Increasing student performance in math, science, and other STEM subjects is an urgent—but often overshadowed—objective for policymakers, often tied to economic growth, international competitiveness, and national security. An effective and robust STEM teacher workforce is critical in providing students access to a strong STEM education.
Yet the STEM teacher workforce is facing strong headwinds as the pipeline of new teachers has been struggling to meet demand. Staffing pressures for STEM teachers have been increasing on multiple fronts over time: dwindling cohorts of graduates from teacher training programs, more classes to cover due to increased STEM requirements in the high school curriculum, and growing compensation for STEM degrees outside of the classroom luring teachers away. This combination of factors is believed to contribute to the staffing problems often observed among STEM teachers, including excessive turnover and low levels of teacher qualifications. Reports of teacher shortages are, often, primarily dominated by STEM teacher shortages, as this specialty (along with special education) accounts for the lion’s share of hard-to-staff vacancies nationwide. These ongoing challenges make a focus on the STEM teacher workforce particularly worthy of policy attention.
Schools serving high-need student populations often report more acute staffing challenges in STEM. Prior research has documented how teachers sort across schools, toward more affluent student populations and the resources available there. Consequently, schools with the greatest needs often have teacher workforces that are, on average, less experienced, qualified, and effective, making the STEM teacher workforce especially vulnerable in these settings. Moreover, a weak STEM teacher workforce in disadvantaged settings has policy-relevant consequences as it implicitly provides fewer opportunities for these students by blocking access to STEM instruction and related high-paying, high-demand jobs of the future.
The Robert Noyce Teacher Scholarship Program (Noyce program) is an initiative of the National Science Foundation (NSF) that aims to develop a STEM teacher pipeline in high-need settings. The Noyce program, first authorized in 2002, is named after the inventor of the integrated circuit and “the mayor of Silicon Valley.” The program institutionalizes Noyce’s desire to ensure that all young people, particularly those from impoverished backgrounds, had opportunities to learn in STEM fields and flourish in the high-tech age that his inventions helped usher in. In the two-plus decades since its inception, the Noyce program has awarded more than $1.6 billion through more than 1,400 grants to teacher-training institutions, scholarship recipients, and researchers in all 50 states, D.C., Puerto Rico, and the U.S. Virgin Islands to strengthen the STEM teaching workforce.
The Noyce program funds the development of interdisciplinary partnerships at universities between teacher education programs and faculty in academic STEM fields to promote high-quality instruction content alongside pedagogical and practicum work for preservice teacher candidates (hereafter, colleges instituting these partnerships are referred to as Noyce institutions). Additionally, individual students at Noyce institutions can apply for scholarship funds to help support their completion of the teacher training program leading to licensure in a STEM field, provided that the candidates agree to a service obligation to teach in a high-need school district upon graduation (hereafter Noyce scholars).
This report summarizes the findings from a multiyear, collaborative research project conducted by researchers at the Brookings Institution, Texas State University, and Florida Atlantic University. They were joined by faculty at four collaborating Noyce institutions located at the University of Texas at Arlington, Florida International University, Texas State University and the University of West Florida. The research team explored questions that were organized around three primary objectives. Objective 1 explored the state of the STEM teacher workforce in high-need settings through descriptive analysis of national survey data spanning nearly three decades. Objective 2 evaluated the Noyce program’s associations with STEM teacher qualifications and vacancies in the school districts located near Noyce programs. Objective 3 conducted a mixed-method study with the project’s four collaborating Noyce institutions to gather and analyze programmatic data, survey alumni about their experiences, and interview stakeholders to understand how these programs prepare and support STEM teachers for high-need settings. Each of the project’s objectives have produced academic papers, some of which have been published in peer-reviewed academic journals, and others are being prepared for academic publication. References to publicly released reports will be provided as relevant in the discussion below; note that authors contributing to each of the supporting analyses vary but for clarity are referred to collectively as the research team in this report.
In summary of this project’s findings, Objective 1 analyses showed a surprising stability in the qualifications and turnover of secondary STEM teachers in high-need settings, which prompted follow-up analyses to investigate how schools are coping despite the apparent staffing pressures. Objective 2 offers some encouraging evidence that participation in the Noyce program is improving STEM workforces in local school districts surrounding them. Objective 3 finds that Noyce scholars are viewed as well prepared for teaching in high-need settings. The following sections of the report briefly describe the motivation, methods, and findings corresponding to each of these three objectives. Finally, the Discussion and Conclusion section presents a synthesis of findings and observations drawing across all three objectives, contextualizing with the broader research literature. Importantly, even though the findings presented here offer encouraging news about the state of the national STEM workforce, there are many reasons to believe that local problems could still be serious and warrant ongoing surveillance and support.
