Abstract
This study is motivated by an ongoing debate about the kinds of schools that make for the best field placements during pre-service preparation. On the one hand, easier-to-staff schools may support teacher learning because they are typically better-functioning institutions that offer desirable teaching conditions. On the other hand, such field placements may leave new teachers unprepared to work in difficult-to-staff schools and with underserved student populations that need high quality teachers the most. Using administrative and survey data on almost 3,000 New York City teachers, their students, and their schools, this study finds that learning to teach in easier-to-staff field placement schools has positive effects on teacher retention and student achievement gains, even for teachers who end up working in the hardest-to-staff schools. The proportion of poor, minority, and low-achieving students in field placements is unrelated to later teacher effectiveness and retention suggesting something beyond student populations explain these results.
Purpose
Researchers and policymakers agree that providing all K–12 students a quality education depends largely upon our capacity to staff schools with highly effective teachers. Recent studies demonstrating the effects of teachers on student achievement gains highlight just how critical it is to improve the quality of our teacher workforce (Rivkin, Hanushek, & Kain, 2001; Rockoff, 2004). Recognizing that teacher education can play an important role in improving teacher quality, a growing number of studies are focusing on the effects of teacher preparation (Cochran-Smith & Zeichner, 2005). Even so, we still lack a strong research base that identifies specific dimensions of teacher education related to the preparation and retention of high-quality teachers (Boyd, Grossman, Lankford, Loeb, & Wyckoff, 2009; Wilson, Floden, & Ferrini-Mundy, 2001).
This article examines the effects of one component of teacher education—the kinds of schools used for field placements during pre-service preparation. In particular, it provides evidence for a commonly debated topic among teacher educators, with limited prior empirical support: Should we place student teachers in difficult-to-staff and underserved field placement schools to learn to teach? The following research questions guide this study:
Are teachers who learn to teach in difficult-to-staff and underserved field placements more or less likely to leave the profession in their first 5 years of teaching?
Do these teachers have higher or lower student achievement gains than similar teachers who had field placements in easier-to-staff, less underserved settings?
Do the effects of learning to teach in difficult-to-staff field placement schools on teacher retention and student achievement gains vary by the student populations in these settings?
Do the results differ for teachers who become teacher of record in difficult-to-staff and underserved schools?
There is no consensus in the literature about what is meant by “difficult-to-staff” and “underserved” schools, or how to measure these kinds of schools (Feng, 2009). As prior literature has shown, schools with higher teacher turnover tend also to have higher proportions of poor, minority, and low-achieving student populations (Hanushek, Kain, & Rivkin, 2004; Lankford, Loeb, & Wyckoff, 2002). Schools with higher teacher turnover also have poorer working conditions (poorer administrative support, induction support, facilities, teaching assignments, etc.) and lower teacher salaries (National Academy of Education, 2009). Although some suggest that teacher turnover is largely explained by student demographics and performance (e.g., Scafidi, Sjoquist, & Stinebrickner, 2007), Loeb, Darling-Hammond, and Luczak (2005) find that the relationship between student demographics and teacher turnover is largely explained by school working conditions and salary.
There is a tendency to conflate schools with high teacher turnover, schools with many underserved students, and schools with poor working conditions, but findings by Loeb et al. (2005) suggest examining the effects of student demographics separate from other school features. Some schools have underserved student populations that are high-functioning institutions; there also exist poor-functioning schools that serve more privileged student populations. This study employs different measures to distinguish “difficult-to-staff” from “underserved” schools. Namely, it uses a school-level measure of the proportion of teachers that stay in the same school from year to year (“stay-ratio”) to signal how easy-to-staff a school may be. As will be described later, easier-to-staff (higher stay-ratio) schools exhibit many of the features one would expect from better functioning schools that offer desirable conditions for professional practice. Separate from this measure for how “difficult-to-staff” a school is, this study also employs measures for the proportions of typically underserved student populations (poor, Black, and low-achieving) to signal “underserved” schools.
Literature Review
A common theme in the existing literature on pre-service teacher education is that new teachers perceive field experiences, including student teaching, to be the most influential part of their preparation (Guyton & McIntyre, 1990; Hollins & Guzman, 2005; Wilson et al., 2001). However, studies on pre-service field experiences typically focus on changes in teacher attitudes, perceptions, and beliefs, thus failing to make direct links to student outcomes (Wilson et al., 2001). Moreover, it is difficult to generalize from research in this area because most studies include small samples and are case studies, typically self-studies, of a single institution that prepares teachers. In their review of the literature, Wilson and Floden (2003) conclude that findings are “thin and inconclusive” and that we lack “reliable and valid measures of impact as well as insights into what specific features of field experiences are more or less effective” (pp. 20–21).
Responding to these gaps in the literature, Boyd, Grossman, Lankford, et al. (2009) examined the effects of various dimensions of teacher education, including field experience, on teachers’ value-added to student achievement across all institutions responsible for preparing teachers in New York City (NYC). They found that program oversight of field experiences was positively and significantly associated with teacher effects. More specifically, new teachers who graduated from programs that were actively involved in selecting field placements had minimum experience thresholds for cooperating teachers and required supervisors to observe student teachers at least five times had higher student achievement gains in their first year as teacher of record (Boyd, Grossman, Lankford, et al., 2009). This study is consistent with prior, qualitative studies identifying field experience as an influential component of pre-service preparation but makes a substantial contribution by providing evidence using student outcomes and by identifying specific aspects of field experiences—namely, features of program oversight—that make a difference.
Boyd, Grossman, Lankford, et al. (2009) suggest that teacher education programs take an active role in selecting and overseeing field sites but do not examine the kinds of schools that make for better or worse field placements. As with much prior work on pre-service field experience, Boyd, Grossman, Lankford, et al. (2009) focus on the relationship, connection, or coherence between the university and K–12 schools used for field experiences rather than the effects of the field placement school characteristics apart from the effects of other program dimensions. Despite it being a practical question faced by field experience coordinators regularly, few studies have examined the types of schools that make for better field placements. To my knowledge, no prior studies have looked across a wide array of institutions that offer teacher preparation to examine the relationship between field placement school-level characteristics and teacher effectiveness and retention.
