Subscribe Today
Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13

Examining the Recruitment, Placement, and Career Trajectories of Secondary Mathematics Teachers Prepared for High-Need Schools

by William Zahner, Suzanne Chapin, Rich Levine, Lingjun A. He & Robert Afonso - 2019

Background: School leaders are challenged by the relatively limited supply and high turnover of qualified secondary school mathematics teachers. In response, policy makers and teacher educators have developed various pathways and incentives to recruit, train, place, and support highly qualified mathematics teachers to work in hard-to-staff schools.

Focus of Study: In this study, we investigate the recruitment, placement, and early career trajectories of 158 Grades 6–12 mathematics teachers from two preparation programs focused on staffing “high-need” schools in the same region.

Setting: The contrasting programs were both supported by the same university in the Northeast United States.

Participants & Programs: The participants were 158 secondary school (Grades 6–12) mathematics teachers. Of these, 48 were recruited and prepared through a teacher education program with financial support from the National Science Foundation-funded Robert Noyce Teacher Scholarship Program. The other 110 school mathematics teachers were recruited and trained through the Greater Boston office of Teach For America. Both programs required two years of service in high-need schools.

Research Design: In this study, we used a comparative design. Descriptive profiles of teachers from each program were created. Then, participants’ early career trajectories were compared using logistic regression and survival analysis.

Data Collection and Analysis: We administered a longitudinal survey and created a database combining survey data and each program’s administrative data.

Findings/Results: The Noyce scholarship-supported pathway successfully recruited individuals with science, technology, engineering and mathematics (STEM) majors, trained them to be mathematics teachers, and placed those individuals as secondary mathematics teachers in high-need schools. The TFA-recruited secondary school mathematics teachers were less likely to have STEM majors than their counterparts in the scholarship pathway, and their attrition rate after completing their service requirement was higher than that of the scholarship-supported teachers. However, TFA recruited a more diverse pool of potential teachers and placed these teachers in schools serving a higher proportion of low-socioeconomic-status students.

Conclusions and Recommendations: This comparison highlights how each program’s design likely attracted prospective teachers who had different long-term career goals before they entered their preparation pathway. The results suggest that policymakers who seek to address staffing shortages in secondary mathematics must balance recruitment criteria, school working conditions, and prospective teachers’ career goals while designing pathways to recruit qualified mathematics teachers for hard-to-staff schools.

The strongest predictor of student learning under the control of school systems is the quality of a students teacher (Darling-Hammond, 2000; Monk, 1994). However, the relatively limited supply of qualified secondary school mathematics teachers challenges school leaders, policy makers, and politicians who seek to increase the quality of school mathematics teaching and learning (National Audit Office, 2016; National Research Council, 2010; Organisation for Economic Co-operation and Development [OECD], 2005; Presidents Council of Advisors on Science and Technology, 2010). At the secondary level, most school systems require mathematics teachers to have extensive subject-specific training (Tatto & Senk, 2011), and school administrators often struggle to find prospective teachers with the required educational background in both mathematics and pedagogy (Guarino, Santibanez, & Daley, 2006; Ingersoll & Perda, 2010). The unfortunate result of this struggle is that the secondary school mathematics teachers with the fewest qualifications are more likely than well-qualified teachers to teach in schools that serve economically disadvantaged students and students who are members of minority groups underrepresented in technical careers (Lankford, Loeb, & Wyckoff, 2002; Peske & Haycock, 2006). Because mathematics functions as a gatekeeper to higher education or technical careers, inequitable access to highly qualified mathematics teachers hinders students educational and economic opportunities (Moses & Cobb, 2001).

In response to persistent and growing shortages of teachers in critical areas such as mathematics, science, and special education (e.g., Darling-Hammond, Furger, Shields, & Sutcher, 2016), teacher educators and policy makers have designed pathways to recruit, train, place, and support highly qualified mathematics teachers to work in hard-to-staff schools (National Audit Office, 2016; National Research Council, 2010). One model for such pathways is to provide new teachers from university-based teacher preparation programs with incentives such as scholarships, loan forgiveness, and salary supplements for agreeing to teach in hard-to-staff schools (Fulbeck & Richards, 2015; Steele, Murnane, & Willett, 2010). Another model pathway involves schools or districts in partnering with alternative certification programs to recruit and train teachers. Alternative certification programs include independent nonprofits such as Teach For America (TFA) in the United States or Teach First in the United Kingdom, as well as internship programs run by schools and school districts (Boyd et al., 2012; Kane, Rockoff, & Staiger, 2006; Papay, West, Fullerton, & Kane, 2012).

We report here on the recruitment, training, and initial career trajectories of 158 secondary school (Grades 612) mathematics teachers who started teaching between 2009 and 2014. These teachers were affiliated with two national programs designed to recruit teachers for high-need schools. The two programs were supported by the teacher preparation program at one university. One program was a traditional postbaccalaureate teacher preparation pathway in which prospective secondary school mathematics teachers received a U.S. government-funded scholarship through the Robert Noyce Teacher Scholarship Program. The Noyce scholarship program was designed to attract individuals with undergraduate mathematics, science, or engineering majors to serve as mathematics teachers in high-need1 schools. The other program was an alternative certification pathway administered by TFA in partnership with the university. TFA is focused on recruiting teachers to work in schools serving economically disadvantaged students, and the university provided academic-year coursework required by the state for the TFA teachers certification. In this article, we focus on the subset of TFA-supported teachers who were teaching secondary school mathematics.

We combined multiple data sources to create a database that included each teachers undergraduate major(s) and university, initial placement school and grade level, job title, length of teaching service (if she or he had left teaching), and responses to a survey with batteries of questions about each teachers decision to teach, perceived working conditions, and career satisfaction. We examined this database to first create a comparative profile of teachers from the two programs. We then used statistical analyses to address two research questions: (1) What are the initial school placements and career trajectories for Grades 612 mathematics teachers who start teaching in each pathway? In particular, do entrants continue teaching after completing their pathways service requirements? (2) How are the career trajectory profiles of each program related to the teachers impressions of school working conditions and administrative support?

In our analysis, we show that the financial incentive of the Noyce scholarship appeared to succeed in persuading individuals with undergraduate science, technology, engineering and mathematics (STEM) majors who were predisposed to teaching to attend the university program and start teaching in high-need schools. The alternative, TFA-supported pathway recruited a more diverse pool (in terms of a teachers undergraduate major, gender, and race/ethnicity) of secondary school mathematics teachers than the scholarship pathway. The alternative program placed these teachers in high-poverty schools that partnered with TFA. In comparing the career trajectories of the two groups of teachers, we found that the teachers who entered via the alternative pathway were more likely to leave teaching after completing the two-year service requirement than the teachers in the scholarship-supported group. Responses to our survey indicate that this difference in career trajectories may be explained in part by the TFA teachers negative impressions of working conditions in the schools where they were initially placed.

In the next section, we describe the conceptual framework of this study and summarize research on the labor market for secondary school mathematics teachers. We then describe the data sources for this study and our analytical technique. Next, we provide descriptive data on the two programs, followed by results of our analysis to answer the two research questions. Finally, we discuss the possible implications of this work for teacher educators, school administrators, and policy makers.



A pathway into teaching is a general term we use to describe a route that an individual may follow to become a teacher. A teacher preparation program is one specific route associated with an intuition or organization that is responsible for teacher preparation (National Research Council, 2010). Each pathway into teaching can be characterized by features such as recruitment methods, entry requirements, incentives for working in particular schools, job-placement assistance, service requirements, and ongoing mentoring or support provided during teachers early years of teaching. Policy makers often distinguish traditional university-based teacher preparation pathways from alternative teacher preparation pathways that are not necessarily housed in a college of education. The combination of a pathways features influences the pool of prospective teachers who enter the profession through that pathway.

On the one hand, some fully integrated, alternative teacher preparation programs such as district- or school-supported internship programs combine incentives, training, and placement as a package, and a prospective teacher cannot receive the training provided without committing to working in local schools (Papay et al., 2012). On the other hand, traditional university-based teacher education programs usually do not integrate training and job placement, and graduates of traditional programs are usually free to work in a wide number of schools or school districts after completing their training.

The confluence of two distinct pathways preparing secondary school mathematics teachers for service in high-need schools at one university2 created the opportunity for this longitudinal, comparative study. Although the data for this study were drawn from two preparation programs housed at a particular university, these programs can serve as a theoretically relevant contrastive case study (Patton, 2002) of different pathways into teaching secondary school mathematics. Patton (2002) described theoretical sampling as a more conceptually oriented version of criterion sampling in which the samples or cases are selected on the basis of their potential manifestation or representation of important theoretical constructs (p. 238). In the present study, the two contrasting programs illuminate how the design of programs to recruit secondary school mathematics teachers for work in high-need schools can shape the pool of entrants who enter a pathway and teachers early career trajectories. This contrastive case study may be informative for researchers, policy makers, school administrators, and teacher preparation program administrators.

