Supporting Early Career Alternatively Certified Teachers: Evidence from the Beginning Teacher Longitudinal Survey

by Christopher Redding & Thomas M. Smith - 2019

Background: Alternatively certified (AC) teachers have generally been found to turn over at higher rates than traditionally certified (TC) teachers. These higher turnover rates are generally attributed to lower levels of preparedness and less of a commitment to remain in teaching than TC teachers, both of which may be compounded by AC teachers’ increased likelihood of beginning their career in schools that enroll traditionally underserved students.

Purpose: Our goal is to better understand the early career professional learning opportunities of AC teachers. We consider the ways in which in-service organizational supports such as mentoring, collaboration with one’s peers, and professional development compensate for alternative certification teachers’ reduced levels of pre-service training.

Research Design: We conduct a secondary analysis of data from the Beginning Teacher Longitudinal Survey (BTLS). BTLS is a nationally representative survey of the cohort of new teachers who began their career in the 2007–2008 school year. Teachers were surveyed annually for their first five years in the teaching profession. We draw on data regarding teachers’ entry pathway, feelings of preparedness, organizational supports, and turnover (i.e., leaving teaching or moving schools).


Data Analysis: We first describe differences in self-reported preparedness, commitment to remain in teaching, and use of in-service organizational supports across beginning teachers across different entry pathways. We then conduct discrete time survival analysis to (1) understand differences in the timing of turnover rates across entry pathways and (2) examine the role induction supports play in improving AC teacher retention.


Findings: We show that AC teachers enter teaching feeling less prepared and, with the exception of mentoring, receive no additional support in their first year of teaching compared to TC teachers. Although we observe a 10-percentage point gap in the turnover between early career AC and TC teachers is, this gap is, in large part, explained by observable teacher and school characteristics. We show some evidence that AC teachers differentially benefit from extra classroom assistance, quality of mentor feedback, and content professional development, which were all associated with lower odds of leaving teaching.


Conclusions: Our findings suggest that schools and districts could do more to target induction supports for novice AC teachers. Given that AC teachers receive most of their training once they begin teaching, alternative certification programs, schools, and districts can customize supports for AC teachers to fit the needs of their local context.

INTRODUCTION

Alternatively certified (AC) teachers have generally been found to turn over (i.e., leave the profession or change schools) at higher rates than traditionally certified (TC) teachers, with even higher turnover rates for one of the most selective programs, Teach For America (Kane, Rockoff, Staiger, 2008; Redding & Smith, 2016). Researchers contend that the higher turnover rates of alternatively certified teachers are linked to less of a commitment to remain in teaching, the lack of pre-service training, and lower levels of preparedness (Darling-Hammond, Chung, & Frelow, 2002; Kee, 2012). Alongside these factors, in-service induction supports must also be considered as a factor influencing the retention of AC teachers, as most of these teachers receive the bulk of their training once they have entered the classroom. This study aims to fill this gap in the literature by examining the early career experiences of AC teachers, including the induction supports they receive that may reduce the likelihood of turnover.

As AC teachers receive less comprehensive pre-service training, it is assumed they receive more on-the-job support (Brantlinger & Smith, 2013; Consuegra, Engels, & Struyven, 2014; Duke, Karson, & Wheeler, 2006). In-service induction supports, such as mentoring, targeted professional development (PD), or extra classroom assistance, may be particularly valuable for supporting the transition of AC teachers into the profession. It could be argued that receipt of effective in-service supports are even more critical for novice AC teachers than for their counterparts who entered through traditional teacher training programs. In this paper, we test this assumption with a nationally representative sample of novice teachers by comparing differences in beginning AC and TC teachers participation in in-service induction supports and the extent to which AC supports differentially reduce turnover among AC teachers during their first four years teaching.

Although there is a trend towards more classroom-based learning as part of teacher training (e.g., more hours of student teaching) (NCATE, 2010) critics contend that, although teacher training should be grounded in practice (Ball & Forzani, 2009), early experience in the classroom often result in procedural, teacher-centered instruction rather than more ambitious instructional practice (Good & Brophy, 2000). In a mixed-methods study of alternative certification programs (ACPs) in Belgium, Consuegra et al. (2014) found that most school-based learning opportunities were incidental and not structured in a way that promoted complex learning by teachers (Feiman-Nemser, 2001). Furthermore, compared to TC teachers, early career AC teachers are more likely to fill vacancies in high-needs schools, particularly those with greater concentrations of racial/ethnic minorities and low-income students. Such schools often have fewer supports needed for AC teachers professional development, including strong leadership, access to resources, and staff collegiality (Johnson, Kardos, Kauffman, Liu, & Donaldson, 2004). Thus, the challenges facing novice AC teachers are often compounded, with less pre-service training combining with more challenging working environments. Yet, the schools in which these teachers begin their careers often lack the organizational resources with which to induct these new teachers into the school and the profession.