Objective 1: Evaluating the state of the STEM teacher workforce in high-need schools
Objective 1 begins with a broad look at trends in the qualifications of the STEM teacher workforce over time, with a focus on high-need schools. Given the multitude of increasing pressures on the STEM teacher workforce—including fewer new teacher candidates in the pipeline and growing demand for STEM instruction—the research team hypothesized that trends in qualification levels for STEM teachers were declining over time. This hypothesis is informed by evidence showing principals and district leaders frequently cope with staffing pressures by lowering hiring standards. Further, prior research shows that high-need schools have particular challenges in maintaining a qualified teacher workforce, and the research team expected declines in these settings to be larger than those in low-need settings. Though prior research studies have explored STEM teachers’ qualifications and their turnover behavior across different settings, this was the first study to document long-term trends with a focus on those in the most high-need settings.
Data and methods
The analysis used eight waves of nationally representative survey data collected by the National Center for Education Statistics from the Schools and Staffing Surveys (SASS) and the National Teacher and Principal Surveys (NTPS) to examine teachers’ characteristics, qualifications, and school contextual variables spanning the 1993-94 to the 2020-21 school years. A teacher follow-up survey administered one year later captures turnover outcomes for a subsample of teachers (up through 2011-12 only). Supplemental data were drawn from the Common Core of Data (CCD) documenting the share of students eligible for free or reduced-price lunch (FRPL) in each school, the study’s proxy measure of student poverty. Since student FRPL shares are constantly increasing with time, the analysis constructed quartiles based on the percentage of FRPL-eligible students in schools for each survey wave. Teachers in the highest-need schools (the top 25% of schools based on FRPL shares) were compared against those in low-need settings (the bottom 25% of schools based on FRPL shares)—hereafter referred to as “high-need” and “low-need” schools.
The analysis sample was limited to secondary teachers in grades 7 to 12 who offer departmentalized instruction; math and science instruction in elementary grades is often led by teachers in self-contained classrooms, employing teachers that do not specialize in STEM instruction. STEM teachers are those who report teaching an academically oriented course in math, science, computer science, or engineering based on its course title (e.g., 8th grade math) in at least one class period in a departmentalized model of instruction (vocational courses that overlap with STEM disciplines, like health sciences, were not counted as STEM). Note that this definition is expansive, including teachers who may focus on a non-STEM subject and only teach one STEM course on the side; these teachers are intentionally included to provide a comprehensive picture of all who lead STEM instruction, not just those hired specifically for that role. Key outcome variables are compared between STEM teachers in high-need schools (the target subgroup of interest) versus STEM teachers in low-need schools and non-STEM teachers in high-need schools (when applicable); non-STEM teachers in low-need settings are different in multiple ways and omitted from most analyses below.
Finding 1: Stability in STEM teacher qualification levels in high-need schools
The first main finding was unexpected and counterintuitive: Qualifications among the STEM teacher workforce in high-need schools has been surprisingly stable despite increasing staffing pressures over the last three decades (full results available in the Journal of Research in Science Teaching). The analysis examined qualifications that are common indicators of quality STEM instruction: years of experience, certification status, holding a master’s degree, and degrees in a STEM field. The qualification levels of STEM teachers in high-need schools were generally stable over time. The share of STEM teachers holding a master’s degree even modestly increased over the period, though many of these new degrees are in subjects other than STEM fields, as the stable share of STEM degree holders indicates (see Figure 1). The share of STEM teachers in low-need settings holding a STEM degree has significantly declined over time, while decreases in high-need settings are not statistically significant.
FIGURE 1
STEM teacher qualifications by school poverty level
The stability of the workforce does not mean that the STEM teacher workforce is healthy overall. STEM teachers in high-need settings are consistently less qualified than those in low-need settings across all measures examined. The gaps in qualifications, however, appear to be driven more by the school’s high-need context rather than whether a teacher works in a STEM or non-STEM field.
Further, the analysis revealed significant variation in teachers’ qualifications across STEM fields (see Figure 2).1 In three of four STEM field categories, teachers in high-need schools show a marked improvement in the share holding an aligned credential with their actual teaching over time. Curiously, biology stands out as the one STEM field where qualification levels are now lower than in recent decades, though other work has singled out biology as the STEM field with the strongest talent pipeline. Physical sciences and computer science show the highest rates of out-of-field teaching, with 45% and 58% of teachers, respectively, having no subject-specific credential in high-need schools in 2020.2 This is juxtaposed against 22% and 34% of teachers in math and biology, respectively. These differences in qualifications across fields (comparing within low- or high-need schools) are generally larger than the differences observed across school settings (comparing low- against high-need schools within a given STEM field). Overall, this evidence suggests that a field-specific approach to the STEM teacher workforce is needed to adequately support the teacher pipeline.