In terms of school-level features receiving attention in the literature on field experience, most studies have looked at the effects of learning to teach in urban, multicultural, or professional development school (PDS) settings. With the growing diversity in national student demographics, alongside a relatively homogenous teaching population, much literature has advocated for including pre-service field experiences in school settings with diverse student populations as part of preparing teachers to meet the changing demands of the profession (Grant & Secada, 1990). A central supposition behind this literature is that new teachers, who are typically White and often from nonurban backgrounds, benefit from guided immersion experiences with students from different backgrounds during professional preparation, especially because many teachers will eventually work in urban schools and with diverse student populations. Studies looking at the effects of requiring these kinds of field placements provide evidence for generally positive shifts in teacher attitudes and beliefs toward multicultural and urban student populations (Adams, Bondy, & Kuhel, 2005; Athanases & Martin, 2006; Burant & Kirby, 2002; Downey & Cobbs, 2007; Hill, Phelps, & Friedland, 2007). However, urban and multicultural field placements are often challenging settings for learning to teach that sometimes perpetuate negative stereotypes and attitudes (Burant & Kirby, 2002; Buehler, Ruggles Gere, Dallavis, & Shaw Haviland, 2009, Cross, 2003; Hill et al., 2007; Mello, 2003). Less is known about the effects of urban field placements on teacher retention and student outcomes. Two recent studies examined whether paid field experiences in urban, high-needs schools increased the likelihood that teacher candidates would take jobs in similar kinds of schools, but results were mixed (Grande, Burns, Schmidt, & Marable, 2009; Helfeldt, Capraro, Capraro, Foster, & Carter, 2009).
Among the existing literature on field placement school characteristics, studies about the effects of PDS are also relatively common. 1 Like teaching hospitals in the medical profession, PDS settings promise to serve the learning goals of professional preparation programs, in addition to the aims of the clinical settings (K–12 schools) themselves (Carnegie Forum on Education and the Economy, 1986; Stallings & Kowalski, 1990). Those who oppose PDS believe that they can be “too far removed from the mainstream of school life to be credible” (Stallings & Kowalski, 1990, p. 262) and, in serving university goals, may come “at the expense of attention to issues of equity” (Boyle-Baise & McIntyre, 2008, p. 324). The literature indicates that teacher preparation in PDS settings has generally positive impacts on teachers’ beliefs, attitudes, practices, recruitment, and retention (Castle, Fox, & Souder, 2005; Clift & Brady, 2005; Latham & Vogt, 2007; Paese, 2003; Ridley, Hurwitz, Davis Hackett, & Knutson Miller, 2005).
Although relatively little is known about the effects of pre-service field placement school characteristics, there is growing evidence that in-service school working conditions significantly impact teachers’ career decisions and induction. Boyd, Grossman, Ing, Lankford, Loeb, and Wyckoff (2009) found that many dimensions of the school working conditions—including staff relations, facilities, students, safety, and teacher influence—significantly influenced teachers’ decisions to transfer and leave NYC schools; of these features, administration quality and support had the largest impact. Similarly, Johnson and Birkeland (2003) found that school working conditions—including administrative support and collegiality—can help teachers achieve a “sense of success” and, thus, a commitment to remain in schools. Looking at alternative route programs, Johnson and Birkland (2008) found that the qualities of the school setting exert a strong influence on induction into the profession and ongoing teacher learning:
In some cases, an unsupportive workplace totally undermined any confidence the candidate had gained in his training and discouraged any further learning on the job. In other cases, a school that was well organized for the induction and continued growth of new teachers encouraged a candidate to feel much better prepared and more optimistic about her work as a teacher. (p. 125)
Little evidence exists on whether these same features in pre-service settings have similar effects.
Competing Theories of Action
Even without a strong empirical base upon which to draw, teacher educators are divided about where to place student teachers to learn to teach. Their debate can be roughly characterized by two competing conjectures on how the settings for student teachers influence teacher learning. These competing conjectures, in turn, represent two theories of action that guide this study’s design.
Because difficult-to-staff and underserved schools are often more challenging settings even for experienced teachers, many teacher educators contend that it is best to place student teachers in easier to staff and less underserved schools to learn to teach. These teacher educators argue that easier to staff schools are typically better functioning and thus promise a more supportive setting for developing professional practice, rather than overwhelming student teachers with classroom management issues, for example. Easy-to-staff schools attract and keep teachers year after year, indicating that they provide the conditions for supporting teachers to do their work effectively. Given they are desirable settings for in-service teacher practice, easier-to-staff schools hold promise as desirable contexts for inducting student teachers into the profession as well.
Other teacher educators argue that learning to teach in easier-to-staff schools, which tend to have more privileged student populations, will leave teachers unprepared for the specific challenges that come with working in difficult-to-staff schools and with the underserved student populations that typically attend such schools. Difficult-to-staff field placements, by contrast, likely provide student teachers more opportunities to learn to be effective under difficult working conditions and with typically underserved student populations. Learning to teach in these settings might better prepare teachers for realities in schools and, thus, lead teachers to be more effective and enduring. Given the need to prepare teachers to work in underserved communities, placing new teachers in difficult-to-staff field placements during pre-service preparation—when university and school mentors are readily available to offer support—may be the best way to prepare them to succeed in similar kinds of schools as they become full-time professionals. Haberman and Post (1998) advocate strongly for this position when they write,
[Teacher] training is most effective when it is offered in the worst schools under the worst conditions of work. Traditional teacher education and state certification agencies make the reverse assumption. They create professional development centers engaged in best practices and then certify graduates universally. The naïve assumption is that graduates will be able to function in the worst school situations because they have observed good practices. (p. 103)
2
Despite the debate among teacher educators, there is little empirical support for either position. This study provides some of the first information on the link between field placement school characteristics and later student and teacher outcomes.
Data
The data for this study come from a survey of NYC first-year teachers and NYC Department of Education (NYCDOE) administrative data on students, teachers, and schools. The Teacher Policy Research group administered a survey in the spring of 2005 to all NYC teachers in their first year, with a response rate of over 70%. 3 The survey was part of a larger study examining the various pathways of teacher preparation in NYC, focusing on the cohort of teachers who went through both alternative and traditional preparation routes in 2003–2004. The survey was designed to collect data on many aspects of this cohort’s experiences in their first year of teaching, in addition to retrospective data on their preparation and backgrounds. The survey asked new teachers to identify the schools in which they did most of their field experience (including summer experiences and student teaching), making it possible to identify field placement schools for 2,860 out of 4,303 survey respondents; this defined the upper bound for this study’s sample. Those teachers for whom it was not possible to identify field placements likely completed field experiences outside of NYC. Survey respondents were then linked to NYCDOE administrative data on these teachers, the schools in which they taught, and their students. These data are summarized the following sections.
Teachers
Table 1 describes the teachers that make up this study’s sample, as well as the population of all NYC first-year teachers in 2004–2005 from which this sample was drawn. The teachers in this study’s sample were, on average, 30 years old, 65% white, and about three-quarters female. Over half of the teachers came from undergraduate institutions that were designated in the top two of Barron’s four-level rating for college competitiveness. More respondents entered teaching through the Teaching Fellows program (47%) than any other preparation route. Teaching Fellows is an early-entry or alternative route program that filled much of the void left when NYC phased out the employment of teachers with temporary licenses. College-recommending (traditional route) teachers composed 37% of survey respondents.