Although in this study we used a comparative design, the results cannot be used to make causal claims about which pathway is more effective because each programs features and its competitive selection processes influenced the pool of prospective teachers who entered the profession through each pathway. For this reason, we provide a thorough descriptive account of each programs recruitment process, incentives, training, and placement process.


The challenge of finding qualified mathematics teachers to teach in hard-to-staff schools involves balancing the supply of new teachers and the labor markets demand (Ingersoll & Perda, 2010). The supply of teachers is influenced by factors such as working conditions (Johnson, Kraft, & Papay, 2012), salaries, and other incentives (Fulbeck & Richards, 2015), all of which can make the profession more or less attractive to potential entrants (Ingersoll, 2011). Although Ingersoll and Perda (2010) argued that the supply of mathematics teachers is sufficient, the number of new teachers finishing certification in university-based programs in the United States is decreasing: In the past 4 years alone, the number of new, university-prepared mathematics teachers has decreased by 32% (Darling-Hammond et al., 2016). However, the supply of teachers is also influenced by the number of teachers who return to teaching after an absence (reentry) or enter teaching through alternative pathwaysthat is, preparation programs other than university-based teacher education programs.

In areas with relatively stable student populations and teacherstudent ratios, the demand for new teachers is driven by teacher attrition (Ingersoll & Perda, 2010). After decades of national attrition rates of about 6% annually, U.S. teacher attrition rates have risen over the past decade to 7.7%8.4% annually (National Center for Education Statistics, 2016). Early-career teachers and those teachers nearing retirement are most likely to leave teaching (Hanushek, Kain, & Rivkin, 2004; Ingersoll, 2001; Kirby, Grissmer, & Hudson, 1991). However, retirement accounts for only one third of attrition, so two thirds of teachers who leave the profession are early- or mid-career teachers (Ingersoll & Merrill, 2012).

Although attrition is one contributor to the overall demand for teachers, another major driver of the local demand for teachers in particular schools, districts, or regions is high teacher turnover. Teacher turnover is defined as major changes in a teachers assignment from one school year (or within a year) to the next (Boe, Cook, & Sunderland, 2008, p. 8). Three examples of teacher turnover are teachers transferring to another school, moving from one classroom or subject area to another, or leaving the profession altogether (Borman & Dowling, 2008; Grissmer & Kirby, 1997; Ingersoll, 2011). In secondary school mathematics, turnover is especially acute in urban public schools with a high percentage of economically disadvantaged students (Ingersoll & Perda, 2010; Ronfeldt, 2012; Ronfeldt, Loeb, & Wyckoff, 2013). Ingersoll (2001) and Johnson et al. (2012) argued that such schools lose teachers not because of student-demographic characteristics per se, but mainly because of school working conditions. In particular, low-quality leadership, discipline challenges, lack of classroom resources, little faculty input into school decisions, and few opportunities for professional development all contribute to teacher turnover (Ingersoll, 2001). We note that teacher turnover is not a priori bad for students: If a low-performing teacher leaves and is replaced by a higher performing teacher, then turnover can be good for a school and for students (Hanushek & Rivkin, 2010).

Hanushek and Rivkins (2010) optimistic stance toward turnover is tempered by the reality that administrators in schools serving economically disadvantaged students are often forced to hire less than highly qualified teachers in the face of staffing difficulties. In an analysis of New York State data on schools and teacher characteristics, Lankford et al. (2002) found that low-performing, low-income, and non-White students had the least qualified teachers. In fact, approximately 50% of high school mathematics teachers in high-poverty and high-minority schools across the United States lack a college major or minor in mathematics or a mathematics-related field (Peske & Haycock, 2006). In high-poverty and high-minority middle schools, the situation is more acute: A total of 70% of mathematics classes are taught by teachers who have neither a major nor a minor in mathematics or a mathematics-related field (Peske & Haycock, 2006).

The association between teacher turnover and student academic achievement is negative (Guin, 2004; Ingersoll & Smith, 2003; Ronfeldt, Lankford, Loeb, & Wyckoff, 2011). This association may result from the loss of institutional knowledge when teachers leave a school. With teacher turnover, studentteacher relationships are affected, as is the remaining teachers institutional knowledge of students, and both appear to affect student achievement (Ronfeldt et al., 2011). Furthermore, teacher turnover has negative effects on the students of both the teachers who leave and the teachers who stay. Teachers who stay at schools with high turnover report feeling overburdened because they are continually searching for colleagues and continually mentoring new teachers (Carroll, Reichardt, & Guarino, 2000; Ronfeldt et al., 2013). Recruiting teachers to fill vacancies, regardless of the reason for the vacancy, is also challenging, costly, and time consuming for school site administrators.

Implicit in the literature on the teaching labor market and pathways into teaching is the issue of teacher effectiveness; a teachers effectiveness is typically defined in terms of improving students academic achievement. Researchers conducting several comparative studies have attempted to measure the effects of a teachers preparation pathway on her students achievement (Constantine et al., 2009; Kane et al., 2006; Shuls & Trivitt, 2015). Kane et al. (2006) found a slight (0.2 standard deviation) advantage in mean test scores for students of TFA corps members in New York City compared with students of teachers from traditional certification pathways. However, Kane et al. (2006) also found that the variation of student achievement for teachers prepared within each pathway was significantly larger than the variation between pathways, indicating that factors besides the teachers certification pathway influence the success of students in particular classrooms. Additionally, Kane and colleagues noted that students of experienced teachers had mean assessment outcomes 0.6 standard deviations higher than students of teachers in their first two years of service, indicating that experience may make up for initial differences by pathways after accounting for attrition. In a similar study, Boyd and colleagues (2012) examined the effectiveness of New York City teachers from traditional and alternative-certification pathways, using value-added modeling to show that students of TFA-prepared teachers made larger academic gains than students of new teachers who were prepared in traditional and district internship programs. However, the relatively high rate of turnover among TFA teachers offset this benefit because all teachers improved with experience, and non-TFA-prepared teachers stayed in the classroom longer.

Whereas Teach For America is a fairly well-known teacher preparation program in the United States, the Robert Noyce Teacher Scholarship Program is more of a sleeping giant in the field of STEM teacher preparation. Since 2002, the U.S. federal government, through the National Science Foundation (NSF), has provided grants to universities to provide scholarships to prospective and current mathematics and science teachers. In 2017, the NSFs Noyce program will distribute $58 million in grants to universities and nonprofit organizations to run Noyce-affiliated teacher preparation and teacher-support programs (National Science Foundation, 2017). Noyce-supported programs vary in focus. Some Noyce programs provide scholarships to undergraduate STEM majors who enter teaching (Worsham, Friedrichsen, Soucie, Barnett, & Akiba, 2014). Other Noyce programs support postbaccalaureate teacher preparation efforts, and still other Noyce programs provide training and stipends to develop teacher leaders in high-need schools. One uniting feature of all the grants in the NSFs Noyce portfolio is a focus on supporting programs that prepare and support highly qualified mathematics and science teachers to work in high-need schools (National Science Foundation, 2017).

Some research has been conducted on Noyce-supported programs, including local qualitative case studies (Ganchorre & Tomanek, 2012; Kelly, Gningue, & Qian, 2015; Kirchhoff & Lawrenz, 2011) and national quantitative evaluations (Liou, Desjardins, & Lawrenz, 2010; Liou, Kirchhoff, & Lawrenz, 2010). For example, Liou, Kirchhoff, and Lawrenz (2010) analyzed a survey completed by 555 Noyce scholarship recipients and showed that Noyce scholarships can be effective in convincing prospective teachers with undergraduate STEM degrees to begin teaching in high-need schools rather than in schools that are not high need. This study provides a focused comparison of Noyce-supported school mathematics teachers and non-Noyce-supported school mathematics teachers who were also specifically recruited to teach in high-need schools. We do not know of a comparable study that was focused on a single region to compare Noyce-supported secondary school mathematics teachers with comparable teachers from another pathway who serve in the same region.