FIGURE 2
Share of teachers with field-specific qualifications by school poverty level
Finding 2: STEM teacher turnover also appears stable over time, though surveys suggest worrisome signs of stress
The research team also explored STEM teachers’ turnover patterns in high-need schools. Most prior work on the teacher workforce shows that, independently of the other, being a STEM teacher or working in high-need schools is associated with an elevated likelihood of turnover. The team was particularly interested in whether the combination of being a STEM teacher in a high-need school was associated with even higher levels of attrition beyond those expected when examining each isolated variable.
Teachers’ actual turnover decisions—including whether they moved to another school or exited the teaching workforce entirely—are only available through the 2011-12 survey wave; follow-up surveys were discontinued. This limits the team’s ability to observe recent turnover outcomes. Up to 2011-12, however, results showed that STEM teachers in high-need schools were both slightly more likely to leave teaching and move to another school in most years (relative to comparison subgroups), though differences between the teacher groups were generally not statistically significant (see Figure 3). There is a modest upward trend in the likelihood of moving to another school for both STEM and non-STEM teachers from high-need settings. No persistent trend is observed in the frequency of leaving the profession over time, nor are differences statistically significant across comparison subgroups.
FIGURE 3
STEM teacher turnover seems stable between 1993 and 2011
The follow-up survey also captured reasons for departure among those who left teaching. STEM teachers in high-need schools stood out from the other teacher subgroups (see Figure 4). Retirement or disability represented the most common reason for departure for all groups, though it motivated far fewer STEM teachers in high-need settings to leave than others. Instead, these teachers showed a far higher share of individuals leaving teaching to further their education, amounting to nearly 15% of departing teachers (compared to 5% or less of other comparison subgroups). At minimum, this evidence points to a sizeable minority of STEM teachers in these settings being younger and potentially driven by motivations that diverge from comparison group norms.
In contrast to the truncated series of actual turnover behavior described above, teachers’ attitudes about their jobs were collected throughout the entire span of surveys. Recent studies have begun to view these attitudes as predictive (though imperfect) indicators of future turnover behavior. The study analyzed two attitudinal variables, which showed divergent results (see Figure 5). The share of teachers indicating they intend to leave teaching as soon as possible is reported infrequently (less than 5% of teachers most years) and also appears to be the most predictive of future departure from teaching. This measure has shown remarkable stability over time, especially since the 2007-08 survey, showing no significant differences across the three teacher subgroups. Conversely, the share of teachers reporting they would leave teaching for a higher-paying job (a more hypothetical question) showed not only a higher level of agreement among teachers but also a consistent, upward trend over time. A large and statistically significant difference is consistently reported between teachers in high-need schools (both STEM and non-STEM teachers) and those in low-need schools. Thus, these changes in teachers’ commitment to teaching may be indicative of a growing underlying fragility, despite the most concrete indicators of turnover appearing to be stable.
FIGURE 5
Teachers’ attitudes on their commitment to the profession
Finding 3: Teachers face large wage penalties in high-poverty schools, outstripping those associated with STEM backgrounds
The project also investigated teachers’ compensating wage differentials—the average salary premium (or penalty) teachers earn based on their assignments and school contexts (available as an Annenberg working paper). Theoretically, hard-to-staff positions, like those in STEM fields or in high-need settings, should command higher salaries. Yet common practices in the K-12 teacher labor market, including the widespread use of single salary schedules for all teachers, are expected to limit the responsiveness of teachers’ earnings to labor market pressures.
The study’s analysis found that K-12 teachers who hold a STEM bachelor’s degree earn a modest salary premium overall—2.77% higher than the salary average—compared to those not holding a STEM degree, all else equal. However, the estimated salary premium shrinks for STEM teachers (and non-STEM teachers) in schools that have higher shares of FRPL-eligible students.
Figure 6 below shows the regression-adjusted differences in salaries for STEM and non-STEM teachers in various settings relative to non-STEM teachers in settings with both low shares of students of color and FRPL-eligible students (baseline teachers, in the top left cell of the figure). As one moves to the right of the columns (increasing school poverty based on FRPL shares), STEM teachers earn significantly less than counterparts in lower-poverty settings. Also, STEM teachers face a larger earnings penalty from teaching in higher-poverty settings than their non-STEM peers in similar settings (18.3% and 12.9% less than baseline teachers, respectively). Further, average salaries seem to increase as the share of students of color in a school increase, which is heavily moderated by poverty in the school, roughly canceling this salary premium when poverty is high (see bottom-right cells). Together, these findings highlight the structural barriers to recruiting and retaining STEM-qualified teachers in high-need schools, where contextual disadvantages erode the modest earnings premium for teachers with STEM degrees.