Better understanding the early career learning opportunities and experiences of AC teachers is important given differences in novice AC and TC teachers students performance. In North Carolina, AC teachers have been found to have a small negative impact on student performance compared to TC teachers, particularly at the secondary level (Clotfelter, Ladd, & Vigdor, 2010; Henry et al., 2014). Research from New York City has shown AC teachers to be less effective in the first years of their career, although these differences fade with experience (Kane et al., 2008). One exception to this pattern is that teachers from the highly selective Teach for America (TFA) program tend to outperform their traditionally certified peers, particularly in mathematics (Glazerman, Mayer, & Decker, 2006; Henry et al., 2014; Xu, Hannaway, & Taylor, 2009). This body of research also generally finds higher turnover rates among AC than TC teachers, although this research tends to be limited to a particular geographic area (e.g. Kane et al., 2008) or use cross-sectional data that fail to distinguish the timing of early career teacher turnover (Redding & Smith, 2016). Using the nationally representative Beginning Teacher Longitudinal Survey (BTLS), we follow a cohort of early career AC and TC teachers through the first four years in the profession to examine the supports they receive and their turnover patterns.

Given our goal of better understanding the early career learning opportunities and experience of AC teachers, we first describe differences in the induction supports received by new teachers across different entry pathways. We then conduct discrete time survival analysis using the BTLS to understand differences in turnover rates across entry pathways and examine the role these supports may play in improving AC teacher retention. This methodological approach allows us to focus not only on whether a teacher had turned over in a given year, as is the case with other studies on AC teacher turnover, but on the timing of that turnover as well (Singer & Willett, 2003).

In addition to examining whether there are differential turnover rates between early career AC teachers, We then examine the in-service induction supports that are associated with a lower risk of turnover among early career AC teachers. To do so we ask the following questions:

(1) What are the differences in self-reported preparedness, commitment, and use of in-service induction supports among early career AC and TC teachers?

(2) To what extent are there differences in the timing of turnover (moving schools or leaving the profession) between early career AC and TC teachers during their first four years of teaching?

(3) To what extent is participation in in-service induction supports differentially associated with a reduction in the likelihood of leaving teaching for early career AC teachers compared to TC teachers?

(4) Among those who left teaching, what are the differences in occupational status between early career AC and TC teachers?

LITERATURE REVIEW

There is great diversity in alternative certification programs (Humphrey & Wechsler, 2007). They can be set up by districts, supported by schools of education in colleges and universities, and administered by state and national non-profit or proprietary agencies. The diversity of program characteristics can also be attributed to the complex state policy contexts and varying professional norms regarding what teachers are expected to know. Yet, in New York City, Boyd et al. (2008) suggest that this diversity in pathways overshadows the similarity of the curriculum across programs. For instance, they find that teachers in both AC and TC programs are often trained in the same educational institutions and take many of the same courses. The differences come from the emphasis and timing of those courses, with ACPs addressing the pragmatic or technical aspects of teaching more than theory. For instance, an AC teacher may focus on classroom management more than learning theory or child and adolescent development. The abbreviated nature of the pre-service curriculum only allows for briefif anystudent teaching during summer school and requires continued coursework while the AC candidate is serving as teacher of record (Constantine et al., 2009; Humphrey & Wechsler, 2007). The literature suggests that entering the classroom with less practice teaching and exposure to teaching methods may increase the risk of turnover (Darling-Hammond, Chung, & Frelow, 2002; Kee, 2012). Yet, the literature has not adequately described the extent to which in-service induction supports received by AC teachers may compensate for the lack of preparedness and improve retention.

With similar design features across ACPs, one key differentiating feature is the types of prospective teachers they recruit. National programs such as TFA or The New Teacher Projects (TNTP) Teaching Fellows program and state programs such as Teach Kentucky make part of their mission the recruitment of teachers from selective universities. Although these AC programs receive the bulk of the media attention, previous research indicates that the majority of AC teachers attended less selective universities (Redding & Smith, 2016). The remainder likely attend less selective programs, such as those that aim to recruit recent college graduates without an education background or midcareer professionals transitioning into teaching (Humphrey & Wechsler, 2007).

Previous research has generally found higher turnover rates among AC teachers, although differences exist between programs. Teachers from more selective programs have been found to leave teaching at higher rates (Boyd et al., 2008; Kane et al., 2008). In New York City, Kane et al. (2008) found much higher attrition rates among TFA teachers with only 18 percent teaching in the district after four years, compared to slightly over 50 percent of traditionally certified teachers. The survival rates for TFA teachers mirror TC teachers for their first year but drop to approximately 40 percent following their second year when their two-year commitment has ended. By the end of their third year, only 25 percent remain in teaching. It is unclear the extent to which this finding generalizes to other programs that recruit teachers from selective universities and without a two-year commitment.