For this study, we report data on 48 teachers who received scholarships from three Noyce-supported grants to Boston University3 (BU) focused on secondary school mathematics teacher preparation (two of the programs have been completed, and the third Noyce program was ongoing at the time of this study). Each of these programs provided prospective teachers with scholarships to cover the cost of a postbaccalaureate teacher certification and a masters degree in the art of teaching (MAT) program to qualified individuals. BUs 12-month program of study for prospective teachers included coursework and student-teaching experiences. Each Noyce scholarship was worth approximately $20,000, with in-kind support from BU covering the remaining tuition expenses. In return for this financial support, scholarship recipients committed to teaching in a high-need school for at least two years.4 Potential Noyce scholars were recruited through information sent to advisors at undergraduate STEM departments in universities across the country and via a special website created to advertise BUs Noyce program.

In line with the requirements of the national Noyce program, recipients of BUs Noyce scholarships are required to teach for at least two years in high-need schools within four or five years of completing their degrees. To enforce the teaching requirement, the Noyce scholarship functions as a no-interest loan that is forgiven when the service requirement is met. Recipients who do not complete the two-year service requirement within five years must repay the scholarship. Schools hire Noyce scholars through their districts usual hiring processes.

Although the three Noyce programs at BU varied in focus, the scholarship recipients of all three programs participated in the same coursework, student-teaching experiences, and professional development. Because of the substantial overlap among the groups of Noyce-supported teachers, we refer to all Noyce-supported teachers as Noyce scholars and distinguish among the BU Noyce programs only when discussing career trajectories and persistence.


Teach For America is a national nonprofit organization with the mission to enlist, develop, and mobilize as many as possible of our nations most promising future leaders to grow and strengthen the movement for educational equity and excellence (Teach For America, 2017). TFA is one of the largest single programs in the United States to recruit, train, and place teachers in all content areas, including Grades 612 mathematics. TFA teachers, known as corps members, are required to teach for two years. The national TFA organization is divided into regional offices, and TFA began placing teachers in the greater Boston area in 2009. Prospective TFA corps members apply to a region in TFA, and their applications are contingent on committing to teach for two years in a school that partners with that regional office of TFA. Corps members are assigned a teaching placement in a partnering school, and the schools that hire TFA teachers pay a fee to TFA for training and support costs. TFA has a network of recruiters at colleges and universities and also uses marketing materials, including a website and social media, to reach prospective teachers who may not be in college classrooms.

When TFA started operations in the Boston area, the organization entered into a partnership with the School of Education at BU. Through this partnership, BU agreed to provide 12 credits of coursework for TFA teachers during their first year of service. This coursework was focused on general methods of instruction, topics in special education, the use of computers in education, and one content-specific course connected to the content area in which the TFA corps member was teaching. For TFAs Grades 612 mathematics teachers, the content course was the same as one of the courses taken by the Noyce Scholars, but for logistical reasons, the two groups had separate sections. TFA teachers received their state teaching licenses directly from the state, and BU did not endorse TFA corps members for licensure. TFA teachers received a federal scholarship, called the AmeriCorps educational benefit, of $5,500 annually. This scholarship, in combination with financial aid from the university, offset the majority of the cost of TFA corps members BU courses.


The two programs that are contrasted in this study were colocated at one university, and some features of the programs overlapped, whereas other features were distinct. Table 1 summarizes the features of each program.

Table 1. Components of the Two Teacher Preparation Programs

Program component

Noyce program at BU

TFA Boston

(Grades 612 mathematics)

Recruitment and admission

Focused on individuals with undergraduate degree in mathematics or STEM with mathematics focus

Recruits high-achieving individuals; no undergraduate major requirement for Grades 612 mathematics teachers.

Preservice preparation

44-credit, 12-month MAT program including summer school, coursework, and supervised student teaching

5-week summer program designed by TFA program administrators

Required in-service coursework


12 credits of coursework at BU during first year of teaching

Additional in-service coursework


Option to enroll in BUs EdM program (24 additional credits)

Programmatic in-service support

Quarterly professional development meetings with all Noyce participants; opportunity to return to summer mathematics institute

Professional development meetings; supervision provided by TFA field office

Service requirement

2 years or 4 yearsa

2 years in placement school

Placement process

Traditional hiring process

TFA acts as recruiter/placement agent

aThe length of service required varied by Noyce program at BU. For this comparison, we have adjusted for this variation when discussing career trajectories.

The data for this study were gathered with support from a grant to study the effect of BUs Noyce programs in mathematics. Zahner was principal investigator of the grant. Zahner and Chapin were faculty members in the mathematics education program at Boston University, whereas Afonso was a graduate student at BU at the time of this study. Zahner was not involved in the design or administration of the original Noyce programs, but Chapin was lead PI on the original Noyce projects. Chapins involvement in this study provided access to existing evaluation data and helped in soliciting the participation of Noyce scholars who graduated before the start of this longitudinal study. As faculty members in the mathematics education department, Zahner and Chapin also worked with TFA administrators on programmatic administration and taught courses for TFA teachers. Levine and He were external to BUs programs and consulted on survey data and statistical analyses. We were able to collect data of sufficient quality and depth for this study because of the authors status as faculty members in the mathematics education program at BU. Simultaneously, external coauthors were included to provide outside perspectives on the analysis. In writing the proposal to study these programs, Zahner secured the agreement of the administrators of Teach for America of Greater Boston to contribute administrative data on TFA corps members and alumni to the study.


For this study, we relied on multiple data sources, including program records and a longitudinal survey. All data were collected and analyzed under the supervision of the institutional review board at Boston University.


Noyce program participants were identified using administrative records. A total of 48 Noyce scholars graduated from 2009 through 2014.5 Four additional potential Noyce scholars were offered the scholarship and enrolled in BUs MAT program, but these individuals either did not complete the course of studies or completed the program but decided to repay the scholarship rather than enter teaching.

The population of TFA teachers was identified using TFA of Greater Bostons administrative records. For the sake of making a meaningful comparison, we narrowed our focus to the TFA Greater Boston corps members whose primary teaching responsibility was Grades 612 mathematics. We identified 110 TFA corps members who were assigned to teach mathematics in Grades 612 between fall 2009 and fall 2014.

Drawing on administrative data from each program enabled us to analyze the teachers demographic data, their academic backgrounds, and their initial school placements. From the Noyce program administrators, we also had access to data on whether Noyce-supported teachers had completed their service requirements. From the TFA alumni survey, we learned whether TFA teachers had completed their service requirements and their current job titles and employers. In Table 2, we summarize basic data about the number of each programs participants who started teaching from 2009 through 2014.

Table 2. Number of Grades 612 Mathematics Teachers by Year and Program









BU Noyce Scholars








TFA Boston Grades
Grades 612 Math Teachers
















a Cohorts are equated by the year that members of each cohort started teaching (or were eligible to start teaching). For example, in fall 2009, seven Noyce scholars entered the field and started teaching mathematics, and 11 TFA Boston corps members started teaching mathematics.

The cohorts in Table 2 are matched by the year the teachers in each pathway entered the classroom. The Noyce teachers who started teaching in fall 2009 entered BUs MAT program in fall 2008. The TFA teachers in the 2009 cohort joined TFA in summer 2009 and started teaching in fall 2009. This choice to match cohorts by years in the classroom follows the conventions used in other studies (Boyd et al., 2012; Papay et al., 2012) and also aligns with our interest in tracking the career trajectories of these mathematics teachers in and through their service requirements. Between 2009 and 2014, TFA placed more than twice as many secondary school mathematics teachers as BUs Noyce program placed. However, note that several other Noyce-supported programs at other universities in the region overlapped with TFA of Greater Bostons service area between 2009 and 2014, and Table 2 does not account for teachers affiliated with those Noyce programs.


We invited all the individuals enumerated in Table 2 to complete an online survey in spring 2014 and again in spring 2015. Following protocols for research with human subjects, we asked the survey takers to consent to the study, and their answers are reported in aggregate to protect the identities of respondents. Each survey taker received a $25 gift card as an incentive to respond, and most respondents completed the survey in 3045 minutes. In the following section, we briefly describe the parts of the survey and provide reliability statistics for the two scales reported in the findings section.

Teaching Situation

In this sequence of 10 items, respondents were asked to summarize their career history, identify whether they had completed their pathways teaching-service requirement, name their current employer and job title, describe their current course load and student populations if they were classroom teachers, and describe future plans if they indicated intention to leave teaching within one year.

Pathways Considered and Self-Evaluation of Preparation to Teach

In this block of nine items, respondents were asked to name the various pathways they considered as routes into teaching and to offer evaluation of their relative preparation. For example, one question on this scale was, When you talk to your colleagues from other teacher preparation programs, do you feel as though your preparation for teaching was generally: (a) better than, (b) about the same as, or (c) not as good as the preparation your colleagues received in their teacher education programs?