Finding 4: High-need schools are increasingly reliant on foreign teachers, targeted district policies to cope with pressures
The surprising stability of STEM teacher qualifications and turnover prompted a follow-up investigation: How have high-need schools maintained their STEM teacher workforces despite external pressures? If schools are not lowering their qualifications standards in response to staffing pressures, schools facing staffing challenges must then adopt other coping strategies. Coping strategies are actions that schools adopt in the face of staffing pressures that allow them to manage instruction when qualified candidates are unavailable. Prior surveys of school leaders facing staffing pressures were referenced to identify a menu of 10 common coping strategies (beyond lowering hiring standards). The analysis found evidence that two of the 10 coping strategies examined have notably shifted in their observed frequency over time, showing a statistically significant divergence between high- and low-need schools.
First, high-need schools have grown increasingly reliant on hiring teachers trained at foreign colleges. This is a variable available in the teacher survey data selected as a proxy to represent individuals recruited to teach on temporary work visas. News and advocacy group reports spanning many years have documented school districts turning to foreign labor sources to fill teacher labor shortages. In the earliest survey wave, foreign-trained teachers accounted for approximately 1% of the STEM workforce in high-need settings but multiplied to exceed 6% by 2020-21. Importantly, foreign-trained STEM teachers in the study data are more qualified than their U.S.-trained counterparts: 79.3% hold a STEM degree (compared to 67.5% of U.S.-trained STEM teachers) and 65.7% hold a master’s degree or higher (compared to 56.6%). Thus, they appear to be critical in buoying the qualifications of the STEM teacher workforce in high-need schools.
Second, the analysis also showed district policies supporting teachers in hard-to-staff subjects and settings also increased in frequency over time. Though the policy variables analyzed for this section were only captured in a few survey waves following the passage of the No Child Left Behind Act, they were an active area of reform at the time. Notably, the frequency of district policies providing incentives or supports targeted to subjects with shortages and teaching in less-desirable locations significantly increased among high-need schools between the first observation in 2003-04 and the final available observation in 2011-12; consequently, gaps between low- and high-need schools grew across these categories (see Table 1). Conversely, other teacher policies that did not promote efforts showed a declining frequency over this time.
Note that schools often do use other coping strategies beyond these two to offer STEM instruction despite the lack of qualified teachers. The remaining eight coping strategies that were explored in detail in the analysis (but not flagged in this summary report) are still evidently adopted in schools, though their frequency of adoption has not significantly changed over time or along the STEM/non-STEM axis to warrant its identification. For example, the study’s analysis found STEM teachers in high-need schools are less likely to offer instruction in advanced STEM courses (25.6% of STEM teachers) than those in low-need schools (35.5%) in the most recent survey. Differences in access to courses were largest in calculus and physics.
Objective 2: Evaluating the association between Noyce institutions and local STEM teacher workforces
Having assessed the state of the STEM teacher workforce broadly in the prior section, this objective narrows the scope of the investigation to focus on the Noyce program’s possible impacts on local STEM teacher workforces. Several operational details about the Noyce program inform key outcomes for this research objective. For example, Noyce institutions are expected to develop interdisciplinary relationships between education departments and STEM faculty. Further, scholarships are intended to attract individuals into teaching, and Noyce scholars are obligated to teach in high-need school districts (this is a difference from the focus on high-need schools in Objective 1, discussed further below). If these features of the Noyce program have the intended effect, then nearby school districts located close to the program should have STEM workforces with stronger qualifications that turn over less frequently and a robust supply of qualified new teachers ready to fill any new vacancies. These and similar STEM workforce outcomes are those that will be analyzed over time to estimate their association with the presence of a Noyce institution in the area.
Most of the existing Noyce literature has not directly addressed these workforce questions. Many prior studies of the Noyce program focus on a single Noyce institution and generally have a small sample size. Two larger-scale studies find that Noyce scholars are more likely to be retained in teaching and to teach in high-need settings after graduation. The most comprehensive study to date focuses on Noyce programs across Texas and finds that while Noyce teachers are more likely to work in high-need schools and are more effective, they are more likely to switch out of these schools or leave teaching entirely, relative to other teachers from their universities. Yet whether some of these patterns bear out at scale is an empirical question that has not been directly addressed. Furthermore, candidates produced from Noyce institutions represent only one of many potential sources for STEM teachers; thus, the magnitude of any Noyce institution’s new STEM teacher supply relative to the local market will also factor into the strength of the estimated association. This analysis addresses these gaps in the literature by investigating whether high-need districts near Noyce institutions face fewer staffing challenges among their STEM teacher workforces.