Average Teacher Characteristics of NYC First Year Teachers in 2004–2005
Table 1 also shows that this study’s sample of teachers was fairly representative of all NYC first-year teachers in 2004–2005 in terms of gender, race, and age. The main difference between groups was in regards to route of preparation. As compared to the overall population of NYC first-year teachers, teachers in this study’s sample included much higher proportions of teachers from the Teaching Fellows and Teach for America preparation routes and much lower proportions of teachers from “other” and independent evaluation preparation routes. Somewhat fewer teachers in this study’s sample also entered teaching through college-recommending programs. These differences may have resulted, at least in part, from Teaching Fellows and TFA teachers having better response rates on the survey than teachers from other preparation routes. 4 Additionally, teachers from the independent evaluation and “other” preparation pathways were less likely to use NYC schools as field placement sites and therefore tended to get dropped from this study’s sample. 5 Differences in routes of preparation probably explain why teachers in this study’s sample also had somewhat higher pre-service qualifications (LAST teacher certification exam scores and college competitiveness), as alternative route programs like TFA and Teaching Fellows typically recruit teacher candidates with better academic qualifications.
Schools
Table 2 summarizes the characteristics of the subsample of NYC schools in which this study’s teachers had been placed during their field experiences, along with comparative statistics on all NYC schools. 6 Because this study focuses on the effects of learning to teach in “difficult-to-staff” and “underserved” schools, the top rows summarize school characteristics used to signal these kinds of schools. More typical measures include the proportion of Black, Hispanic, poor, and lowest performing students. In field placement schools, about three quarters of students were either Black or Hispanic, with over two thirds qualifying for free or reduced priced lunch and 14% of students taking the lowest level math exam.
Average NYC School Characteristics (2001–2002 to 2006–2007)
Note. ELL = English language learners.
Even though schools that enroll typically underserved populations also tend to be harder to staff, it is important to recognize that there exist schools with high proportions of underserved populations that are also easy to staff. By including a separate measure for “difficult to staff” that goes beyond student demographics, I hoped to distinguish poorly functioning schools, or schools with undesirable conditions for teacher practice, from schools with higher proportions of underserved students. Specifically, I generated the “stay-ratio,” which is essentially a school-level measure for average teacher turnover. The assumption here is that better functioning schools with more desirable conditions for practice will be ones in which teachers tend to want to stay (easier to staff). Using NYCDOE work history files, for every school, I identified the number of teachers who stayed in the same school (stayers) and the number of teachers who moved to a new school (movers) from one year to the next. In its simplest form, the stay-ratio is the proportion of stayers out of the combined total of stayers and movers, averaged over 5 schools years (2003–2004 to 2007–2008). Field placement schools, on average, kept 92% of their teachers out of all who remained in teaching in NYC schools from one year to the next. This statistic is almost the same as it is for all NYC schools, 91%. I transformed the measure using the exponential, centered it by school level, and standardized it to have a mean of 0 and standard deviation of 1. 7
In addition to measures that signal “difficult-to-staff” and “underserved” schools, Table 2 summarizes other school characteristics used as controls in this study. Of the schools used as field placements by teachers in this study’s sample, 56% were elementary schools, 23% were high schools, and 21% middle schools. These field placement schools had average enrollments of almost 1,000 students, attendance rates of about 91%, and 4% of students receiving suspensions. A fifth of these schools were new since 1998. Additionally, field placement schools had an average expenditure per pupil of $10,701 and 51% of faculty with 5 or more years of experience.
Table 2 also demonstrates that teacher education programs may have used field placement sites that differed somewhat from NYC schools in general. Compared to all NYC schools, those used for field placements during pre-service preparation had slightly lower concentrations of Black students and students who qualified for free or reduced-priced lunch, but slightly higher concentrations of Asian, English language learners (ELL), and lowest performing students. Although over one third of all NYC schools were new since 1998, only about one fifth of schools used for field experiences were new. Compared to all NYC schools, field placement schools had larger enrollments, a higher proportion of experienced teachers, and slightly higher stay-ratios (by about 13% of a standard deviation). Student attendance rates and expenditure per pupil were similar between groups.
Stay-Ratio Measure Validity Check
As an initial validity check for whether the school stay-ratio signaled better functioning schools with more desirable conditions of practice, I linked administrative data to survey data to see what in-service teachers reported about higher stay-ratio schools in which they were currently teaching. 8 Table 3a summarizes results from linear regression models that estimate school stay-ratio as a function of individual survey factors signaling different dimensions of school context. 9 Results indicate that teachers in schools with higher stay-ratios (easier to staff) reported many of the features we would expect from better functioning schools with desirable conditions for teacher practice—higher quality administration and support, better staff relations and collegiality, fewer incidents of teachers being threatened or attacked, cleaner and more adequate facilities, and better student habits, behaviors, and parental support. These results are consistent with prior research demonstrating that lower turnover schools have better administrative support (Darling-Hammond & Prince, 2007; Ingersoll, 2001); facilities, parental involvement, and professional development (Loeb et al., 2005); enrollment and student discipline (Ingersoll, 2001); and staff relations and support (Bryk & Schneider, 2002).
Estimating Current School Stay-Ratio as a Function of Teachers’ Perceptions of School Context (School-Level Analysis Using School Context Factors From Survey)
p < 0.1. *p < 0.05. **p < 0.01. ***p < 0.001.
Estimating Current School Stay-Ratio as a Function of Other School Features (School-Level Analysis Using Measures From Administrative and Survey Data)
p < 0.1. *p < 0.05. **p < 0.01. ***p < 0.001.
Table 3b presents results from models that include administrative data on school characteristics entered together with the survey factors from Table 3a. As before, easier-to-staff (higher stay-ratio) schools were associated with features that are typical of better functioning schools—more experienced faculty, higher attendance and enrollment rates, and fewer suspensions. 10 Consistent with prior literature, schools with higher turnover (lower stay-ratio) also tended to enroll poor, low-achieving, and minority students (Scafidi et al., 2007). Model 3 shows that staff relations and adequacy of facilities were no longer statistically significant when controlling for other survey predictors for school context. After controlling for administrative data in Model 4, administrative quality/support was the only survey measure that predicted school stay-ratio at statistically significant levels.
In addition to providing in-service teachers with more desirable conditions for professional practice, it is possible that higher stay-ratio schools also provide pre-service teacher candidates with better conditions for professional learning. To examine this possibility, I analyzed survey questions that asked first-year teachers to retrospectively rate their experiences in their field settings during pre-service preparation, particularly regarding feedback and supervision. After linking school stay-ratio data to survey respondents’ field placement sites, I used linear regression to estimate the field placement school stay-ratio as a function of teachers’ perceptions of their experiences in their placement sites. As summarized in Table 3c, teachers who had field placements in higher stay-ratio schools were more likely to report observing excellent teachers and role models, being observed regularly by K–12 school and/or university faculty, receiving useful feedback, and having opportunities to try out strategies from pre-service coursework. These findings suggest that easier-to-staff schools are indeed associated with many of the characteristics one would expect from more desirable settings for student teacher learning, including better modeling, opportunities for practice, and feedback.