Professional Roles and Collaboration

This researcher-created block of six questions related to current teachers leadership roles at their schools (e.g., department chair or club advisory roles) and opportunities for professional collaboration. The questions were based on items from a survey developed at the University of Minnesota for a national Noyce evaluation (NSF Grant No. 0514884, Frances Lawrenz, PI).

School Context and Administrative Support

Responses to the 12 items in this section about administrative conditions and support at the school where participants last taught were on a 5-point Likert scale. One example was, I feel that the school administrators provide support by spending money on adequate classroom supplies. These items were adapted from Boyd et al. (2012). We found that the items had strong internal reliability, Cronbachs [39_22554.htm_g/00002.jpg] = 0.89.

Access to Resources

This researcher-created block of 15 questions was designed to follow up on the school-context and administrative-support items by identifying whether participants had access to particular mathematics teaching tools such as calculators or mathematics manipulatives.

Satisfaction With Career Pathway

Participants responded to this block of nine items about their satisfaction with their career pathways using a 5-point Likert scale. Four of the questions on the scale were from Boyd and colleagues (2012) survey. We found high internal reliability with this scale, Cronbachs [39_22554.htm_g/00004.jpg] = 0.87.

Orientation to Mathematics

Responding to this block of 21 items using a 5-point Likert scale, participants were asked to agree or disagree with statements such as, Mathematics is a collection of rules and procedures that prescribe how to solve a task. These items were adapted from Kim and McCrorys (2007) research on mathematics teachers orientations to the subject area. Factor analysis of this scale indicated that at least two distinct factors were measured in this scale. To narrow the scope of this article, we do not discuss this block of items further.

Effective Mathematics Teaching Practices

For this block of 15 items, participants were asked to agree or disagree, using a 5-point Likert scale, with statements such as, To be an effective mathematics teacher, it is most important to pose open-ended problems that have multiple solution methods. These items were adapted from Boyd and colleagues (2012) survey, and we modified them to align with the block of items on the orientations-to-mathematics scale. To narrow the scope of this article, we do not discuss this block of items further.

Orientation to Mathematics (Qualitative)

Participants were asked to respond to a series of three comic-style vignettes showing mathematics teaching in a classroom situation. In these open-ended items, teachers were asked to identify the mathematics in the situation, discuss options a teacher might pursue in response to the situation, and then describe what they would do as teachers in response to the situation. Again, to maintain the focus of this article, we do not discuss the results from the qualitative analysis here.

Opportunities to Learn in Preparation Pathways

In responding to this block of eight items, adapted from the survey used in Boyd et al. (2012), participants rated the extent to which their teacher preparation pathways included opportunities to learn particular mathematical content, foundational topics in education such as the relationship between social justice and education, and to develop pedagogical strategies for particular student populations. The TFA teachers responded to this block twice, once concerning their preparation with TFA, and a second time about their coursework at BU. Reliability was not calculated on this multidimensional scale. Because these questions were program specific, we do not discuss this block of items further in this article.

Finally, we note that we did not collect or analyze student-level academic achievement data (as was done by Boyd et al., 2012, and Papay et al., 2012). One logistical hurtle that prevented such an analysis was accessing and reconciling student achievement data from multiple districts and, in some cases, from different states. Additionally, at the time of this study, Massachusetts had statewide mathematics assessments for Grades 38 and 10. Because the 10th-grade mathematics exam included content from middle school as well as 9th- and 10th-grade mathematics, we could not meaningfully interpret students 10th-grade assessment scores.


Statistical analyses were performed in the R statistical software environment (R Core Team, 2016). For analysis of contingency table data, we used Fishers exact text, a type of chi-square test procedure for contingency tables with small counts. For the analysis of career trajectories, we also used logistic regression and survival analysis. These analyses are described in more detail in the findings.

The administrative data were our primary resource for answering Research Question 1, and we had access to some form of administrative data for all participants. Some records had missing data, which we account for in our analyses. For our analysis for Research Question 2, we relied on our surveys. Counting all the participants who responded to at least one of the longitudinal surveys, the overall survey response rate was 64%. The response rate among Noyce teachers was 100%, and 49% of the TFA group responded. The disparity in response rates reflects the fact that Noyce scholars agreed to participate in follow-up studies as a condition of their scholarships. We were unable to solicit responses from many of the TFA corps members from the 2009, 2010, and 2011 cohorts who were beyond the window of their two-year service requirements before the study started. For these individuals, we did have TFAs records of where they were initially placed, and data from TFAs alumni survey, which included the alumnis current employers and job titles in 20142015 academic year. To identify whether nonresponses from the TFA group were likely to affect our results, we ran several preliminary analyses comparing data known about all TFA participants (e.g., data provided by the TFA program administrators) and whether an individual responded to our surveys. We found nonsignificant differences on cross-tabulations of teaching status by survey response (p = 0.909 Fishers exact test), and undergraduate STEM major by survey response (p = 0.573, Fishers exact test). We interpret these results to indicate that the group of TFA school mathematics teachers who responded to our survey is likely sufficient for making the inferences we draw in this study. However, to remind readers that the survey responses represent a subset of the TFA group, we use the term TFA respondents in the discussion of survey responses.



According to each programs demographic data, the Noyce scholars were more likely (p < 0.0001 using Fishers exact test) than TFA corps members to be female (70% vs. 60%). TFA corps members were also more racially diverse than their Noyce counterparts (p < 0.0001 using Fishers exact test): 56% of TFA teachers identified as White, 11% identified as African American, 11% as Hispanic, 12% as multiracial, 4% as Asian Pacific Islander, and 6% as other. Noyce scholars identified as 90% White, 2% African American, 4% Hispanic, and 4% Asian Pacific Islander. Both the Noyce pathway teachers and TFA teachers were placed in schools in which more than 50% of the students were identified as African American and Latino or Latina. Thus, although neither Noyce nor TFA recruited a teacher population that was reflective of the student demographics at target schools, TFA did recruit a more diverse pool of secondary school mathematics teachers than the scholarship-supported program did.

Both the Noyce program and TFA had competitive selection processes, but their differing recruitment criteria are clear when one examines the teachers undergraduate majors. The Noyce scholarship recipients were significantly more likely than their TFA counterparts to have undergraduate degrees in mathematics or another STEM6 field. Forty-six of the 48 Noyce mathematics teachers completed a STEM major. The remaining two Noyce teachers completed undergraduate studies in technical fields (actuarial sciences and music technology) and completed the postbaccalaureate equivalent of a STEM major. TFA did not have a corresponding undergraduate major requirement for secondary school mathematics teachers. (TFA mathematics teachers without STEM majors fulfilled the state-mandated content-preparation requirements by passing a state mathematics assessment for prospective teachers.) Among the 110 TFA secondary school mathematics teachers in this study, 33 completed a STEM major, 52 majored in a non-STEM field, and 25 were missing undergraduate-major data. This difference in the distribution of STEM majors was significant (p < 0.0001, using Fishers exact test), and we note that the missing data would not affect this result: The difference in the distribution of STEM and non-STEM majors would be significant even if all the TFA corps members with missing data were STEM majors. Additionally, if we focus on undergraduate mathematics majors, the trend is even more pronounced: A total of 4 of the 84 TFA corps members for whom we have undergraduate major information studied mathematics, contrasted with 36 of the 48 Noyce teachers.


Research Question 1 is, What are the initial school placements and career trajectories for grades 612 mathematics teachers who start teaching in each pathway? In particular, do entrants continue teaching after completing their pathways service requirements? Retention is a major driver of the high demand for secondary school mathematics teachers in hard-to-staff schools (Ingersoll & Perda, 2010). Therefore, an analysis of retention is critical for understanding the dynamics of supplying qualified teachers in hard-to-staff schools.

Placements and Placement Process

Using the program administrative data for all 158 teachers, we observed several major differences in the types of schools in which Noyce and TFA teachers initially worked. Secondary school mathematics teachers span Grades 612. First, considering placement in a middle school (Grades 68) or high school (Grades 912), the TFA pathway mathematics teachers were more likely to be placed in a middle school than their Noyce pathway counterparts (p < 0.0001, Fishers exact test). Among the Noyce group, 38 began teaching in high schools, and 10 began teaching in middle schools. Among the mathematics teachers from the TFA pathway, 67 were placed in middle schools, 42 were placed in high schools, and one was placed in a combined middle and high school.

Second, Noyce and TFA pathway teachers started teaching in different types of public schools. Charter schools are publicly funded schools that operate outside of the control of the local school district. Thirty-one percent of the Grades 612 mathematics teachers in TFA were initially placed in charter schools, compared with 12.5% of their Noyce counterparts, a significant difference in distribution by school type (p = 0.025, Fishers exact test). This sorting pattern reflects the TFA teachers placement process, in which schools contract with TFA to supply teachers rather than hiring teachers on the open market.