The evidence is mixed regarding the turnover rates of AC teachers entering teaching through less selective programs. In New York City, Kane et al. (2008) find no differences in turnover between traditionally certified teachers and participants in the New York City Teaching Fellows (NYCTF). In North Carolina, Bastian and Henry (2014) report higher turnover rates among alternate entry teachers compared to any other entry pathway. In a study using the Schools and Staffing Survey (SASS), we find evidence of a growing turnover gap between AC and TC teachers. In 20072008, 25 percent of early career AC teachers turned over annually, compared to only 17 percent of TC teachers (Redding & Smith, 2016).

INDUCTION SUPPORTS AND NOVICE TEACHER TURNOVER

With evidence of higher turnover among AC teachers, in-service induction supports such as mentoring, collaboration with ones peers, and professional development may all help in developing teachers instructional practices and also integrate them into their school. Previous research links in-service induction supports such as mentoring and collaboration with colleagues to lower turnover levels (Ingersoll & Strong, 2011). In the study most similar to the present one, DeAngelis, Wall, & Che (2013) examine evidence of the extent to which novice teachers intention to remain in their current school is related to the interaction between pre-service program quality and induction supports. Among novice teachers trained in traditional credentialing programs across 12 public education schools in one state, the researchers find that novice teachers intend to move schools at lower rates when more satisfied with their pre-service program and assigned to a mentor they perceive to be more helpful. The results do not hold when they examine teachers actual departure from their school. We build on their analysis by drawing on a nationally representative cohort of new teachers and examining turnover differences across entry pathway. For AC teachers, lacking substantial pre-service training may elevate the need for and provision of in-service induction supports such as mentoring. In-service organizational supports may also compensate for novice AC teachers lower levels of preparedness, differentially benefitting these teachers and resulting in higher retention rates.

Given that AC teachers receive a significant amount of training once they begin teaching, supporting AC teachers is not only the responsibility of their training program but the school as well. On the one hand, this arrangement provides schools with a unique opportunity as supports can be customized to the local context. Teacher residency programs provide a unique example of a way in which teacher training and retention becomes the shared responsibility of the preparation program and district (Papay, West, Fullerton, & Kane, 2012; Silva, McKie, Knechtel, Gleason, & Makowsky, 2014). This approach, whereby aspiring teachers complete coursework and supervised fieldwork in a school over the course of a year or two, has been linked to higher retention rates within the district (Papay et al., 2012). These results suggest that the training of alternatively certified teachers could be structured in a way that is sensitive to the local district and even school context.

Youngs et al. (2012) argue for the need to acknowledge the social organization of schools when considering induction programs, as most induction supports come from peer mentors, grade- or subject-level teams, department chairs, and administrators. These induction activities aid in instructional improvement but also signal school stakeholders expectations for new teachers. When new teachers beliefs about effective instruction align with the expectations of others in their school they are more likely to experience higher levels of commitment to teaching and retention. For alternatively certified teachers, the reduced levels of pre-service training may even enhance their receptivity to these localized supports.

On the other hand, there are reasons why induction supports may not help reduce AC teacher turnover. First, as AC teachers are more likely to work in hard-to-staff schools with difficult working conditions, in-service supports offered by their school may not be adequately address the increased challenges faced by these new teachers. Second, AC teachers who have entered teaching with little pre-service preparation may have unique developmental needs as teachers. Given the critique that professional development often fails to be differentiated (Hill, Rowan, & Ball, 2005), beginning AC teachers may not receive the types of in-service supports they require. Third, the provision supports for new teachers may have little relationship with their retention, unless the supports are of high quality (Youngs, 2007). Finally, regardless of the supports, it is possible that AC teachers bring less of a commitment to remain in teaching that is unresponsive to induction supports.

DATA AND METHODS

DATA

The data for this paper come from the BTLS. This study began in 20072008 with the cohort of beginning public school teachers who participated in the SASS. BTLS followed a cohort of 1,990 teachers annually for their first five years of teaching (Gray, Goldring, & Taie, 2015). In the first year, survey response rates varied between principals and teachers but were consistently above 80 percent. Response rates were higher in subsequent school yearsabove 84 percentuntil they dipped to 77 percent in the final year of the study. To reduce nonresponse bias and compensate for schools selection probability, this study uses longitudinal survey weights to conduct analysis on a sample that most closely resembles the target population of new public-school teachers in 20072008.