Data and methods
The primary data are drawn from the same series of nationally representative teacher surveys used in Objective 1 spanning nearly three decades. This study utilizes information on STEM teachers, their qualifications, the school contexts, and their turnover decisions in the follow-up survey collected one year later as key measures. Additionally, questionnaires that were simultaneously administered to school staff as part of these survey efforts collected data on teacher vacancies that schools posted in the period just before taking the survey and the difficulty filling them (by subject specialty). Together, these variables provide information about the level of STEM teacher qualifications, their turnover, teacher vacancies, and the difficulty in filling those vacancies.
Proximity to a Noyce institution is the treatment of interest for this analysis, an assumption bolstered by prior research showing that teacher labor markets are geographically very small and that a candidate’s student teaching location is highly predictive of their first job placement. To capture information about the presence of Noyce institutions, key data were extracted from the NSF awardee web portal. The publicly accessible portal contains information on all funded Noyce institutions, the years of operation under the Noyce program, and the number of expected Noyce scholars (based on the initial proposals).3 This provides all the relevant time and place information for Noyce institutions, which were merged with the survey data using geolocation data in both sources. The Noyce program’s definition of high-need school districts has varied over time, starting with a very inclusive definition in the early years to a more restricted one today. For this study, a high-need school district (HNSD) is defined as having an FRPL student share exceeding 50%. Given the Noyce program’s focus on high-need school districts, the research design also compares differences among the STEM teacher workforces in high-need school districts near Noyce institutions (comparing against those in districts not considered high-need).
The analysis estimates a regression model where the outcomes of interest are the labor market variables including STEM teacher qualifications, turnover, and STEM vacancies. The explanatory variables are indicators of universities that eventually became Noyce institutions, the years following approval as a Noyce institution, and high-need school districts, plus other school and district control variables. It is possible that Noyce institutions may be unique on other dimensions that cannot be directly controlled in the model (e.g., having a strong teacher training program, concurrent adoption of other programs). Consequently, the results presented here do not reflect causal relationships.
Finding 1: High-need school districts near Noyce institutions are more likely to employ teachers with a STEM degree
The first key finding showed that the presence of an operating Noyce institution is associated with more qualified STEM teachers in nearby high-need school districts (see Figure 7). STEM teachers in high-need school districts were about 1.5 to 1.6 percentage points more likely to hold a STEM degree following the introduction of a Noyce institution in the local area (notated in the figure as Treatment x Post x HNSD), controlling for other teacher and school characteristics. The analytical model that uses state fixed effects showed a statistically significant estimate; the estimate from a model with a district fixed effect was slightly smaller and not statistically significant. Meanwhile, no significant differences were observed in the qualifications of nearby non-high-need school districts, suggesting that the Noyce program’s high-need focus may have played a role in narrowing qualification gaps across settings.
The research team conducted additional analyses on STEM teacher turnover, considering the district’s proximity to an active Noyce institution. Again, though the turnover data has some limitations due to data availability, the team considered both actual turnover based on the follow-up survey (for the years available) and a turnover indicator based on teachers’ willingness to become a teacher again. Neither outcome showed statistically significant differences among STEM teachers in nearby high-need school districts.
Finding 2: Fewer STEM vacancies in nearby school districts, and those vacancies tend to be easier to fill
Next, the team analyzed schools’ reports of STEM teacher vacancies. The results showed the introduction of a nearby Noyce institution is associated with a statistically significant reduction in the number of open teacher vacancies in all three STEM fields in the data: math, physical sciences, and biology (see Figure 8). The results vary modestly across STEM fields, with the largest reduction of nearly 8 percentage points for math vacancies, though all fields point in the same direction of fewer vacancies. It is unclear why nearby districts have fewer STEM vacancies when there is no evident association with STEM teacher turnover, as described above; plausible reasons for this discrepancy may be due to data availability (vacancy data and observed turnover are measured in many, but not all, survey waves) or differences in enrollment growth patterns.
Furthermore, when analyzing the difficulty filling these STEM vacancies, nearby school districts reported significantly lower levels of difficulty for all three STEM fields. The difficulty-to-fill estimates, however, show little apparent variation across STEM fields. Combined, the Objective 2 results presented here suggest the presence of a Noyce institution helps to support the STEM teacher pipeline in nearby school districts, especially those serving high-need student bodies.
Objective 3: A mixed-methods study looking inside Noyce institutions
This final objective narrows the study’s scope even further to Noyce institutions themselves. The research team worked closely with four collaborating Noyce institutions in Texas and Florida that have established track records of producing Noyce scholars and collectively serve diverse geographical labor markets. The purpose of this investigation was to learn about the operations of each institution and whether and how they were adapted for their settings. Further, the research team was interested in understanding how the effectiveness of the local Noyce institution was perceived based on the views of Noyce scholar alumni, local school districts that hire program graduates, or other adjacent stakeholders in the university.