The percentage of students in a particular school who receive a free or reduced-price lunch was used by the state of Massachusetts until 20137 as a metric for measuring the SES of a school population. In a third contrast of initial teaching placements, we found that TFA teachers were initially placed in schools with a higher proportion of students who qualified for free or reduced-price lunch (M = 79.16, SD = 13.72) than the schools in which Noyce teachers began teaching (M = 63.21, SD = 23.8). This difference in means is significant at the 0.05 level, t(48.4) = 3.841, p < 0.001. Partially explaining this difference is the definition of a high-need school: Student poverty is only one aspect of the U.S. federal definition. High-need schools can also be characterized by high teacher turnover or a relatively high proportion of out-of-field teachers. Thus, although both Noyce scholars and TFA teachers were required to work in high-need schools, the programs criteria for placement differed. Some Noyce scholars elected to complete their service requirements at schools with a shortage of qualified mathematics teachers but a relatively low percentage of students who qualified for free or reduced-price lunch.

Career Trajectories

TFA required two years of service from its teachers, and the Noyce programs at BU required at least two years of service (see Note 4). Given the literature on the importance of mathematics teacher retention (Ingersoll & Perda, 2010) and the overall harmful effects of teacher turnover (Guin, 2004; Ingersoll, & Smith, 2003; Ronfeldt et al., 2011), one key question with policy implications is, Do teachers who are recruited to work in high-need schools continue teaching after completing their service requirements? Table 3 shows the academic year 20142015 job placement of TFA and Noyce teachers who were beyond their programs service requirements. In the 20142015 school year, the teachers in the 2009 cohort were (or would be) in their sixth year of teaching. In general, a higher percentage of Noyce-prepared teachers than TFA teachers remained teaching beyond their required service (p = 0.0002, Fishers exact test), with the caveat that the number of individuals in the early cohorts was small.

Table 3. Current Position in 20142015 for Teachers Beyond Service Requirement


BU Noyce

TFA Boston


Classroom teacher






Other educationa






Not in educationb










a Teachers who had left the classroom for administrative/counseling/coaching roles, and some teachers who had moved to work for nonprofit organizations that provide support to schools.

b Former teachers who had moved into jobs in the business sector, some graduate students, and some parents who were on leave for the birth of a child at the time our of survey.

Noyce and TFA program participants listed as classroom teachers in Table 3 were in their third through sixth years of teaching. In line with other evaluations of teacher preparation pathways and career trajectories (e.g., Boyd et al., 2012), we found that many TFA alumni who stayed in education moved into administrative roles. For example, one former TFA mathematics teacher became the director of operations at a charter school by the end of his third year in education. The Noyce-prepared teachers who moved into other education roles include an administrator, two community college instructors, and a counselor. For TFA Boston-affiliated secondary school mathematics teachers, the results presented in this section align with TFAs national statistics (Boyd et al., 2012).

The data in Table 3 provide a snapshot of the secondary school mathematics teachers career placements at one point in time. However, this cross-sectional view is not a longitudinal track of each cohort. Table 4 shows the full job-placement data for 153 of the teachers in our full sample, including members of cohorts that were in the midst of their service requirements.

Table 4. Job Title for BUs Noyce Program and TFA Boston Grades 612 Mathematics Teachers in 20142015 School Year by Cohort, With Column Percentages

Position in 201415 Academic Year

Cohort year

(Started teaching in fall)





























































Specialist, administrator, or other nonteaching education role






























Not currently in education
















































One potential weakness of the data in Tables 3 and 4 is that the sample size of teachers in the early cohorts is small. To account for the sample-size differences and to study the significance of the relationship between programs and career longevity, we created a variable to represent teaching-discontinuation status throughout the study period. The variable contained four levels: (a) continuously teaching (62.20% of participants), (b) teaching discontinued (28.05% of participants), (c) never started teaching (6.71% of participants), and (d) delay, then continuously teaching (3.05% of participants). We found a significant relationship between the teaching-discontinuation variable and the program category (Noyce or TFA; p = 0.0070; Fishers exact test). A logistic regression model accounting for the initial school percentage of students classified as low SES and an individual teachers STEM major identifier shows that being in the TFA group increased the probability of teaching discontinuation by 45.4% (p = 0.003).

To shed additional light on length of service, we used survival analysis (Klein & Moeschberger, 2005), a method that provides more information than creating and analyzing a binary-outcome variable to study time to teaching discontinuation. Survival analysis is frequently used in medical studies on time-to-event (e.g., disease outcome) data. The advantage of this modeling approach is that we may include both those participants who have left teaching (and whose outcome is known) and those who have persisted in teaching to date and whose length of service is unknown (censored). Cox proportional-hazards regression models were applied to predict the relative risk in teaching discontinuation. The predictors in this model were the program category (Noyce or TFA), school low-SES statistics, and STEM-major identifier. The Noyce and TFA groups present discontinuation rates of 8/47 (17%) and 43/106 (41%), respectively; see Tables 3 and 4. Regression diagnostics indicated no violations of the Cox model assumptions. The analysis results indicated that after controlling for school-level data about student SES and a teachers STEM-major identifier, the risk of discontinuing in teaching for the TFA group is 5.1 times as large (95% CI [1.5, 17.1]) as for the Noyce group (p = 0.009; Cox regression model inference).

Finally, to investigate how retention might change longitudinally in the future, we also examined the job-placement data by cohort and program as well as survey responses given by 60 classroom teachers in spring 2015 (among 80 survey responses we received in spring 2015, 60 were from current classroom teachers). In that survey, we asked current teachers how long they planned to stay at their current schools. A total of 22 of the 25 (88%) TFA corps members and TFA alumni who answered this question indicated that they planned to leave their current schools within two years. In contrast, 11 of the 35 (31%) Noyce pathway respondents indicated that they planned to leave their current schools within two years. These survey responses provide additional data to indicate how the attrition data in Tables 3 and 4 may develop over time.


In the final facet of this analysis, to answer Research Question 2, we examine how individual- and school-level factors differed across the groups of mathematics teachers prepared in each pathway. To investigate these questions, we examined teachers responses to the survey items about their decisions to enter teaching, about school-level administrative conditions, and about their career satisfaction. Recall that each of these batteries of items was adapted from prior studies of teacher preparation pathways (Boyd et al., 2012) and earlier studies of the national Noyce-related programs, and these scales had high-reliability statistics. The items on administrative conditions and career satisfaction appeared on both the 2014 and 2015 surveys. If an individual completed both surveys, we used her or his more recent responses for this analysis.

Noyce-pathway teachers who responded to our survey were more likely (p = 0.025; Fishers exact test) than the TFA-pathway respondents to choose teaching as a career before seeking a preparation program. Seventy-five percent of Noyce teachers who responded to our survey reported that they probably or definitely would have entered teaching without the Noyce scholarship, and 15% were unsure. Conversely, 50% of the TFA respondents to the survey indicated they would definitely not or probably not have become teachers without joining TFA, and 34% were unsure. Although the majority of Noyce-supported teachers reported that they would have become teachers without the Noyce scholarship, 23% reported they would have definitely not or probably not worked in a high-need school without the scholarship, and an additional 30% were unsure. On the bases of the responses to these questions together, the TFA program appears to have induced people who were unsure about becoming teachers to teach, whereas the Noyce scholarship was effective at directing prospective mathematics teachers with strong disciplinary preparation into teaching in high-need schools (aligning with the results reported by Liou, Kirchhoff, & Lawrenz, 2010).

Individuals with STEM majors who want to pursue a career in teaching are relatively rare. The Noyce scholarship was successful in enticing potential teachers with STEM majors to attend BUs MAT program; 75% percent of the Noyce-pathway respondents indicated that they probably or definitely would not have attended BU without the Noyce scholarship. Most of these individuals stated that they considered attending multiple universities or even entering teaching via alternative pathways such as TFA. These responses indicate that the Noyce scholarship, in combination with financial aid from the university, was an important tool to recruit highly sought STEM majors to attend BUs MAT program. The Noyce-pathway teachers also reported a high level of satisfaction with BUs MAT program. More than 75% of Noyce teachers who completed the 2015 survey said that they would recommend their teacher preparation program to a friend, and 50% of Noyce teachers reported feeling that they were more prepared to teach than were their colleagues who entered via other pathways.