Estimating Field Placement School Stay-Ratio as a Function of Teachers’ Perceptions of Prior Pre-Service Field Experiences, Supervision, and Feedback (Teacher-Level Analysis Using Survey Item A13 a, b, c, e)
p < 0.1. *p < 0.05. **p < 0.01. ***p < 0.001.
Outcome Measures
For retention analyses, I created a binary outcome variable for whether teachers had left NYC schools by the beginning of each academic year from 2005–2006 to 2008–2009. Teachers were assigned a “0” if they continued to teach in the same or a different NYC school and a “1” if they left NYC schools (this includes teachers who retired, resigned, transferred out of district or out of state, etc.). After a teacher was coded as having left NYC schools, she or he was dropped from subsequent years in the sample. Those who took leave (paid or not), became substitute teachers, or were of unknown status were also dropped from the sample.
Student achievement analyses were based upon test performance data from New York State and the NYCDOE. Prior to 2006, New York State administered examinations in mathematics and English language arts to Grades 4 and 8. Additionally, the NYCDOE tested third, fifth, sixth, and seventh graders in these subjects. All exams were aligned to the New York State learning standards, and IRT methods were used to convert raw scores into scale scores. New York State began administering all the tests in 2006. 11
Increasingly, researchers and policymakers are estimating teacher effectiveness by turning to value-added modeling strategies because these approaches look at growth in student test scores rather than raw scores, which are highly correlated with prior achievement. Despite remarkable advancements in using value-added modeling to estimate teacher effectiveness based on students’ test scores, there are many concerns over whether such models generate valid or reliable measures of teacher effectiveness. Among them, researchers have expressed concern over year-to-year variation in scores (McCaffrey, Sass, Lockwood, & Mihaly, 2009) and bias due to nonrandom assignment of students to teachers (Rothstein, 2009). Nevertheless, this study uses teachers’ value-added to student test scores as a proxy for “teacher effectiveness.” Any mention of teacher effectiveness refers narrowly to teachers’ effectiveness at raising students’ test scores as measured by value-added models described subsequently.
This study used the student test data described above to model math achievement gains from 2004–2005 to 2007–2008 as a function of student time-invariant and time-varying controls, school time-invariant and time-varying controls, and classroom controls. Student characteristics included race/ethnicity, poverty status, days absent during the prior year, and suspensions in the prior year. Class characteristics were the aggregate (teacher by grade by school by year) of these same student characteristics but included also the average of student test scores and their standard deviations in the prior year. The school-level characteristics used as controls are outlined in Table 2, excluding the stay-ratio and school age measures. The models also include fixed-effects for grade, year, and the school that students attended, so that the comparisons for each teacher were with other teachers in the same school, grade level, and year. After adjusting for measurement error with teacher-by-year fixed-effects, I used these as measures for teacher effectiveness in models of the relationship between field experience and teacher effectiveness. The benefit of this approach over a student-level analysis was that it used data on all teachers of students with test scores, not just teachers who responded to the survey about field experiences, to model teachers’ value-added to math achievement. Math was tested only in Grades 3 through 8, and not all teachers taught math, thus reducing the number of observations available for analysis of teacher effectiveness. The analyses of teachers’ value-added then depends on a relatively small subsample of teachers that excludes those who teach grades pre-K–2 or 9–12 and those who either do not teach mathematics or who teach mathematics but could not be linked to value-added data. Both value-added measures and field placement data were available for 828 teachers. 12 Of these 828 teachers included, 39% of them had 1 year of value-added scores, 17% had 2 years, 23% had 3 years, and 21% had 4 years.
Method
A major challenge to studying the effects of field experiences is that teacher candidates are not randomly assigned to placement sites. To isolate the effects of features of field placements on teacher outcomes, it is then necessary to account for alternative factors likely to influence these outcomes. For retention analyses, I used discrete-time hazard models to estimate the likelihood of leaving NYC schools as a function of teacher characteristics, field placement school characteristics, and current school characteristics. Equation 1 describes these analyses:
Here, the probability (P) that teacher t, in year y, in current school s, having been trained in field placement school f, leaves NYC schools is a function of time-invariant and time-varying teacher characteristics including teacher preparation program fixed-effects (T), current school characteristics (S), field placement school characteristics (F), a school year indicator variable (δ y ), and a random error term (ε tysf ). I also ran models with current school fixed-effects replacing current school controls. Each teacher had a separate observation for each year that she or he remained teaching full-time in any NYC school. After a teacher left NYC schools, she or he was dropped from subsequent years. Tables 1 and 2 outline the teacher and school variables included in these analyses.
Because there are substantive differences between routes of preparation, teacher education program fixed-effects were included in models that control for teacher characteristics.13 Among these differences, many alternative route programs require field experiences that occur over the summer, and so are substantially shorter in length than are traditional route field experiences, and in a different set of schools (those open during the summer). Including program fixed-effects allows within-program comparisons of the effects of different field placement school characteristics. The effect of a given field placement school characteristic on a given teacher’s retention (or effectiveness) is compared to other teacher candidates who graduated from her same program. Thus, this analytic strategy separates the effect of field placement school characteristics from other dimensions of professional preparation that are often difficult to disentangle.
I used linear regression to model a teacher’s value-added to math achievement as a function of characteristics of the teacher, his or her field placement school, and his or her current school. Because value-added scores were available from 2004–2005 to 2007–2008, I clustered standard errors at the teacher level to account for non-independence of observations from teachers for whom multiple years of data were available. Equation 2 describes these analyses:
In the above equation, the value-added to student achievement (V) of teacher t, in year y, in school s, having been trained in field placement school f, is a function of time-varying and time-invariant teacher characteristics including teacher preparation program fixed-effects (T), current school characteristics (S), pre-service field placement school characteristics (F), a school year indicator variable (δ y ), and a random error term (ε tysf ). Tables 1 and 2 outline the teacher and school characteristics included in these analyses. In some models, I replaced current school controls with current school fixed-effects, leaving the rest of the model unchanged.
Results
Are Teachers Who Learn to Teach in Hard-to-Staff, Underserved Field Placements More or Less Likely to Leave NYC Schools Once They Become Teacher of Record?
Table 4 presents logit model estimates for the likelihood of leaving NYC schools as a function of pre-service field placement school characteristics. Coefficients are odds ratios, so any estimate less than 1 indicates a negative effect. All models include teacher preparation program fixed-effects, teacher characteristics, and school year indicators. Models 1–5 include controls for current school characteristics. Models 1–4 show results for models that enter predictors one at a time, estimating the likelihood of leaving NYC schools as a function of individual field placement characteristics that signal hard-to-staff and underserved schools. Model 5 combines all of these predictors together along with many other field placement school controls (e.g., attendance rates, other student demographics, whether a new school). Model 6 is the same as Model 5 except it replaces current school fixed-effects for current school characteristics as controls.