For this analysis, we draw on the annual BTLS surveys as well as data on school working conditions from the SASS conducted during teachers first year of teaching. Although the BTLS surveys provide a rich understanding of these teachers employment status, continuing education, and certification status, these surveys do not ask teachers to provide the same level of detail of the conditions within their school as SASS. Merging the 20072008 SASS allows us to account for the perceptions of school working conditions among all teachers in the SASS sample. When these data are merged, our analytic sample includes 1,570 teachers in the base year. Initial missing data is largely a result of several variables from the SASS school survey, of which 7% and 9% of values are missing.

MEASURES

In this section, we describe the variables used in the descriptive and survival analysis. The dependent variable comes from teacher reports of their employment status starting their year after participation in the base year SASS sample. Each questionnaire asked teachers about their current occupation status. From several response options, we collapse teacher employment status into one of three categories: stayers, movers, and leavers. In some models, we also measure turnover as including teachers who have ever moved schools or left teaching. After teachers first year, BTLS followed up with teachers who had left and returned to teaching after the 20082009 school year. Given the definition of turnover as ever moving or leaving the school, we ignore these returners in this analysis.¹

Our key independent variable is teachers entry pathway into teaching. Entry pathway is measured by teachers answer to the question Did you enter teaching through an alternative certification program?² We also focus on a variety of in-service supports available to new teachers. Induction supports includes participation in an induction program, assignment of a mentor, a reduced teaching schedule, common planning time with other teachers in the same subject or grade level, beginning teacher seminars, extra classroom assistance, regular communication with an administrator or department chair, and ongoing feedback from a master or mentor teacher. With the exception of mentoring, all variables come from the base year SASS survey and are coded as dichotomous variables, indicating whether or not the teacher participated in each support in their first year. Receipt of mentoring is the only support measured in each survey year and is included in all analysis as a time-varying predictor.³ Participation in professional development is measured by teachers responses to a series of questions about whether or not they have participated in professional development for a variety of topics in the past year. These topics include content related to the subject they teach, the use of computers for instruction, reading instruction, classroom management, teaching students with disabilities, and teaching limited-English proficient (LEP) students. Similar to the induction supports, these variables are included in the analysis as dichotomous variables. In addition, we also create a measure assessing the extent to which professional development matched teachers needs. Teachers report on their top priorities for their own professional development. We create an indicator variable if teachers report needing a particular type of PD and they actually received it.

We also consider self-reports of teachers preparedness and commitment to teaching. Preparedness is a scale measure of six questions of teachers rating of how well prepared they were in their first year. The tasks include classroom management, using a variety of instructional methods, teaching their subject matter, using computers in classroom instruction, assessing students, and selecting and adapting curriculum materials. This measure was generated and standardized for the sample of first-year teachers (α = 0.81).

We also examine descriptive evidence of differences in the commitment and career satisfaction at the end of teachers first year. Teacher commitment is measured by the response to the question: How long do you plan to remain in teaching? We examine five response options: as long as I am available, until I am eligible for retirement benefits or Social Security, until a specific life event occurs, until a more desirable job opportunity comes along, and undecided. For career satisfaction, we focus on the question If you could go back to your college days and start over again, would you become a teacher or not?, answer on a five-item Likert scale, which we reverse-code to indicate teachers greater certainty of becoming a teacher again.

In the survival analysis, we include a number of teacher, classroom, and school variables. Teacher characteristics include, gender, race/ethnicity, age, marital status, their salary (logged), and indicators of whether they teach an in-demand subject, their education level, whether they teach part-time, and union membership. In-demand subjects include those typically found to have the most frequent turnover, including mathematics, science, special education, or English as second language (ESL) teachers. School context is measured by the proportion of racial/ethnic minorities and students on free or reduced price lunch (FRPL), the schools average absentee rate and dichotomous variables for urbanicity (city, town, and rural), whether or not the school is a charter school, or a secondary school, and the principal report of whether or not the school made AYP in the previous school year.

Six variables are used to characterize working conditions that have been used in previous research: class size and hours teaching per week, principal reports of student behavior, and teacher reports of principal effectiveness, staff collegiality, and faculty influence (Ingersoll, 2001). Student conflict is measured by principal reports of the frequency of the following behaviors: physical conflicts among students, robbery or theft, vandalism, student use of alcohol, student use of illegal drugs, student possession of weapons, physical abuse of teachers, student racial tensions, student bullying, student verbal abuse of teachers, widespread disorder in classrooms, student acts of disrespect for teachers, and gang activities. Each item was reverse coded to have an increased frequency of problem behaviors corresponding with a higher value in the student behavior problems scale and standardized (α = 0.84). The measure of principal effectiveness uses teacher reports of their principal’s school leadership. These questions are related to principal support for school staff, enforcement of school rules, principal communication to the staff, recognizing staff for accomplishments, and teacher satisfaction with how the principal manages the school (α = 0.88). Like the other scale measures used in this analysis, this measure was reverse coded, averaged at the school-level for all teachers sampled in the SASS and standardized across the teachers in the analytic sample. Two questions make up the collegiality measure: Most of my colleagues share my beliefs and values about what the central mission of the school should be and There is a great deal of cooperative effort among the staff members (α = 0.70). Faculty influence is measured by teacher influence over selecting textbooks, selecting content, selecting teaching techniques, evaluating students, disciplining students, and determining the amount of homework assigned (α = 0.73).