Those studying teacher retention highlight the critical influence of school working conditions on teachers job satisfaction (Ingersoll, 2001; Johnson et al., 2012). As noted in its description, the survey included a battery of questions on teachers impressions of school conditions and administrationteacher relations. As a whole, the scale showed strong internal consistency (Cronbachs [39_22554.htm_g/00006.jpg] = 0.891). The Noyce-pathway teachers who responded to our survey consistently reported more positive evaluations of their school administrations than did their TFA-pathway counterparts who responded to our survey. An independent-samples t test showed that the difference in means was significant at the 0.05 level for five of the 12 items on the scale (those identified by asterisks in Table 5). The p values for the set of 12 tests have been adjusted using the false-discovery-rate procedure of Benjamini and Hochberg (1995).

Table 5. Means and Standard Deviations on School-Administrative-Conditions Scale


Rate the following on a 5-point scale from strongly disagree (1) to strongly agree (5)

BU Noyce

Mean (SD)

(n = 47)

TFA Boston

Mean (SD)

(n = 47)

I feel that teachers work with the administrators to make school-related decisions.**

3.19 (1.10)

2.66 (1.24)

I feel that administrators appreciate teacher input when making school-related decisions.*

3.38 (1.11)

2.92 (1.23)

I feel that administrators seek out information from teachers when making school-related decisions.**

3.15 (1.08)

2.62 (1.31)

I feel that administrators tend to make school-related decisions without consulting the teachers.a

3.06 (1.15)

3.45 (1.19)

I feel that the school administrators provide support by spending money on adequate classroom supplies.

3.383 (1.05)

3.404 (1.06)

I feel that the school administrators provide support by providing discipline to students beyond the classroom.*

3.53 (1.14)

3.00 (1.29)

I feel that the school administrators provide support by providing constructive feedback on how I can improve as a teacher.

3.30 (1.20)

3.15 (1.23)

I feel that the school administrators provide support by providing professional development opportunities.

3.51 (0.95)

3.23 (1.05)

I feel that the school administrators provide support by ensuring the maintenance of a school with updated technology.


3.55 (1.08)

3.25 (1.19)

I feel that the school administrators provide support by maintaining the physical plant of the school (e.g., the building, classrooms, other spaces inside the school, and the school grounds).

3.55 (1.04)

3.32 (1.04)

I feel that the school administrators provide support by providing a welcoming social environment for teachers.**

3.62 (1.07)

3.00 (1.10)

I feel that the school administrators provide support by providing an environment where I feel like a professional.

3.59 (1.06)

3.39 (0.99)

a Item reverse coded.

*p < 0.10. **p < 0.05.

One possible explanation for the difference in the teachers reported satisfaction with school administrative conditions may be that Noyce-affiliated teachers went through a traditional hiring process, whereas TFA teachers were placed in their schools by the TFA program administrators. Thus, Noyce scholars had opportunities to assess the administration and working conditions of the schools to which they applied to teach before accepting a position. In contrast, TFA corps members were not given a choice in selecting the schools where they would work.

Another block of items on our surveys focused on the teachers career satisfaction. This scale also had high internal reliability (Cronbachs [39_22554.htm_g/00008.jpg] = 0.871). Similar to the scale on school conditions, in general, the responding Noyce teachers reported higher levels of career satisfaction than their TFA counterparts who responded to our survey. The means for Noyce teachers were higher on each item on the scale, and the difference in means was significant at the p = 0.05 level on three items (t test adjusting for multiple comparisons using false discovery rate; identified by asterisks in Table 6).

Table 6. Means and Standard Deviations on Career-Satisfaction Scale


Rate the following on a 5-point scale from strongly disagree (1) to strongly agree (5)

BU Noyce

M (SD)

(n = 47)

TFA Boston

M (SD)

(n = 47)

If I had to do it all over again, in view of my present knowledge, I would become a teacher.*

4.19 (0.85)

3.81 (1.12)

If I had to do it all over again, I would choose the same teacher preparation program and/or route into teaching.**

4.13 (1.06)

3.60 (1.29)

If I had to do it all over again, I would rather teach in a different school.a

2.64 (1.15)

2.72 (1.33)

I am as happy about teaching as I thought I would be.

3.51 (1.04)

3.33 (1.07)

In most ways, being a teacher is close to my ideal.


3.477 (1.02)

3.09 (1.06)

My working conditions as a teacher are excellent.**

2.96 (1.12)

2.49 (0.95)

I am satisfied with being a teacher.**

3.70 (0.95)

3.15 (1.12)

So far I have gotten the important things I wanted out of being a teacher.

3.64 (1.03)

3.68 (0.91)

If I could choose my career over, I would change almost nothing.**

3.46 (1.18)

3.06 (1.07)

a Item reverse coded.

*p < 0.10. **p < 0.05.


In this article, we have described outcomes from two pathways into secondary school mathematics teaching. Given the enduring and widespread challenge of recruiting qualified school mathematics and science teachers (National Audit Office, 2016; National Research Council, 2010; OECD, 2005), the results of this study, in combination with prior research, can inform the design of incentive programs and teacher preparation pathways to fill shortages of STEM teachers. We compared early career-trajectory outcomes of teachers from two pathways at one university to study how possible design features of teacher-recruitment pathways might lead to differing recruiting outcomes and career trajectories of entrants to the field.

Our data illustrate that the Noyce scholarship-supported pathway was generally successful in recruiting individuals with STEM majors, training them to be mathematics teachers, and placing those individuals as secondary school mathematics teachers in high-need schools. The comparison of the scholarship-pathway teachers with the secondary school mathematics teachers in the alternative-certification pathway provides a useful contrast. On the one hand, the alternatively certified secondary school mathematics teachers were less likely than the scholarship-pathway teachers to have STEM majors, and the attrition rate for the alternatively prepared teachers was higher than the attrition rate for the scholarship-supported teachers, particularly after they had completed the two-year service requirement. On the other hand, the alternative-certification program recruited a more diverse pool of potential teachers and placed these teachers in schools serving a higher proportion of low-SES students. On our longitudinal survey, the responding traditional-pathway teachers indicated the intention to remain in teaching longer than their alternatively prepared counterparts who responded to our survey. The career-trajectory results may be explained in part by differences in the perceived levels of administrative support at the schools in which teachers from both pathways started teaching (Johnson et al., 2012). Our results parallel the results of prior studies of pathways into teaching and incentives (Boyd et al., 2012; Papay et al., 2012; Ronfeldt et al., 2013; Steele et al., 2010).

This comparison highlights how the design features (e.g., recruitment methods, incentives, and job-placement processes) for each program influenced the observed outcomes for each program. For example, one strength of the scholarship-supported pathway is that the scholarship successfully enticed individuals with STEM majors to attend the universitys one-year teacher preparation program. For these potential teachers, the scholarship lowered the cost of attending the university-based teacher preparation program (though we note that they faced the opportunity cost of attending school for a full-year, postbaccalaureate degree program). The scholarships model as a no-interest loan then provided those individuals with strong incentives to teach mathematics in a high-need school for at least two years. This longitudinal study shows that the majority of the scholarship recipients subsequently remained in teaching beyond their service requirements. In light of the research showing that low-income students and students from groups underrepresented in STEM careers often have mathematics teachers with little content background (Lankford et al., 2002; Peske & Haycock, 2006), this result appears to be a major benefit of the Noyce program. This finding is tempered, however, by the fact that Noyce-scholarship-supported teachers also tended to work in schools that met the federal definition of high-need but not necessarily in schools serving the highest percentages of low-income students.

In contrast, the local TFA program recruited a more diverse group of individuals through a highly selective admissions process and placed these individuals as teachers in high-poverty schools. This recruitment and placement process filled an immediate need in many schools that struggled to recruit qualified teachers. Additionally, compared with the Noyce-supported program, the alternative TFA-supported program recruited a more diverse pool of teachers in terms of race/ethnicity and gender, which likely helped bring more balance between student demographics and teacher demographics at the schools in which they were placed (Cherng & Halpin, 2016). In essence, this study illustrates that both programs largely achieved their goals, but the goals of the two programs differed slightly, resulting in different long-term outcomes. The difference in the programs recruiting strategies also highlights the potential tension in balancing teacher demographics with their content-area preparation and is reminiscent of Martins (2007) reflections on what factors make teachers qualified to work in urban schools serving students of color.