Data and methods
Programmatic data were collected from the four collaborating Noyce institutions in 2021. The information gathered included recruiting and application processes for the Noyce scholarship, required coursework and experiences, and supporting features for students during both preservice preparation and after completing their training. Additional data probed Noyce institutions’ interactions with area school districts for student teaching and employment placements, as well as opportunities to gather feedback on these points (a full report is available in the Journal of Science Education and Technology).
An alumni survey was designed to gather information on alumni’s preservice interest in the teaching profession, their experience as Noyce scholars, and their commitment to teaching. This survey was implemented during 2022 to 226 alumni with a 46.5% response rate. Additionally, the team conducted interviews with 19 adjacent stakeholders to the Noyce institutions in 2023. These stakeholders held a variety of roles including cooperating teachers, school leaders where Noyce scholars completed their student teaching, human resources personnel in the hiring school district offices, and program support personnel from the four collaborating universities.
This analysis builds on an extensive literature review that selected and analyzed prior Noyce program evaluation reports. Consistent Noyce program objectives emerged from the review, namely: to train teachers that are effective at teaching STEM subjects in high-need schools and to provide mentoring and other supports to ensure program graduates improved and were retained during their early teaching years. Another theme that was echoed across the evaluations was the importance of the Noyce scholarship funds in attracting prospective students. Items in the alumni survey were motivated by similar questions from the prior Noyce literature. For example, one item in the present study’s alumni survey asked respondents to identify the point at which they seriously considered becoming a teacher (see Figure 9). A large majority of respondents (71%) indicated that their occupational choice was made well before applying to the Noyce scholarship program; however, a sizable minority indicated decision points based on either the availability of Noyce scholarship funds or early teaching experiences that are frequently used in Noyce programs to recruit potential candidates.
After gathering data across these various sources (programmatic data, interviews with stakeholders, and alumni surveys), the research team looked for common themes triangulated across them. The three findings presented below are those that surfaced from this analysis.
Finding 1: Noyce institutions provided strong opportunities for preservice preparation, tailored to high-need settings
Analysis of the programmatic data revealed that Noyce institutions are providing preparation opportunities rich with teaching and training experiences that are tailored for the program’s intended high-need setting. All institutions required scholars to have meaningful preservice experiences that included student teaching and opportunities to work on a limited basis in a high-need school (see Figure 10).4 Preparation for a high-need setting was also a common preparation theme, where three of four programs required scholars to participate in specific development around culture, diversity, and inclusion, and provided opportunities for interaction with children from different cultures; two of four institutions required the scholars’ student teaching experience to be completed in a high-need school. The remainder of programs offered these activities but did not require them, allowing for some flexibility in how scholars engaged with these preparation experiences.
Committing to teach in a high-need school district is an explicit requirement to receive the Noyce scholarship. Prior research shows that teachers also differently sort across schools according to race: Teachers of color are significantly more likely to teach in school settings that serve high shares of students of color, which also tend to be in high-need settings. This requirement may, therefore, implicitly attract more individuals of color into teaching or may shift white candidates’ preferences about where they teach. However, alumni survey responses did not reveal notable differences by scholar race. About 85% of scholars of color and 82% of white scholars indicated they would have considered teaching in a high-need school even without receiving the Noyce scholarship (see Figure 11). Thus, the program may attract candidates already disposed toward serving in high-need settings, and the scholarship funds simply provide the financial and preservice support that enables them to act on that disposition.
Stakeholder interviews affirmed that Noyce scholars demonstrate strong skills in both content knowledge and pedagogical practice and are well prepared to teach in high-need settings. Most interview participants felt this was a clear area of strength for graduates of Noyce institutions. Alumni themselves felt well prepared by their programs (based on survey results), and districts and schools receiving Noyce graduates echoed this sentiment (in interviews), indicating that the scholars were appropriately prepared for the classroom. Noyce scholars were welcomed in schools and shaped a positive image of the Noyce scholarship program among hiring districts.
Finding 2: Ongoing supports to Noyce scholars identified as critical elements
Across all data sources, support systems for Noyce scholars emerged as a critical theme, though with notable variation in how support was conceptualized and delivered. In the programmatic data, all institutions claimed to offer (and some required) ongoing support following Noyce scholars’ graduation (see Figure 12).
However, the perceptions and meaning of mentoring supports appeared to shift depending on who was being asked. For example, there was no mention of a mentoring program that directly led to the success or retention of scholars, nor was there a common definition of mentoring used by participants across programs. Interview findings revealed that some school-based participants felt feedback was beneficial and appreciated the classroom visits from university supervisors. Others felt that ongoing support from Noyce institutions was inconsistent or lacking.
Data underscored the importance of Noyce funding as a support mechanism. All program coordinators reported that Noyce funding improved their capacity to provide a reliable source of STEM teachers to partner districts, improving their working relationships. Also, the data showed that Noyce funding was associated with the retention of teachers in high-need schools, suggesting the financial support enabled scholars to persist in teaching through the challenging early years.