Estimating the Probability of Leaving NYC Schools in the First 5 Years of Teaching as a Function of Field Placement School Characteristics (Estimates as Odds Ratios)
p < 0.1. *p < 0.05. **p < 0.01. ***p < 0.001.
Of the school characteristics used to signal difficult-to-staff and underserved field placements, only school stay-ratio was consistently significant across model specifications. The results show that teachers who learned to teach in field placement schools with higher stay-ratios (easier to staff) were less likely to leave NYC schools in their first 5 years of teaching. Models 1 and 5 indicate that the odds of leaving were between 14% and 22% lower for teachers prepared in schools with 1 standard deviation higher stay-ratios. The substantially higher standard errors and bigger effects in Model 6 raise some concern about the precision of the estimates in models using school fixed-effects. Model 6 estimates suggest that teachers trained in schools with 1 standard deviation higher stay-ratios were more than 50% less likely to leave NYC schools.
None of the other measures signaling “underserved” field placement schools were significantly related to retention in Models 1 through 4. That is, teachers who learned to teach in field placement schools with higher proportions of poor, Black, and lowest achieving students were no more or less likely to remain in NYC schools. In Model 6, the percentage of students on free or reduced priced lunch at field placement schools significantly predicted teacher retention. Because the direction of the coefficient changes across different model specifications, and given the concerns expressed above about using current school fixed-effects, it is unclear if this is a true effect.
Are Teachers Who Learn to Teach in Hard-to-Staff, Underserved Field Placements More or Less Effective Once They Become Teacher of Record?
Table 5 presents the results for models estimating teachers’ value-added to student achievement as a function of field placement school characteristics that signal difficult-to-staff, underserved schools. As with retention analyses, all models include teacher preparation program fixed-effects, teacher characteristics, and school year indicators. Models 1 through 5 include controls for current school characteristics. Models 1 through 4 display estimates of teacher effects as a function of individual field placement characteristics that are proxies for hard-to-staff, underserved schools. Model 5 shows estimates for teacher effectiveness as a function of the field placement characteristics from Models 1 through 4 combined, in addition to many other field placement school controls. Model 6 is the same as Model 5 except it replaces current school controls with current school fixed-effects.
Estimating Teacher Value-Added as a Function of Field Placement Characteristics
p < 0.1. *p < 0.05. **p < 0.01. ***p < 0.001.
Similar to retention analyses, school stay-ratio was the only field placement characteristic consistently significant across model specifications. The coefficients were always positive, demonstrating that teachers who learned to teach in field placements with higher stay-ratios (easier-to-staff schools) had better student test score gains after becoming teacher of record. A 1 standard deviation increase in field placement stay-ratio was associated with between 7% and 11% of a standard deviation increase in teachers’ value-added to student achievement.14 Based on estimates from Boyd, Grossman, Lankford, Loeb, and Wyckoff (2006) and Boyd, Lankford, Loeb, Rockoff, and Wyckoff (2008), this is roughly comparable to the effect of between a half and a full year of initial teaching experience on student achievement gains.
According to Table 5, estimates for the effect of the proportion of lowest achieving students were consistently positive across models and statistically significant in Model 6. These findings provide some weak evidence that learning to teach in field placements with more low-performing students is associated with being more effective at raising test scores as teacher of record, after controlling for the effects of other field placement school characteristics. The instability of the estimates across model specifications demands some caution in interpreting the results, however, especially because the estimates reach levels of statistical significance only in models with current school fixed effects. Especially given the relatively small subsample of teachers for whom value-added data was available, there is good reason for concern about the precision of the estimates employing school fixed effects. Results from Table 5 demonstrate also that teachers who learned to teach in schools with higher proportions of Black and poor students were no more or less effective as teacher of record.
School stay-ratio was the only measure for difficult-to-staff or underserved field placement schools that was statistically significant across model specifications and outcome measures. Teachers who learned to teach in field placement schools with higher stay-ratios were more likely to remain in NYC schools and were also more effective. However, from both a policy and research perspective, it would be helpful to know if the effect of being prepared in higher stay-ratio schools varies by the concentration of poor, minority, or lowest achieving students in these schools. Might the effect of learning to teach in easier-to-staff (higher stay-ratio) schools, for instance, be stronger if these schools also have higher concentrations of historically underserved student populations? I turn to this question in the section that follows.
Do the Effects of Learning to Teach in Difficult-to-Staff Field Placement Schools on Teacher Retention and Effectiveness Vary by the Student Populations in These Settings?
To begin to examine the relationship between the effects of stay-ratio and other field placement school characteristics, I created interaction terms to include in both retention and value-added analyses. Table 6 presents the estimates for the likelihood of leaving NYC schools as a function of interactions between field placement stay-ratio and other field placement school characteristics. None of the interaction estimates were statistically significant, suggesting that the effect of field placement stay-ratio on retention does not vary significantly as the proportion of Black, poor, and lowest achieving students varies.
Estimating the Probability of Leaving NYC Schools as a Function of Interactions Between Field Placement School Characteristics (Estimates as Odds Ratios)
p < 0.1. *p < 0.05. **p < 0.01. ***p < 0.001.
Table 7 examines the effect of these same interaction terms as they relate to teachers’ value-added to student achievement instead of retention. The positive and significant estimate on the interaction between stay-ratio and percentage of Black students suggests that the positive effect of stay-ratio on student achievement gains increases as the proportion of Black students within the field placement setting also increases. To explore this relationship further, Table 8 juxtaposes the effect of stay-ratio for teachers who were trained in schools above and below the field placement school median for percentage of Black students (Models 1 and 2, respectively). The results demonstrate that the positive relationship between stay-ratio and teacher effectiveness is four times stronger in schools above the median on percentage Black, as compared to those below the median. In other words, the positive effects of learning to teach in easier-to-staff schools are more pronounced when these schools also serve large populations of Black students.
Estimating Teachers’ Value-Added to Student Achievement as a Function of Interactions Between Field Placement School Characteristics
p < 0.1. *p < 0.05. **p < 0.01. ***p < 0.001.
Exploring the Relationship Between the Effect of Field Placement Stay-Ratio and Percentage Black on Teacher Effectiveness
p < 0.1. *p < 0.05. **p < 0.01. ***p < 0.001.
To explore this interaction from another perspective, Table 8 also presents estimates for the effect of Black student percentage on teacher effectiveness at low (below the median) and high (above the median) stay-ratio field placements. The results show that among high stay-ratio (easier-to-staff) field placement schools, increasing the percentage of Black students has a moderately positive effect on teachers’ value-added to student achievement, whereas there is no effect in lower stay-ratio field placement. This is weak evidence that using field placements with more Black students may have positive effects on teacher effectiveness, conditional on these field placements also being easier-to-staff schools.