The survival analysis based on a full sample and our longitudinal-survey responses showed that mathematics teachers recruited and placed through the alternative pathway were more likely to leave teaching than their scholarship-supported counterparts. The relatively short teaching careers of the TFA teachers in this study is not necessarily surprising in light of prior research that has examined the career pathways of TFA teachers in other cities (Boyd et al., 2012; Kane et al., 2006). On the one hand, in findings consistent with prior analyses of the career pathways of TFA teachers, we found that many of the TFA corps members in our sample who left classroom teaching remained involved with education, taking on roles in administration or working for nonprofit organizations with educational objectives. On the other hand, the general trend of persistence of the scholarship-supported teachers as classroom teachers in high-need schools beyond their service requirements was relatively surprising in light of prior research that indicated that mathematics teachers with STEM majors have high rates of attrition (Hanushek & Rivkin, 2010). From a purely economic standpoint, the persistence of these teachers was especially striking because most of these individuals were qualified for higher paying career opportunities in STEM-related careers outside of teaching. For example, one potential Noyce scholar who decided to repay the scholarship pursued a highly paid career in finance.

From a systematic perspective, we also note that this evaluation has underestimated the overall effect of the government-supported Noyce scholarship program in the Boston area. At the same time that BU supported a Noyce scholarship program, several other neighboring universities (including the University of Massachusetts Boston, Tufts University, and Boston College in the immediate metro area) also had grants from the Robert Noyce Teacher Scholarship Program. However, as noted in the description of the national scholarship program, the specific foci of the Noyce programs differed slightly across institutions. The results we have presented here provide some evidence for the efficacy of one government-supported grant program in Boston and indicate that more research is warranted on the regional effect of such government-supported scholarship and grant programs.

We note that the differences across the pathways observed in this study cannot be interpreted causally. Rather, we conjecture that each pathway attracted different types of prospective teachers who had different long-term career goals before they started teaching. This distinction was most evident in responses to the survey item that asked whether entrants would have become teachers even if they had not entered their teacher preparation programs. Seventy-five percent of Noyce teachers probably or definitely would have still entered teaching without the Noyce scholarship, compared with 50% of the TFA respondents who said that they would definitely not or probably not have become teachers without joining TFA. This difference highlights the point that, on the one hand, the Noyce program succeeded in drawing individuals with STEM majors and an interest in teaching to teach in high-need schools. TFA, on the other hand, appeared to draw many individuals who would not otherwise teach into teaching careers; though regarding attrition, ascertaining whether the teachers who remained in the classroom beyond their service requirement were also the ones who had already committed to teaching is difficult.

In addition to attracting different pools of potential teachers, the two programs had initial job-placement processes that differed significantly, which may help account for the observed differences in career trajectories. The Noyce-supported teachers had a full year between their decisions to teach and their first jobs. During this time, the Noyce teachers had opportunities to learn which schools or districts that met the program requirements were also good places to work. They could then focus their job searches in those schools and districts. Additionally, Noyce program alumni who were working in local high-need schools would contact BU about openings at their schools to recruit other Noyce scholars. This pattern of recruitment created clusters of Noyce-supported teachers in particular schools. In contrast, TFA teachers placements were determined by program administrators. That the Noyce and TFA teachers impressions of administrative support differed was clear in their survey responses. Given the importance of school working conditions on teacher retention, this study, in line with Johnson et al. (2012), shows that creating a positive working environment may be a critical step to address teacher shortages in particular schools.

Finally, we end with the caution that the results presented here should not be interpreted in terms of which of the two pathways is better. Rather, we hope that this study can inform administrators, teacher educators, politicians, and policy makers about how the design of teacher-recruitment pathways can lead to differing program effects. Although significant research has been conducted to compare outcomes from traditional and alternative pathways into teaching, less research is found on the programmatic features of these pathways that might lead to different pools of potential teachers and different long-term career outcomes. The findings we present here show that further research on government-grant and scholarship-supported pathways such as the Noyce Teacher Scholarship Program is warranted to explicate the regional and national value of scholarship-supported programs designed to recruit, train, and place STEM majors as teachers in high-need schools.


1. The definition of high-need local educational agency is used from Title I of the Elementary and Secondary Education Act. The criteria for a high-need school or educational agency include more than 50% low-SES students, a high rate of teacher turnover, or a high percentage of out-of-field teachers.

2. The university also has an ongoing undergraduate program that prepares and certifies secondary mathematics teachers. That program is not included in this evaluation because graduates of the program were neither required, nor given incentives, to teach in high-need schools.

3. We decided to name the university in this study because the name of the grant supporting this research reveals the name of the university. To protect the privacy of the teachers who participated in this study, we do not provide specific teacher-level data, such as the names of the schools and districts where they worked.

4. Scholarship recipients from one of the BU Noyce Programs committed to teaching in a high-need school for four years, and they received an additional living stipend ($25,000 during the year of studies) and salary supplements during their term of service ($12,500 per year for each year of service). However, the teachers in these programs participated in the same coursework and student-teaching experiences as those who committed to teaching for two years. Thus, we treat the Noyce-supported teachers as a group in most of this study and distinguish within the Noyce group only when discussing persistence.

5. One member of the 2009 Noyce scholar cohort is deceased. In our reports of career trajectories, the percentages reported for Noyce teachers exclude this individual and use n = 47.

6. Following the NSFs guidelines, we defined STEM majors to include mathematics, computer science, engineering, and the natural sciences, as well as psychology, economics, and political science.

7. Massachusetts has since stopped using this statistic to measure SES.


This article is based on work supported by the National Science Foundation under Grants No. 1240057 and 1459792. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. We are grateful to the administration at Boston University School of Education and to the directors of Teach for America Greater Boston for collaborating with this study. We are also grateful to Tim Heeren for assistance with survey design and to Bonnie Schappelle for editorial assistance.


Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B 57, 289300.

Boe, E. E., Cook, L. H., & Sunderland, R. J. (2008). Teacher turnover: Examining exit attrition, teaching area transfer, and school migration. Exceptional Children, 75(1), 731.

Borman, G. D., & Dowling, N. M. (2008). Teacher attrition and retention: A meta-analytic and narrative review of the research. Review of Educational Research78(3), 367409.

Boyd, D., Grossman, P., Hammerness, K., Lankford, H., Loeb, S., Ronfeldt, M., & Wyckoff, J. (2012). Recruiting effective math teachers: Evidence from New York City. American Educational Research Journal, 49, 10081047.

Carroll, S., Reichardt, R., & Guarino, C. (2000). The distribution of teachers among Californias school districts and schools. Santa Monica, CA: RAND.

Cherng, H.-Y. S., & Halpin, P. F. (2016). The importance of minority teachers: Student perceptions of minority versus White teachers. Educational Researcher, 45(7), 407420.

Constantine, J., Player, D., Silva, T., Hallgren, K., Grider, M., & Deke, J. (2009). An evaluation of teachers trained through different routes to certification (No. NCEE 2009-4043). Washington, DC: National Center for Education Evaluation and Regional Assistance.

Darling-Hammond, L. (2000). Teacher quality and student achievement: A review of state policy evidence. Education Policy Analysis Archives, 8(1), 144.

Darling-Hammond, L., Furger, R., Shields, P. M., & Sutcher, L. (2016). Addressing Californias emerging teacher shortage: An analysis of sources and solutions. Palo Alto, CA: Learning Policy Institute.

Fulbeck, E. S., & Richards, M. P. (2015). The impact of school-based financial incentives on teachers strategic moves: A descriptive analysis. Teachers College Record, 117(9), 140.

Ganchorre, A. R., & Tomanek, D. (2012). Commitment to teach in under-resourced schools: Prospective science and mathematics teachers dispositions. Journal of Science Teacher Education, 23(1), 87110.

Grissmer, D., & Kirby, S. (1997). Teacher turnover and teacher quality. Teachers College Record99(1), 4556.

Guarino, C. M., Santibanez, L., & Daley, G. A. (2006). Teacher recruitment and retention: A review of the recent empirical literature. Review of Educational Research, 76(2), 173208.

Guin, K. (2004). Chronic teacher turnover in urban elementary schools. Educational Evaluation and Policy Analysis, 12(42), 125.

Hanushek, E. A., Kain, J. F., & Rivkin, S. G. (2004). Why public schools lose teachers. Journal of Human Resources, 39(2), 326354.

Hanushek, E. A., & Rivkin, S. G. (2010). The quality and distribution of teachers under the No Child Left Behind Act. Journal of Economic Perspectives, 24(3), 133150.

Ingersoll, R. M. (2001). Teacher turnover and teacher shortages: An organizational analysis. American Educational Research Journal, 38(3), 499534.

Ingersoll, R. M. (2011). Do we produce enough mathematics and science teachers? Phi Delta Kappan, 92(6), 3741.

Ingersoll, R., & Merrill, L. (2012). Seven trends: The transformation of the teaching force (CPRE Working Paper No. WP-01). Philadelphia: Consortium for Policy Research in Education, University of Pennsylvania.

Ingersoll, R. M., & Perda, D. (2010). Is the supply of mathematics and science teachers sufficient? American Educational Research Journal, 47, 563594.