Finding 3: Communication between Noyce institutions and local school districts demonstrated a discrepancy with stakeholder beliefs
Despite the strengths identified in scholar preparation and support systems, perceptions about communication success between Noyce institutions and their partner school districts differed across the two sources. From the perspective of Noyce institutions, communication with local school districts is productive and spans multiple domains (see Figure 13). Programs reported ongoing communication with teachers supervising Noyce graduates, coordination on certification requirements, and provision of advice as needed.
District-based interview participants, however, spoke of limited communication and several were unaware that some of their student teachers were Noyce scholars. The research team observed there was little mention of substantive engagement with universities other than transactional correspondence around placement. `
The placement process itself also reflected some communication challenges. Alumni expressed mixed satisfaction with their placement experience, and stakeholder interviews identified specific challenges that interfered with scholars’ placement in high-need settings. Thus, the institutional linkages between universities and school districts appeared underdeveloped. This represents an area of potential growth, as stronger communication between Noyce institutions and nearby hiring school districts could facilitate better placement matching, more targeted support for scholars, and feedback loops to continuously improve their program’s approach.
Discussion and conclusion
This multiyear collaborative research project offers several contributions to the field’s understanding of the STEM teacher workforce and the role of the Robert Noyce Teacher Scholarship Program in supporting it. Several themes emerging from this work warrant further discussion.
Perhaps the most striking finding is the unexpected stability of STEM teacher qualifications in high-need schools. Despite mounting pressures—declining teacher preparation program enrollments, increased STEM curriculum requirements in schools, and higher outside earnings for STEM graduates—the qualifications of STEM teachers in high-need schools have remained stable or slightly improved on multiple dimensions, narrowing qualification gaps relative to low-need schools in some cases. In three of four STEM fields, teachers in high-need schools were more likely in 2020 to hold some credential aligned with their classroom instruction than they were in 2003 (Figure 2). Turnover outcomes among STEM teachers were also generally stable over time, showing no clear differences compared with other teacher groups (Figure 3). These turnover findings run counter to other research—notably, two recent empirical studies that conclude STEM teachers in high-need settings are significantly more likely to both move and exit teaching (though methodological details may account for the discrepancy).
This theme of overall stability is especially surprising given the financial barriers that STEM teachers face. Figure 6 shows that STEM teachers in disadvantaged settings face salary penalties rather than the premiums predicted by theory. Recent decades also coincide with an increasing “teacher pay penalty” relative to earnings in other professions. These two penalties combined should theoretically indicate greater staffing pressures, yet stability is observed across multiple outcomes.
The surprising workforce stability may reflect the cumulative impact of targeted policy interventions intended to counteract market pressures drawing STEM talent away from teaching in general and away from high-need settings in particular. The Objective 1 investigation of coping strategies sheds light on the increased implementation of school district policies offering targeted incentives for teachers in shortage fields and in hard-to-staff locations (Table 1). Though this is an admittedly narrow domain of policy interventions, other sources also point to similar increases for targeted policy interventions. For example, over 30 states have now established pathways and financial incentives for teachers in hard-to-staff schools or subjects and most large districts have adopted more differentiated compensation policies.
The resilience of STEM teachers’ qualification levels appear to be due in part to the growing role of foreign-trained teachers in shoring up the teacher corps, especially in high-need schools. This finding corroborates other scholarship that documents U.S. schools’ increasing reliance on labor from developing countries (the Philippines is a major source country) to cope with policy pressures from the No Child Left Behind Act. The U.S. is not alone, either; other industrialized countries, including Australia and Ireland, have also turned to foreign teachers in staffing their classrooms. Yet this reliance implies a growing vulnerability to events that could disrupt this labor supply, including pandemics and diplomatic relations. Protecting this pipeline of STEM talent should be a priority to maintain instruction in high-need schools.
Another reason for workforce stability could be due to the expansion of programs seeking to prepare STEM teachers to work in high-need settings. The results from Objective 2 provide suggestive evidence that the Noyce program is achieving its intended effects. High-need school districts near Noyce institutions show increased shares of STEM-trained teachers (Figure 7). Nearby districts showed reduced vacancy rates and reported fewer difficulties filling STEM positions (Figure 8). It would be premature, however, to conclude that the estimates are due to the presence of the Noyce program only, as it is not alone in attempting to shore up teacher pipelines. For example, UTeach is also a program that supports collaboration between STEM university faculty and those in teacher training programs, recruits individuals into K-12 STEM teaching, and has a national scope that began within the period of this study. It has also influenced STEM teacher supply: In 2021, an estimated 6% of newly trained STEM teachers came from UTeach programs. Further, many Noyce institutions, including some of the project’s collaborators, also participate in UTeach; thus, isolating the causal effect of Noyce separate from UTeach is impossible with the data available. Additionally, Beyond100K (launched in 2011 as 100Kin10) is a national network initiative that mobilizes organizations to prepare, retain, and support STEM educators; its work also appears to have helped maintain and expand the STEM teacher pipeline in recent years.5 These national initiatives, combined with the increasingly robust (if fractured) policy landscape on service scholarship and loan forgiveness programs targeting high-need subjects and settings, further bolster the claim that targeted supports are making a difference at scale in supporting the STEM teacher pipeline.