Although learning to teach in higher stay-ratio field placements (easier-to-staff schools) is associated with being more effective and enduring in the aggregate, a possible concern remains to be studied directly—whether learning to teach in easier-to-staff schools leaves new teachers unprepared to work in the hardest-to-staff and most underserved schools. In the section that follows, I examine this topic.
Are the Results Different for Teachers Who Become Teacher of Record in the Most Difficult-to-Staff and Underserved Schools?
As described in the introduction, some have speculated that learning to teach in easier-to-staff schools may help candidates who become full-time teachers in similar types of schools but may disadvantage those who end up in difficult-to-staff and underserved schools. Many who hold this view believe that preparing teachers in difficult-to-staff and underserved schools, particularly during pre-service preparation when mentorship and guidance is most readily available, is a better way to support new teachers to become effective in similar settings. Those who promote field experience in multicultural settings, for example, often believe that new teachers—who are typically White and often from suburban backgrounds—need exposure to students from different backgrounds to support the development of cultural competency. Others like Haberman and Post (1998) believe that if our goal is to prepare teachers to function effectively in schools with poor working conditions, then teacher training should occur in “schools under the worst conditions of work” (p. 103). Given that schools with underserved students tend also to have poor working conditions, people often conflate these kinds of schools. By including predictors that signal underserved student populations alongside an indicator for working conditions (stay-ratio), the analyses described here are able to disentangle the effects of these different kinds of schools.
For those who become in-service teachers in schools that are difficult-to-staff or have large underserved student populations, is it better to prepare them in similar kinds of field placement settings? That the stay-ratio effect holds in models with current school fixed-effects suggests otherwise. Such fixed-effects models indicate instead that, among teachers who ended up becoming teacher of record in the same school, including difficult-to-staff and underserved schools, those prepared in field placement schools with higher stay-ratios (easier to staff) were more effective and more likely to persist in NYC schools.
A more direct analysis, though, is to focus only on those teachers who became teacher of record in the hardest-to-staff and most underserved schools to consider the effects of prior training in different kinds of field placement settings. To this end, I reproduced previous retention and value-added analyses by limiting the sample to only those who became teachers of record in schools below the stay-ratio median (hardest-to-staff schools). I then repeated analyses by limiting the sample to only in-service teachers in schools with the highest concentrations of poor, Black, and lowest achieving students, always using the current school median score as the cutoff. 15 For comparison’s sake, I also examined the effects of field placement stay-ratio on those who became teacher of record in the easiest-to-staff schools and in schools with the fewest Black, poor, and lowest achieving students (again using the median as the cutoff).
The left side of Table 9 presents the results from models estimating the likelihood of leaving NYC schools as a function of field placement stay-ratio for teachers working in the hardest-to-staff and most underserved schools. Results are presented as odds ratios, so coefficients less than 1 indicate negative effects. The top row shows the results for those teachers who became teacher of record in the lowest stay-ratio (hardest-to-staff) schools. The effect is negative and statistically significant, suggesting that even for teachers who ended up in the hardest-to-staff schools, those who learned to teach in higher stay-ratio (easier-to-staff) schools were significantly less likely to leave NYC schools. The results were similar for those who became teachers of record in schools with the highest proportion of Black and poor students. For teachers who ended up working in schools with the most lowest achieving students, the direction of the stay-ratio effect was in the same direction, though nonsignificant. The right side of Table 9 demonstrates that there was no significant effect of field placement stay-ratio on retention for those who became teacher of record in the easiest-to-staff schools and schools with typically more privileged populations. Teachers who learned to teach in easier-to-staff field placements were no more or less likely to leave NYC when they ended up working with typically more privileged populations and in easier-to-staff settings.
Estimating the Likelihood of Leaving NYC Schools as a Function of Field Placement Stay-Ratio for Teachers in the Hardest- and Easiest-to-Staff Schools (Estimates as Odds Ratios)
p < 0.1. *p < 0.05. **p < 0.01. ***p < 0.001.
Table 10 presents the estimates for the effect of field placement stay-ratio on teachers’ value-added to student achievement separately for those who became teachers of record in the hardest-to-staff and most underserved schools (left side) and for those who became teachers of record in the easiest-to-staff and least underserved schools (right side). Across the table, point estimates are positive, suggesting that learning to teach in higher stay-ratio (easier-to-staff) field placement schools tended to increase teachers’ effectiveness regardless of where they became teachers of record. This positive effect is statistically significant for teachers who ended up working in the lowest stay-ratio (hardest-to-staff) and the poorest schools. On the other hand, the effect was also statistically significant for teachers who ended up in schools with the fewest Black students.
Estimating Teacher Effectiveness as a Function of Field Placement Stay-Ratio for Teachers Who Work in the Hardest-to-Staff Versus Easiest-to-Staff Schools
p < 0.1. *p < 0.05. **p < 0.01. ***p < 0.001.
Because this study does not use an experimental design with random assignment of teachers to easy-to-staff and difficult-to-staff field placements, it provides evidence for, but cannot establish a direct causal link between, features of pre-service field settings and later in-service teacher outcomes. In the section that follows, I consider some possible alternatives to a causal explanation.
Considering Alternative Explanations for Stay-Ratio Main Effects
This section presents other sorting explanations that could account for the field placement stay-ratio effect on teacher retention and effectiveness. In it, I consider that the field placement stay-ratio effect may actually be a current school stay-ratio effect, an effect of teacher characteristics (including professional preparation), or an effect of “cherry-picking” the most promising teacher candidates. For each, I describe the explanation and why it seems unlikely to account for the stay-ratio results presented here. By ruling out alternative explanations, I intend to build a stronger case for a causal relationship between field placement stay-ratio and teacher retention and effectiveness.
A current school stay-ratio effect?
One possible alternative explanation for the stay-ratio effect is that teachers from easier-to-staff field placements differentially sort into easier-to-staff current schools to become teacher of record. Table 3a suggests that easier-to-staff current settings are more desirable and supportive settings in which to work. It is reasonable then to expect that teachers who sort into such schools will be more effective and less likely to leave. Thus, the field placement stay-ratio effect may simply be a proxy for a current school stay-ratio effect. In other words, if this type of sorting exists, it would be difficult to separate out the stay-ratio effect of pre-service placements from the stay-ratio effect of in-service placements. However, the field placement stay-ratio effect holds even in models that have current school fixed-effects, suggesting that, among in-service colleagues who teach in the same school, those who had field placements in easier-to-staff schools are more effective and persistent. Because school characteristics are identical for all teachers within the same current school, something else likely explains any effect of field placement stay-ratio.
An effect of teacher characteristics or preparation?