METHODS

We begin our analysis with descriptive statistics on the differences between teachers who entered teaching through a traditional or alternative pathway. To better understand the early classroom experiences of these teachers, we report on their feelings of preparedness, commitment to remain in teaching, and in-service supports.

We then conduct discrete-time survival analysis using logistic regression (Singer & Willett, 2003). We estimate the following model using logistic regression:

 (1)

This model estimates the odds that teacher i turned over in year j. The  to  parameters include time indicators for teachers second to fourth years of teaching ( to ). To allow the effect of entry pathways to differ in each year, we interact entry pathway () with the time period indicators. Without this series of interactions, the more basic model would assume a constant hazard rate, that is, that teachers have an equal risk of turning over in a given school year. Given the two- to three-year commitment in many alternative certification programs, this fully interactive specification in Model 1 is the more appropriate modeling strategy. Below, we tested for, and found evidence of, better model fit with the fully interactive model. To this model, we add time-invariant controls for a vector of teacher and school characteristics (). To account for the correlated error term associated with repeated observations of the same teacher, all models cluster robust standard errors at the teacher level. All models include longitudinal survey weights. We estimate this model separately for overall turnover, moving schools, and leaving teaching altogether.

This discrete time survival analysis allows us to understand differences in the timing of turnover for early career AC and TC teachers. To answer our next research question of the relationship between participation in in-service induction supports and a decreased likelihood of turnover, we estimate a series of logistic regression models that examine the interactive effect these new teacher supports have on AC teachers. As the in-service induction supports are only measured in teachers first year, Model 2 predicts whether a teacher ever left teaching during their first four years. This model can be estimated as:

		(2)

Model 2 includes the in-service organizational supports () for new teachers. These supports include time-invariant supports teachers received in their first year. In this model, standard errors are clustered at the school level.

A limitation of this analysis that should be noted is that there is likely unobserved heterogeneity in teachers employment decisions. For example, there are likely to be unobservable differences in the working conditions between the schools in which AC and TC teachers are more likely to teach that could relate be related to turnover. This could bias the relationship between entry pathway and turnover, in that the pathway is getting credit for increased turnover when the cause is really the school placement. One possible solution to this problem is the inclusion of school fixed effects, which would allow us to account for unobserved school characteristics that are likely associated with teachers turnover decisions. Given that few novice teachers are sampled in the same school from different entry pathway, this estimation approach is unpractical for the current study. That said, we believe the rich set of school control variables available in the BTLS allow us to account for much of the variation in school characteristics and working conditions that prior literature suggests is related to turnover.

RESULTS

To answer our first research question regarding self-reports of preparedness and commitment to remain in teaching, Table 1 presents conditional means comparing AC and TC teachers during their first year of teaching. In line with previous research, AC teachers in 20072008 entered the profession feeling much less prepared than TC teachers. We find a 0.58 standard deviation difference in the level of preparedness reported by first-year AC teachers compared to TC teachers (p < 0.001). Across the two measures of commitment to remain in teaching, we find mixed evidence. In terms of how long they plan to remain in teaching, we find no practically or statistically significant differences between novice AC and TC teacher. AC teachers are less likely than TC teachers to report that they would become a teacher again if they were to start their career over. On a five-point scale, the average teacher response was 4.26 (σ = 1.00), indicating that teachers typically reported that they would become a teacher again. Responses varied by teachers’ entry pathway, with an average for TC teachers of 4.40 compared to 4.10 for AC teachers. Beginning AC teachers felt less prepared when they entered teaching and slightly less committed to the profession but they did not leave teaching at higher rates during their first four years of teaching. AC teachers, did, however, move between schools at higher rates. 39% of early career AC teachers moved schools compared to 28% of TC teachers.

