The access symbiosis was foundational—a 90 years-long central dynamic—for the U.S.’s world-leading capacity to participate in a period of unprecedented growth in science discovery. Over this period, the undergraduate enrollment rate and the number of universities with STEM+ graduate programs each doubled three times, and the annual volume of new scientists doubled six times. We argue that these trends influenced one another. Not only did the U.S. expand STEM+ training, U.S. universities achieved significant growth by enrolling graduate students from a variety of racial, gender, and social backgrounds, and national origins later in the century.

Access to enroll in higher education, access by a range of actors to freely found universities, access to flexible mission charters and autonomous development of those universities, and access to advanced science training, were all symbiotically related. This is perhaps most clear in the fact that the widening diversity of individuals seeking advanced STEM+ training likely rescued the U.S.’s capacity from a leveling off as science discovery became ever more global beginning in the 1980s. The federal government—through the efforts of the Nation Science Foundation and the Department of Education—encouraged universities to increase female and racial or ethnic minority enrollments in STEM+ fields, but universities were not mandated to do so. Yet even soft federal power does not explain the admittance of international STEM+ students (and faculty), a large source of new scientists that was neither a national strategy nor policy. Analysis of doctoral student-level data reveals that from mid-century on diversifying enrollment across STEM+ graduate programs, as imperfectly and haltingly as it was, not only ensured the ability of American universities to fully participate in the Century of Science, it placed the U.S. at the center of an unprecedented rise in the production of new scientific discovery.

Alternative Historical Scenarios

The graphical trends are consistent with historiography, but the access symbiosis was not an inevitable outcome. American federalism left educational development relatively unrestricted, but the European model of a core of educated elite trained in a limited number of small universities was known and attractive in the U.S. (Trow 2010). Even the first American President, George Washington, assumed that an elite national “University of the United States” would be created, and various attempts to do so well into the 19^th century failed not because of a rejection of an elite model, but because of oppositional lobbying by established institutions, such as those in the Elite-16 (Trow 2010). Also, as noted above, the U.S. could have adopted some form of the more restricted European model of higher education, particularly since it was experienced first-hand by a number of 19^th century aspiring American scientists who obtained the PhD from German and British universities. Finally, there have always been arguments in the U.S. against mass higher education, which were accompanied by cautionary tales of allowing too many universities, too many undergraduates, and too many PhDs (Ben-David 1977; Berelson 1960; Clark 1995; Hunter 1988).

Related, there is an underappreciation of the access symbiosis itself, largely because of a misinterpretation of what the Elite-16 and a few other universities like them represent. They are often assumed to reflect a more elitist model, and understandably, historiography underscores their early spectacular success, just as contemporary assessments praise them as a singular group of “world-class” science producing institutions (e.g. Yudkevich, Altbach, and Rumbley 2016). But their abundant resources, world-ranked prestige, and mostly private control leads to over-identifying them as “the secret” behind American university-based research; what is missed is that these universities were actually a forerunner to the broader access symbiosis that drove the century-long process (Geiger 2009; Labaree 2017). Until now, scholars have theorized about elements of the access symbiosis without using systematic data to provide evidence of the phenomenon through the latter half of the 20^th century.

The majority of Ivy League universities were mostly founded by various Protestant denominations, The Johns Hopkins University and Stanford University by educationalists and aging capitalists, University of Michigan by the state of Michigan, and the other public universities in this group by a blend of everything from federal land-granting to local interests in agricultural training and science (Williams 2018). They were also not limited in the direction of their future development and scholarly missions, which in most cases evolved in directions divergent from original local charters, and by the early 20^th century, they were too large a group to survive by solely educating a small elite even if that was their intention. While remaining prestigious, the Elite-16, followed by a larger set of universities who self-identified with them, nevertheless set many of the dynamics of the earliest stages of the access symbiosis. The early research-intensive universities represented a goal that many other universities, realistically or not, aspired to—providing the aspirational target for a kind of unrestrained institutional academic drift.

The Access Symbiosis as Historical Explanation

Earlier scholarship conceptualizes the history as one of inclusive undergraduate education cross-subsidizing research capacity within the improbable result of a “perfect mess” of decentralized educational development (Brint 2019; Labaree 2017). And while both images are helpful, there is likely a concurrent deeper process at work. The U.S. was always at the historical forefront in the expansion of education and in the penetration of education into its society, and this high level of unchecked educational development perpetuates its own legitimation (Baker 2014; Meyer 1979). The access symbiosis, then, can be thought of as a result of a broader integration of education that yields a kind of societal cross-subsidy between educational expansion and science production. Greater social inclusion is also an important part of the story and stems from the same broad cultural integration.