Another possible explanation for the effect of field placement stay-ratio is that certain kinds of teachers sort into higher stay-ratio field placements, and these same kinds of teachers are more effective and more likely to remain teaching in NYC schools regardless of where they completed their field experiences. For purposes of explanation, let’s imagine younger teachers tend to sort into higher stay-ratio schools and are also more effective and more likely to persist, regardless of where they had field placements. In this case, the apparent effect of higher stay-ratio field placements might simply reflect the effects of the kinds of teachers (younger) who happen to sort into these kinds of schools. However, the stay-ratio effects on teacher retention and effectiveness persist in models that control for teacher characteristics and preparation, thus providing evidence against this alternative explanation.
Nevertheless, it is possible that unobserved teacher characteristics—like social skills or charisma, for instance—predict the kinds of schools in which teachers do their field experiences as well as their effectiveness and persistence in the profession. It would be impossible to identify and control for all possible teacher characteristics that could explain the stay-ratio effect. Even so, I decided to be as exhaustive as possible given available data: I examined all items from the Teacher Policy Research first-year survey—given to all teachers in this study’s sample—that might signal teacher characteristics or background.
Because a premise of this alternative explanation is that teachers sort by certain characteristics into higher stay-ratio field placements, I began by testing this premise. More specifically, I used linear regression to estimate teachers’ field placement stay-ratio as a function of teacher characteristics—including measures used in prior analyses, as well as 36 items from the Teacher Policy Group first-year survey. Survey items included measures about their undergraduate major, high school coursework, income, prior experiences working in schools, number of children, marital status, and so forth. Although some of these teacher characteristics predicted sorting into higher stay-ratio schools, 16 the critical issue was whether these same items would also predict teacher retention or effectiveness. If so, these teacher characteristics may explain the effects of field placement stay-ratio. To test this, I reproduced both retention and value-added analyses but included controls for all survey items that were associated with sorting into higher stay-ratio schools at statistically significant levels. The stay-ratio estimates remained similar, however, providing further evidence that the stay-ratio effects on teacher retention and effectiveness are not explained by certain kinds of teachers sorting into higher stay-ratio field placements. 17
Another explanation may be that the stay-ratio effect reflects differences between teacher preparation pathways or programs. More specifically, some pathways or programs may prepare more effective and persistent teachers, and these same programs may differentially place teachers in higher stay-ratio field placements. Of particular concern is that alternative route programs, like Teaching Fellows and TFA, are deliberate about placing their teachers in harder-to-staff schools and with underserved student populations. Moreover, these pathways typically require abbreviated field experiences in the summer and therefore are much shorter in duration and likely utilize a different set of schools from other pathways—namely, ones that are year-round or include summer programs. Complicating the analysis of alternative route programs further, summer schools may not reflect the average school characteristics used in this study’s analyses, which are based on administrative data from the academic year.
The regression models presented in this study, however, include preparation program fixed-effects, 18 suggesting that within-program, rather than between-program, differences account for the findings. As compared to candidates prepared in the same preparation program, those who learned to teach in field placement schools with higher stay-ratios were more effective and more likely to persist in NYC schools. Given concerns about the nature of field experiences in alternative routes, I also tried limiting the sample to those who entered teaching through college-recommending (traditional route) pathways and reproduced both retention and value-added models. Even with this reduced sample, results were similar—among college-recommending teachers only, those trained in higher stay-ratio field placements were more likely to persist and were more effective. 19
The above results suggest that the stay-ratio effects are not explained by differences between teacher characteristics or preparation programs. Even though between-program explanations are unlikely, alternative, within-program explanations are important to consider further.
“Cherry-picking” candidates within programs
A final explanation for the stay-ratio effect could be that those who were responsible for assigning field placements within a given teacher education program were “cherry-picking” candidates. More specifically, field placement coordinators were able to anticipate which teacher candidates would end up being more effective and more likely to stay in NYC schools, and then differentially placed these same candidates into higher stay-ratio field placements. In this scenario, the positive effect of field placement stay-ratio would simply reflect sorting by more promising candidates into higher stay-ratio schools rather than demonstrating the positive causal effects that learning to teach in easier-to-staff schools might have on teacher retention and effectiveness.
There are a number of reasons why this explanation seems unlikely to account for the robust and substantial stay-ratio effects. First, the argument assumes that field placement coordinators and others responsible for assigning field placements are capable of forecasting which candidates will persist in NYC schools and become more effective. However, field placements are typically made early on, often before field placement coordinators and other faculty would have had an opportunity to observe prospective teachers interact with K–12 students. In such cases, teacher educators would need to depend on observable characteristics, like teacher candidate records and background information, to make these kinds of projections. As described in the last section, however, the stay-ratio effects persisted even after controlling for an extensive number of observable teacher characteristics, thus casting doubt on this “cherry-picking” explanation.
Even assuming program faculty are able to forecast which teachers would persist and become more effective as in-service teachers, the “cherry-picking” explanation is still problematic. Namely, it assumes that program faculty would place the candidates with the most potential in the easiest-to-staff placements. There is a good rationale for why this could occur: Field placement coordinators might place more promising candidates in higher stay-ratio settings to ensure the site will be amenable to future placements. Schools that received less promising candidates, for example, might reject future requests to take on student teachers. However, there is an equally strong rationale for placing candidates with the least potential in the easiest-to-staff placements. Assuming higher stay-ratio field placement schools are indeed better environments for learning to teach, field placement coordinators might assign less promising candidates to such settings to ensure that those candidates who need help the most will receive the best training.
Discussion
The purpose of this study was to examine the effects of learning to teach in difficult-to-staff and underserved schools on teacher retention and effectiveness. It found that teachers who learned to teach in higher stay-ratio (easier-to-staff) field placement schools were more effective at raising test scores and more likely to stay in NYC schools during their first 5 years of teaching. Moreover, learning to teach in easier-to-staff schools was associated with better retention and achievement gains even for teachers who became teachers of record in the hardest-to-staff (lowest stay-ratio) schools and for those who ended up working with the most underserved student populations.
Though studies more typically examine the effects of school settings on K–12 student learning, this study suggests that certain kinds of schools may have positive effects on student teacher learning. Because alternative sorting explanations considered above are unlikely to account for the robust effects of stay-ratio, the findings indicate that prospective teachers are learning something from easier-to-staff schools that helps them become more effective and better able to persist as teachers of record in NYC schools. Although not technically “professional development schools,” easier-to-staff schools may serve a similar function in being sites for professional learning.
That teachers prepared in easier-to-staff settings were more effective at raising student test scores as they became full-time teachers, regardless of where they ended up working, suggests they may have learned something about instructional practice that could translate into positive outcomes for various kinds of students and circumstances. Although school administration, students, parents, and even facilities likely play important roles in creating successful learning environments, prospective teachers are probably learning most about effective instructional practice from other teachers. This is consistent with Jackson and Elias (2009), who find evidence that about one fifth of a teacher’s effectiveness, as measured by student achievement gains, can be explained by the effectiveness of prior teaching colleagues to whom they had been exposed and from whom they could have learned professionally. This is consistent also with the fact that high stay-ratio schools are places where teachers want to come and stay year after year. Teachers will gravitate toward educational communities where they will grow and thrive professionally, and where they feel like students are receiving the learning they deserve. It should not be surprising that these same settings are where new teachers may thrive as well.