Ingersoll, R. M., & Smith, T. M. (2003). The wrong solution to the teacher shortage. Educational Leadership, 60(8), 3033.

Johnson, S. M., Kraft, M. A., & Papay, J. P. (2012). How context matters in high need schools: The effects of teachers working conditions on their professional satisfaction and their students achievement. Teachers College Record, 114(10), 139.

Kane, T. J., Rockoff, J. E., & Staiger, D. O. (2006). What does certification tell us about teacher effectiveness? Evidence from New York City. National Bureau of Economic Research Working Paper Series, 12155.

Kelly, A. M., Gningue, S. M., & Qian, G. (2015). First-year urban mathematics and science middle school teachers: Classroom challenges and reflective solutions. Education and Urban Society, 47(2), 132159.

Kim, Y. Y., & McCrory, R. (2007, April). Effects of mathematics courses on prospective elementary teachers learning mathematical content knowledge for teaching. Paper presented at the annual meeting of the American Educational Research Association, Chicago, IL.

Kirby, S., Grissmer, D., & Hudson, L. (1991). Sources of teacher supply: Some new evidence from Indiana. Educational Evaluation and Policy Analysis, 13, 256268.

Kirchhoff, A., & Lawrenz, F. (2011). The use of grounded theory to investigate the role of teacher education on STEM teachers career paths in high-need schools. Journal of Teacher Education, 62(3), 246259.

Klein, J. P., & Moeschberger, M. L. (2005). Survival analysis: Techniques for censored and truncated data (2nd ed.). New York, NY: Springer.

Lankford, H., Loeb, S., & Wyckoff, J. (2002). Teacher sorting and the plight of urban schools: A descriptive analysis. Educational Evaluation and Policy Analysis, 24(1), 3762.

Liou, P.-Y., Desjardins, C. D., & Lawrenz, F. (2010). Influence of scholarships on STEM teachers: Cluster analysis and characteristics. School Science & Mathematics, 110(3), 128143.

Liou, P.-Y., Kirchhoff, A., & Lawrenz, F. (2010). Perceived effects of scholarships on STEM majors commitment to teaching in high need schools. Journal of Science Teacher Education, 21(4), 451470.

Martin, D. B. (2007). Beyond missionaries or cannibals: Who should teach mathematics to African American children? The High School Journal, 91(1), 628.

Monk, D. H. (1994). Subject area preparation of secondary mathematics and science teachers and student achievement. Economics of Education Review, 13(2), 125145.

Moses, R., & Cobb, C. (2001). Radical equations. Boston, MA: Beacon Press.

National Audit Office. (2016). Training new teachers. Retrieved from https://www.nao.org.uk/report/training-new-teachers/

National Center for Education Statistics. (2016). Digest of education statistics. Retrieved from https://nces.ed.gov/programs/digest/d14/tables/dt14_210.30.asp?current=yes

National Research Council. (2010). Preparing teachers: Building sound evidence for sound policy. Washington, DC: National Academy Press.

National Science Foundation. (2017). Robert Noyce Teacher Scholarship Program. Retrieved from https://www.nsf.gov/pubs/2017/nsf17541/nsf17541.htm

Organisation for Economic Co-operation and Development. (2005). Teachers matter: Attracting, developing, and retaining effective teachers. Paris, France: Author.

Papay, J. P., West, M. R., Fullerton, J. B., & Kane, T. J. (2012). Does an urban teacher residency increase student achievement? Early evidence from Boston. Educational Evaluation and Policy Analysis, 34(4), 413434.

Patton, M. Q. (2002). Qualitative research & evaluation methods. Thousand Oaks, CA: Sage.

Peske, H. G., & Haycock, K. (2006). Teaching inequality: How poor and minority students are shortchanged on teacher quality: A report and recommendations by the Education Trust. Washington, DC: Education Trust. Retrieved from http://eric.ed.gov/?id=ED494820

Presidents Council of Advisors on Science and Technology. (2010). Prepare and inspire: K12 education in science, technology, engineering, and math (STEM) for Americas future. Retrieved from https://obamawhitehouse.archives.gov/sites/default/files/microsites/ostp/pcast-stem-ed-final.pdf 

R Core Team. (2016). R: A language and environment for statistical computing [Computer software]. Retrieved from https://www.R-project.org

Ronfeldt, M. (2012). Where should student teachers learns to teach? Effects of field placement school characteristics on teacher retention and effectiveness. Educational Evaluation and Policy Analysis, 34(1), 326.

Ronfeldt, M., Lankford, H., Loeb, S., & Wyckoff, J. (2011). How teacher turnover harms student achievement (NBER Working Paper No. 17176). Cambridge, MA: National Bureau of Economic Research.

Ronfeldt, M., Loeb, S., & Wyckoff, J. (2013). How teacher turnover harms student achievement. American Educational Research Journal 50(1), 4­36.

Shuls, J. V., & Trivitt, J. R. (2013). Teacher effectiveness. Educational Policy, 29(4), 645675.

Steele, J. L., Murnane, R. J., & Willett, J. B. (2010). Do financial incentives help low-performing schools attract and keep academically talented teachers? Evidence from California. Journal of Policy Analysis and Management, 29(3), 451478.

Tatto, M. T., & Senk, S. (2011). The mathematics education of future primary and secondary teachers: Methods and findings from the Teacher Education and Development Study in Mathematics. Journal of Teacher Education, 62(2), 121137.

Teach for America. (2017). Our mission. Retrieved from https://www.teachforamerica.org/about-us/our-mission

Worsham, H. M., Friedrichsen, P., Soucie, M., Barnett, E., & Akiba, M. (2014). Recruiting science majors into secondary science teaching: Paid internships in informal science settings. Journal of Science Teacher Education, 25(1), 5377.

Cite This Article as: Teachers College Record Volume 121 Number 2, 2019, p. 1-36
https://www.tcrecord.org ID Number: 22554, Date Accessed: 10/16/2021 7:30:37 PM

Purchase Reprint Rights for this article or review
Article Tools
Related Articles

Related Discussion
Post a Comment | Read All

About the Author
  • William Zahner
    San Diego State University
    E-mail Author
    WILLIAM ZAHNER is an assistant professor of mathematics education in the Department of Mathematics and Statistics and the associate director of the Center for Research in Mathematics and Science Education at San Diego State University. His research focuses on understanding principles for designing equitable mathematics classroom learning environments in diverse secondary classrooms. Zahner was formerly on the faculty in the School of Education at Boston University. Zahner’s published work has appeared in Learning and Instruction, Educational Studies in Mathematics, The Journal of Mathematical Behavior, and Mathematical Thinking and Learning.
  • Suzanne Chapin
    Boston University
    E-mail Author
    SUZANNE CHAPIN is professor of mathematics education at Boston University, where she teaches graduate-level courses and conducts research. Her work covers the areas of gifted education, curriculum design, teacher professional development in mathematics, and teacher and student discourse in mathematics. Dr. Chapin is principal investigator of the Elementary Mathematics Project, where she is involved in writing problem-based mathematics tasks and researching the effects of these tasks, in conjunction with discourse-based pedagogy, on elementary education majors’ knowledge of mathematics for teaching. Chapin is the author of multiple books and textbooks, including Classroom Discussions: Using Math Talk to Help Students Learn and Math Matters: Understanding the Math You Teach Grades K–8 (both published by Math Solutions).
  • Rich Levine
    San Diego State University
    E-mail Author
    RICH LEVINE is professor of statistics and institutional research faculty advisor at San Diego State University. He is a Fellow of the American Statistical Association, with research interests in learning analytics and computationally intensive machine learning methods. He is currently working with his collaborators and students on applications of random forests in educational data mining settings. This work has recently been published in the Journal of Educational Data Mining, International Journal of Artificial Intelligence in Education, and Technology, Knowledge, and Learning.
  • Lingjun He
    San Diego State University
    E-mail Author
    LINGJUN A. HE is a Ph.D. data scientist at the Analytic Research and Institutional Research Center and an adjunct faculty member in the Computational Science Research Center at San Diego State University. Her primary research interests include learning analytics, educational data mining, semiparametric mixed-effects modeling, and financial time series. Her work has appeared in publications such as International Journal of Artificial Intelligence in Education and Journal of Biological Chemistry.
  • Robert Afonso
    Boston University
    E-mail Author
    ROBERT AFONSO is a doctoral student in mathematics education at Boston University. He currently teaches middle school mathematics in Rhode Island. In addition to studying the career trajectories of secondary mathematics teachers, Afonso has contributed to research on the development of curriculum materials for mathematics courses for prospective elementary teachers.
Member Center
In Print
This Month's Issue