Table 1. Conditional Means for Traditionally and Alternatively Certified Teachers, First Year

	Traditional Certification	Alternative Certification
Ever left teaching	0.17	0.18
Ever moved schools	0.28	0.39*
Earned additional degree	0.22	0.31+
Ever finished alternative certification program		0.91
Preparedness	0.14	-0.44***
Remain in teaching?
As long as able	0.64	0.63
Until benefits	0.11	0.09
Until life event	0.07	0.05
Until more desirable job	0.02	0.03
Leave as soon as can	0.01	0.01
Undecided	0.14	0.19
Would become teacher again	4.40	4.10**
School Characteristics
Fraction FRPL students	0.45	0.54***
Fraction minority students	0.45	0.62***
Organizational Supports
Mentor	0.83	0.90*
Participated in induction program	0.81	0.81
1st year support
Reduced schedule	0.13	0.21+
Shared planning time	0.63	0.67
New teacher class	0.81	0.79
Extra assistance	0.31	0.26
Regular communication	0.88	0.84
Mentor feedback	0.83	0.90*
Participated in content PD	0.80	0.76
Participated in technology PD	0.50	0.50
Participated in reading PD	0.53	0.49
Participated in classroom management PD	0.56	0.51
Participated in PD for students with disabilities	0.32	0.49**
Participated in PD for LEP students	0.22	0.30
PD matched needs	0.74	0.76
Principal effectiveness	0.13	0.06
Staff collegiality	0.09	0.05
Observations	1180	390
Notes. Estimates adjusted using BTLS survey weights. Statistical significance based on a t test, adjusting for clustering at the school level with robust standard errors. + p < .10 * p < 0.05 ** p < 0.01 *** p < 0.001. Sample sizes rounded in accordance with NCES nondisclosure rules.

Table 1 also presents descriptive evidence of differences in the organizational supports allocated to first-year AC and TC teachers. The design of alternative certification policies shifts the bulk of a teachers training to when they are serving as teacher of record in the classroom. As a result, we expect evidence of greater use of various organizational supports, particularly since first-year AC teachers feel much less prepared to enter the classroom than teachers who received more extensive pre-service training. Outside of higher assignment to a mentor (which could be someone with the school or district, as well as someone from the AC preparation program), we find little consistent evidence of additional organizational supports being accessed by AC teachers. In their first year, 90% of AC teachers report being assigned to a mentor compared to 83% of TC teachers. In results not reported, we find that in the subsequent year, AC teachers tend to be assigned to a mentor at slightly higher rates, although the differences narrow over time and are never significant after the first year.

Among the other organizational supports, we find few differences in the professional development opportunities available to AC and TC teachers. The one exception to this pattern is that AC report receiving more PD to work with students with disabilities than TC teachers (49% versus 32%; p < 0.01). This difference is likely driven by the increased share of AC teachers working as special education teachers (18% versus 10%, p = 0.07). Finally, as the supports new teachers receive may also come from the social conditions of the school itself, we examined teacher-reports of principal effectiveness and staff collegiality. We found no significant differences across TC and AC teachers for either of these measures.

Although we find no evidence to support the hypothesis of more organizational supports for novice alternatively certified teachers, it may be that the types of training they receive is not captured in questions pertaining to school-based supports. For instance, their training may be situated at a local university (e.g., while they work on a masters degree) rather than their school. We find some evidence for this proposition. Compared to the 22 percent of TC teachers who earned an additional degree during their first five years of teaching, 31 percent of AC teachers earned a degree. With little evidence of differential receipt of supports by early career AC teachers, they still may differentially benefit from the supports they do receive because of the reduced level of preparedness.

PREDICTING TURNOVER AMONG EARLY CAREER ALTERNATIVELY CERTIFIED TEACHERS

To better understand the timing of moving schools or leaving teaching for early career AC and TC teachers, our second research question is answered using discrete time survival analysis. We begin by presenting the unadjusted hazard and survival rates obtained from Model 1. Results shown in Table 2 point to important differences in the turnover rates of these two groups of teachers over time. AC and TC teachers have a similar risk of moving schools during their first three years. Conditional on remaining in their first school, AC teachers have twice the risk of moving schools following their fourth year compared to TC teachers. In terms of leaving, TC teachers are at greater risk of leaving teaching following their first year while. Conditional on not having turned over, AC teachers are at much greater risk of leaving following their third year. In terms of survival probability, after four years, 38% of TC teachers remain in the same school compared to only 28%. This difference is driven almost entirely by the higher rates with which AC teachers move schools. Half of AC teachers have moved schools by the end of their fourth year compared to 40% of TC teachers.