The mixed-methods study of collaborating Noyce institutions in Objective 3 illuminated both strengths and areas for improvement in program implementation. Noyce scholars are esteemed as well prepared in content and pedagogy, build strong relationships with peers, and are welcomed by schools. Their apparent readiness for working in high-need settings conflicts with other recent work on the Noyce program that found Noyce scholars were often underprepared for such settings. Communication between Noyce institutions and school districts, however, emerged as an area warranting attention. Further, the finding that most Noyce scholars had already considered teaching before applying (Figure 9) suggests the program is most effective at supporting and retaining committed individuals rather than recruiting new candidates to the profession. Increasing the scholarship amount or otherwise modifying the terms of eligibility may help enhance the program’s external attractiveness, expanding its reach to more prospective teacher candidates.
Yet continued policy and research attention is needed for the STEM teacher workforce, despite the relative stability overall. Large and persistent gaps in subject qualifications between STEM teachers in high-need schools and their counterparts in low-need schools remain (Figure 1). Moreover, the stark differences in qualification rates across STEM fields—with physical sciences and computer science showing particularly high rates of out-of-field instruction—suggest STEM instruction in these fields is likely underwhelming for most students regardless of their school setting (Figure 2). These persistent gaps indicate that whatever supports are helping to maintain the STEM teacher workforce now, they are inadequate to fully address the need for qualified STEM instruction for all students.
The STEM workforce in high-need schools also shows unique signs of underlying fragility. Not only is it increasingly reliant on foreign-trained teachers and many who express a growing willingness to leave for more pay, but it is also reliant on a disproportionately large share of young teachers who leave the classroom to pursue further their own education. This may signal the presence of Teach For America (TFA), though TFA affiliations are not captured in the data. TFA is a nonprofit staffing organization that recruits promising young professionals to teach for a two-year commitment in high-need schools, disproportionately in STEM fields. They also exhibit high turnover with 85% leaving the profession within five years, nearly a quarter of which leave to further their studies for career aspirations (like the patterns shown in Figure 4). And on a cautionary note, TFA peaked in 2013 with nearly 6,000 new teachers placed and shrunk by two-thirds over the next 10 years, demonstrating how quickly these labor sources may collapse; low-quality alternative certification providers appear to be expanding quickly to fill the gap.
The landscape of alternative certification programs unaffiliated with institutions of higher education has quickly grown in recent years. This is the only category of teacher training programs that has experienced growth over the last decade (traditional or university-based alternative programs have shrunk). Recent evidence from Texas, which accounts for an outsized share of these newly trained teachers, indicates they have significantly lower value added for students in comparison to those from traditional or nonprofit alternative certification programs; the study also highlights explosive growth in the share of uncertified teachers starting in 2022. These developing trends in Texas certification highlight several limitations of this project’s findings. First, the national scope employed here does not necessarily represent the health of the workforce in any given setting, and some localities or states may vary quite dramatically from these general patterns. Second, the survey data is only available through the 2020-21 school year and newer developments since then are unobservable. And third, not all credentials for teaching should be considered interchangeable. Texas’ recent experience shows how dramatically some locations may depart from the overall patterns documented here, how quickly major changes may happen, and the folly of failing to differentiate across various pathways into teaching.
In summary, the evidence presented here shows both optimism and caution are warranted regarding the status of STEM teaching in high-need settings. Optimism stems from the overall stability of teacher qualifications and turnover rates, as well as the creative coping strategies schools employ despite increasing pressures. This optimism extends to the Noyce program and adjacent initiatives—both public and private—that are shoring up the STEM teacher pipeline. However, caution is warranted due to persistent gaps in access to qualified teachers and the various fragilities exposed within the STEM teacher workforce. Current pay and incentive structures appear to be insufficient to address these vulnerabilities. Rather, more substantial financial supports—whether through direct salary differentials targeted to high-need settings and hard-to-staff subjects, loan forgiveness, or enhancing scholarship programs like Noyce—appear to be necessary to both improve the STEM teacher workforce overall and achieve an equitable distribution of teaching talent.
The Brookings Institution is committed to quality, independence, and impact.
We are supported by a diverse array of funders. In line with our values and policies, each Brookings publication represents the sole views of its author(s).