Historians describe other contributing developments within the American university that are not covered in this analysis, such as development of the academic department, the role of rural outreach and agricultural science, and university-government partnerships (e.g. Geiger 2009). And American science productivity is often attributed to federally-sponsored research grants, even though universities did not benefit from large federal grants until the latter half of the 20th century, and a relatively small number of universities received large amounts of the government’s largesse in both STEM+ and social science fields (Cantwell and Taylor 2013; Geiger 1988, 1997; Kennedy 2014; Stephan 2012). Similarly, scholars continue to examine the most research-intensive universities in the U.S. and the importance of private higher education (e.g., Taylor and Cantwell, 2015). The access symbiosis idea may help scholars expand or complement traditional analyses, particularly in crediting both private and public sectors, considering the importance of widening access to undergraduate education, appreciating the contributions of less-research-intensive universities, and considering multiple ways that U.S. universities broadened participation in mass graduate education.

As noted above, American educational development did not occur within a vacuum as many non-educational events factored in to the broader history. The focus here on institutional and organizational factors of higher education enriches conventional focus on economics, arms technology, and federal funding patterns. The mutually propelling trends of the access symbiosis likely enabled other more distal factors to play a role in the country’s science capacity. Of course, competition for resources, historical circumstances, and a degree of chance also dictated the emergence of this historically unusual arrangement compared to so many other nations. Given these other factors, future research should examine the degree to which the access symbiosis set the foundation for other educational and non-educational developments related to science capacity.

Future Viability

If the access symbiosis is one underappreciated engine behind American science, will it remain so into the future? Given recent trends in American postsecondary education there are many challenges, perhaps none more salient than increased financial costs, renewed exclusionary elitism, and limits to growth. Rising financial costs of undergraduate education and student loan debt threaten the long-term sustainability of the process. All century long, a component of the access symbiosis was privately paid tuition, but there were numerous lower cost options at large public institutions that kept higher education affordable for many families. Responding to major declines in public funding for universities, average costs of postsecondary attendance have grown and may limit undergraduate enrollments. Therefore, students are accruing greater amounts of debt to pay the higher costs of earning baccalaureate degrees, and students with even relatively small amounts of loan debt (e.g. $5,000) have lower odds of applying to graduate school (Millett 2003).

Often as a sector of education expands well above 50% of age cohorts, stratification among institutions intensifies. While historically there always was prestige stratification among universities, there is recent evidence of a tightening association between prestige and the socioeconomic standing of families of attendees (Hearn and Rosinger 2014; Marginson 2016a). The “California Idea of higher education,” which incorporated free or low-cost access to flagship universities, may have too many opposing cultural forces to remain viable into the future (Marginson 2016b).

The U.S. model of mass graduate education likely was, and is, not exactly transferable to STEM+ hubs in other national or regional contexts. Higher education policies must attend to global, national, regional, and local dynamics, and it is difficult to pluck deep cultural characteristics out of one nation (Marginson and Rhoades 2002). As noted, mass STEM+ doctoral education in the U.S. was not the result of coordinated national policy efforts, instead, the identified elements of the access symbiosis are dependent on an agglomeration of national dynamics. These include increasing supply and demand pressures between students and universities, as well as flexible arrangements between government and universities. For example, scholars would need to include national and regional contextual factors if they are to consider how a limited access symbiosis in a country such as China increased it rapidly growing science capacity (Zhang, Sun, and Bao 2017).

At the same time, there is evidence that the expansion of university attendance in many regions of the world, regardless of how it occurred, is one component of the rising scientific capacity behind enormous output of the Mega Science period (Powell et al., 2017). For a host of reasons still to be explored, the U.S. was the earliest and perhaps the most extreme example of what became a general world trend. Even though Germany tried to decouple instruction from research production by using two organizational forms (universities and independent research institutes), it has been the faculty-scientists of the universities who were responsible for a significant majority of the scientific output of the world’s third most scientifically productive nation (Dusdal et al. 2020). Now rivalling the U.S. in STEM+ publications, countries in Asia—South Korea, Japan, Taiwan, and China—all rapidly expanded secondary and then postsecondary enrollments and universities supporting massive growth in science publishing over just the last few decades. For example, emerging evidence that expansion of higher education, compared to excellence funding to stimulate elite universities, ends up being a partial access symbiosis that increased science capacity in Taiwan and South Korea (Fu, Baker, Zhang 2018; Kim 2019; Nam 2020).

Lastly, all rapidly expanding processes, particularly societal ones, have upper limits to growth, and this is certainly true for the American postsecondary system. Even though later modest increase is of a very large base, growth rates of all the components of the access symbiosis have declined from highs during the middle decades. Indeed, Big Science, owing so much to the U.S.’s contribution, was widely predicted to hit natural upper limits by about the 1980s, these predictions, of course, underestimated the global carrying capacity of science. Even newcomers to a kind of access symbiosis (e.g., South Korea and Taiwan) will eventually face the same challenges as the U.S. Nevertheless, American university-based scientists still contribute the largest share to Mega-Science, although financial constraints and limits of growth trends could cause the supporting symbiosis to break down. For example, several decades of a low fertility rate is already limiting postsecondary development in South Korea and foreshadowing challenges to maintain its robust science capacity. Yet, regardless of what occurs in the near future, it is clear that the access symbiosis is one central component of the U.S.’s, and likely other countries, crucial contributions to the Century of Science.