Understanding what and how teachers may be learning from easier-to-staff field placements are critical issues beyond the scope of this study. Even so, schools with higher stay-ratios possess many of the features associated with well-functioning schools where quality teaching and learning thrive, including higher quality administration and support, better staff relations and collegiality, and a more experienced faculty. As suggested by Johnson and Birkeland (2003), better working conditions may promote in teachers a “sense of success” and a commitment to remain in schools.
The findings of this study warn against conflating the effects of learning to teach in hard-to-staff (low stay-ratio) schools with schools that represent underserved student populations. This study employs a number of different measures intending to signal hard-to-staff and underserved schools, and the results vary according to which measure is used. Because learning to teach in harder-to-staff schools is associated with worse teacher retention and effectiveness, one might expect learning to teach in schools with more poor, Black, and lowest achieving students to have similar effects, as schools with more underserved student populations typically are harder to staff. However, this is not the case, as the proportion of these underserved populations in field settings is not significantly related to teacher retention or effectiveness. If anything, this study provides weak evidence that learning to teach in schools with more low-achieving and Black students—controlling for other school characteristics—is associated with better, rather than worse, teacher effectiveness.
The effect of stay-ratio is positively associated with teacher retention and effectiveness even after controlling for the effects of poor, minority, lowest achieving student populations, as well as many other school characteristics. Even though easier-to-staff schools tend to be schools with fewer poor, minority, and lowest achieving students, something beyond having typically more privileged student populations appears to make them better environments for learning to teach. This is consistent with the explanation that high stay-ratio schools may offer better working conditions and teaching faculty who are better equipped to mentor prospective teachers.
Implications and Future Research
What are the implications of these results for the debate described in the introduction of this study? Should we place student teachers in difficult-to-staff and underserved schools to learn to teach? The main findings suggest teacher education programs should avoid placing prospective teachers in difficult-to-staff schools. Learning to teach in difficult-to-staff field settings was associated with lower teacher effectiveness and retention. This is not to suggest that programs seek out schools with fewer poor, minority, and low-achieving students. As described in the last section, holding stay-ratio constant, the proportions of underserved student populations at field sites have no significant effects on teacher retention or effectiveness. If anything, this study provides weak evidence that higher proportions of lowest achieving students may even have positive effects. Moreover, it also provides weak evidence that larger populations of Black students may be associated with better teacher effectiveness, provided that field sites are also easy to staff. In other words, training teachers in settings with more underserved student populations may improve teacher effectiveness given that these settings also offer desirable conditions for professional practice.
Were these findings true also for those who ended up becoming in-service teachers in the most difficult-to-staff schools and with the most underserved student populations? As described in the introduction, some have suggested that to succeed in difficult-to-staff, underserved schools as teacher of record, it may be better to learn to teach in similar kinds of field placement schools. For example, Haberman (1995) argues that we may actually be doing new teachers a disservice by preparing them in easier-to-staff settings, as this will leave them unprepared for difficult-to-staff schools; instead, we should prepare teachers in “the worst schools and under the poorest conditions of practice” (p. 778). Contrary to this position, this study finds that learning to teach in field placements under better, rather than poorer, conditions of practice has positive outcomes for teachers who become teachers of record in the hardest-to-staff and most underserved schools. In fact, teacher training in easier-to-staff field settings had the strongest benefit on retention and achievement gains for those teachers who ended up working in the hardest-to-staff schools.
Although this study’s results are suggestive of a causal relationship between field placement stay-ratio and teacher retention and effectiveness, the evidence is by no means definitive. More studies are needed to reproduce these findings. Given the very unique student demographics and teacher labor market of NYC, these should include studies in different kinds of districts and states to see if the effects of field placement stay-ratio still hold. Moreover, well-designed experimental studies with random assignment of teachers to easy-to-staff and difficult-to-staff field placements would be useful in bolstering, or countering, the case for a causal relationship presented here.
Because it can be a challenge for teacher preparation programs to find enough field placements in NYC schools, it may be unrealistic to expect programs to find placements at easier-to-staff schools. One reason for the short supply of field placement schools is the potential cost—schools are often hesitant to allow less experienced teachers to take over classroom responsibilities from more experienced professionals. Even so, the field placement stay-ratio effects on teacher retention and effectiveness appear to be separately quite large in magnitude, so their combined effect has strong potential for improving teacher quality in NYC schools. Although there may be short-term losses to teacher quality in some classrooms that take on student teachers, there are tremendous long-term potential gains. As such, districts may consider ways to offset short-term costs to schools that take on student teachers in order to achieve these longer term benefits.
Although this study suggests a relationship between field placement stay-ratio and teacher retention and effectiveness, it does not explain the nature of this relationship. This begs the question, What about higher stay-ratio (easier-to-staff) schools make them positive sites for teacher learning and development? Might it be the case, for example, that easier-to-staff schools attract and keep higher quality mentor teachers who can model or share best practices and principles that help prospective teachers become more effective and persistent wherever they go? Do these kinds of schools provide new teachers more supportive or complimentary settings for trying out the practices they are learning in their university coursework? Is it that these schools simply have better administrative support and leadership that provides a learning environment where quality instructional practice can thrive and get passed on to aspiring teachers? Future studies should examine easier-to-staff schools to uncover the mechanisms by which they support teacher learning. What is learned from studies might guide teacher educators in assigning better field placements and also might inform efforts to design PDS.
It is important to acknowledge that the results presented here are average effects based on school-level measures. In other words, this study has provided a blunt signal for identifying quality placements, and more research is needed to understand the specific features that give rise to these average effects. It may be, for example, that the school stay-ratio is signaling better average mentorship by senior faculty. This would not imply, however, that only high stay-ratio schools have worthy mentors or make for quality placements. There may be quality mentors in schools that have low average scores that would still provide ideal field placement settings. Although stay-ratio may be a useful average proxy for selecting placements, identifying the specific mechanisms by which high stay-ratio schools support teacher retention and effectiveness can have more direct implications for how we prepare teachers.
This study introduces a program of research with promising policy implications. Given that school stay-ratio is fairly straightforward to calculate from administrative data that districts routinely collect, this may be a low-cost, average indicator for identifying high-quality field placements. Assuming the effects and their magnitudes hold up in future studies like the ones outlined above, this study’s results indicate that teacher education programs should consider organizing field placements at easier-to-staff schools a priority. Doing so may help in responding to growing calls from researchers and policymakers alike to staff K–12 schools with effective teachers and to keep them there.