Table 2. Unadjusted Hazard and Survival Rates for Traditionally and Alternatively Certified Teachers
Unadjusted Hazard Rates						Unadjusted Survival Rates
	20072008	20082009	20092010	20102011		20072008	20082009	20092010	20102011	20122013
Traditionally Certified Teachers
Turnover	0.26	0.13	0.12	0.11		1	0.74	0.61	0.49	0.38
Mover	0.15	0.10	0.08	0.07		1	0.85	0.75	0.67	0.60
Leaver	0.10	0.04	0.04	0.04		1	0.90	0.86	0.82	0.78
Alternatively Certified Teachers
Turnover	0.20	0.14	0.20	0.18		1	0.8	0.66	0.46	0.28
Mover	0.15	0.12	0.09	0.14		1	0.85	0.73	0.64	0.50
Leaver	0.05	0.03	0.11	0.05		1	0.95	0.93	0.82	0.77

Importantly, the differences in the hazard rate between AC and TC teachers may be explained by differences between AC and TC teachers or the schools in which they begin their career. To account for such factors, in Table 3, we present the results from the discrete time hazard model, controlling for teacher and school factors. For each outcome, we specify a simplified model before adding interactions between survey period and entry pathway. We then add controls for teacher and school characteristics. In the simplified model in Column 1, we find no overall evidence to support the hypothesis that AC teachers are more at risk of turning over during their first four years of teaching than TC teachers.⁴ When we allow the relationship between certification and turnover to vary over time, we still find no evidence of differences in the risk of turnover between AC and TC teachers, although the interaction between AC and the 20092010 wave is nearing statistical significance at conventional levels (p = 0.14) and we have better model fit (D₀ D₁  = 920, df = 6).

Table 3. Discrete Time Hazard Models of the Odds that a Teacher Will Turn Over, Move Schools, or Leave Teaching

	Turnover vs. Stay				Move vs. Stay				Leave vs. Stay
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Alternatively certified (AC)	1.24	0.89	0.90	0.73	1.31	1.07	1.16	0.94	1.05	0.62	0.56	0.54
	(1.12)	(-0.38)	(-0.32)	(-0.94)	(1.12)	(0.18)	(0.37)	(-0.16)	(0.16)	(-1.17)	(-1.37)	(-1.46)
2008-2009	0.52**	0.46***	0.51**	0.52**	0.59*	0.52*	0.59*	0.60*	0.48*	0.45*	0.49*	0.51*
	(-3.14)	(-3.48)	(-3.12)	(-3.05)	(-2.04)	(-2.42)	(-2.01)	(-1.99)	(-2.45)	(-2.22)	(-2.13)	(-1.98)
2009-2010	0.50**	0.40***	0.45**	0.48**	0.46*	0.44*	0.53*	0.56	0.68	0.42*	0.44*	0.46*
	(-2.87)	(-3.60)	(-3.17)	(-3.02)	(-2.55)	(-2.48)	(-2.01)	(-1.89)	(-1.09)	(-2.56)	(-2.49)	(-2.34)
2010-2011	0.35***	0.32***	0.37**	0.39**	0.36**	0.32**	0.40*	0.42*	0.42*	0.40	0.43	0.45
	(-3.70)	(-3.49)	(-3.03)	(-2.87)	(-2.92)	(-2.78)	(-2.29)	(-2.17)	(-1.98)	(-1.82)	(-1.74)	(-1.64)
AC*2008-2009		1.66	1.50	1.62		1.65	1.50	1.63		1.30	1.15	1.10
		(0.98)	(0.75)	(0.91)		(0.81)	(0.64)	(0.78)		(0.44)	(0.23)	(0.15)
AC*2009-2010		2.34	2.14	2.27		1.20	1.11	1.13		5.40*	5.01*	4.87*
		(1.47)	(1.27)	(1.35)		(0.23)	(0.13)	(0.16)		(2.25)	(2.10)	(2.43)
AC*2010-2011		1.41	1.35	1.35		1.47	1.39	1.33		1.19	1.16	1.22
		(0.51)	(0.43)	(0.42)		(0.47)	(0.39)	(0.34)		(0.18)	(0.16)	(0.21)
Constant	0.30***	0.33***	0.14	0.34	0.18***	0.19***	0.00	0.00	0.08***	0.09***	19.68	21.22
	(-7.29)	(-6.64)	(-0.55)	(-0.30)	(-8.47)	(-7.85)	(-1.44)	(-1.34)	(-10.72)	(-10.39)	(0.67)	(0.68)
Teacher Controls			X	X			X	X			X	X
School Controls				X				X				X
Observations	2890	2890	2890	2890	2890	2890	2890	2890	2890	2890	2890	2890
Deviance	290121	289201	280539	273921	225600	225318	218428	211528	138934	137391	132469	127504
Notes. Estimates adjusted using BTLS longitudinal probability weights. Logistic regression coefficients reported as odds ratios. T statistics (in parentheses) are based on standard errors clustered at the teacher level. * p < 0.05 ** p < 0.01 *** p < 0.001.

In fact, although the direction of the odds ratios suggests that alternatively certified teachers have a greater risk of turnover than TC teachers after their second, third, and fourth year, this difference is only significant for leaving teaching after the third year (i.e., the 20092010 school year). Following their third year, AC teachers risk of leaving teaching is 440% greater than TC teachers. Controls mitigate the overall turnover differences between novice TC teachers and AC teachers, although in this one instance of a significant difference, the results remain significant. In supplementary analysis presented in Table A1, we separate AC teachers by whether they graduated from a selective university as a proxy for the selectivity of their AC program. In general, these results are comparable with what is presented in Table 3. Additionally, this analysis shows that AC teachers high risk of leaving teaching after their third year is driven by AC teachers who graduated from unselective universities.

To facilitate interpretation of these results, in Figure 1 we plot the fitted hazard and survivor functions based on this fully interactive discrete time hazard model, holding all variables in the model at their mean and separating the results by entry pathway. Compared to the unadjusted hazard and survival probabilities in Table 2, the addition of teacher and student controls narrows the observed differences between AC and TC teachers. Table 2 showed a 10-percentage point gap in the turnover rate between AC and TC teachers after four years. When controlling for teacher and school characteristics, the survival probability is 39% for TC teachers compared to 35% for AC teachers (top right panel of Figure 1). The difference in survival probability is driven by AC teachers higher risk of turnover following the 20092010 and 20102011 school years (top left panel of Figure 1).

Figure 1. Fitted Hazard and Survival Probabilities Describing the Risk of Turnover (top panel), Moving Schools (middle panel) and Leaving the Teaching Profession (bottom panel)

Results in the middle panel of Figure 1 show that the difference in survival probability between AC and TC teachers is comprised completely of differences in moving schools. By the end of their fourth year, 40% of TC teachers had moved schools compared to 45% of AC teachers. This finding implies that AC teachers may adopt what Evans (2011, p. 286) refers to as a get in the door approach to getting hired, whereby they begin their career in difficult to staff schools with intentions to move once they had earned their full certification.

The bottom panel of Figure 2 shows that, consistent with results from Table 3, AC teachers had a much greater risk of leaving teaching after their third year, conditional on not having previously turned over. Yet, given that TC teachers tended to have a higher risk of leaving following their first year, there were no overall differences in the survival probability between AC and TC teachers at the end of four years.

IN-SERVICE SUPPORTS FOR TEACHER RETENTION

We have found that early career AC teachers are more slightly more likely to move schools and, following their third year, more likely to leave the profession compared to TC teachers. We have also shown that as much as half of the turnover gap between AC and TC teacher can be explained by observable differences in teacher and school characteristics. With less commitment to teaching and less preparation, in-service induction supports may be particularly influential for the retention of novice AC teachers. We examine the role that in-service induction supports play in reducing the attrition rates of AC teachers in Table 4a and professional development in Table 4b.

Table 4. Logistic Regression Analysis of the Differential Effect of In-Service Induction Supports on Leaving Teaching for Alternatively Certified Teachers
Panel A. Induction Supports
	Mentor	Induction Program	Shared planning time	New teacher class	Extra assistance	Regular commun-	Mentor feedback
						ication
Alternatively certified (AC)	0.58	0.72	0.62	0.54	0.74	0.75	1.64
	(-0.58)	(-0.54)	(-0.96)	(-0.95)	(-0.64)	(-0.41)	(0.67)
Organizational Support	1.52	0.76	0.97	0.75	1.64	0.55	1.56
	(1.18)	(-0.82)	(-0.10)	(-0.86)	(1.52)	(-1.25)	(1.19)
Alternatively certified (AC)*Support	0.85	0.65	0.74	0.89	0.24*	0.59	0.26*
	(-0.20)	(-0.76)	(-0.56)	(-0.19)	(-2.29)	(-0.79)	(-1.96)
Observations	2860	2890	2810	2810	2810	2810	2810
Panel B. Professional Development
	Content	Technology	Reading	Classroom management	Students with disabilities	LEP students	PD matched needs
Alternatively certified (AC)	0.91	0.46	0.61	0.30	0.37	0.57	0.47
	(-0.16)	(-1.62)	(-0.99)	(-1.95)	(-1.85)	(-1.28)	(-1.10)
Organizational Support	0.85	0.58	0.88	0.86	0.67	0.98	0.86
	(-0.39)	(-1.74)	(-0.40)	(-0.53)	(-1.20)	(-0.05)	(-0.46)
Alternatively certified (AC)*Support	0.40	1.35	0.72	2.54	2.33	0.87	1.18
	(-1.56)	(0.51)	(-0.57)	(1.59)	(1.47)	(-0.20)	(0.26)
Observations	2890	2890	2890	2880	2880	2880	2840
Notes. Models include full controls. Estimates adjusted using BTLS longitudinal probability weights. Logistic regression coefficients reported as odds ratios. T statistics (in parentheses) are based on standard errors clustered at the teacher level. * p < 0.05 ** p < 0.01 *** p < 0.001.

In Table 4a we focus on induction supports for new teachers. Among early career AC teachers, we find extra classroom assistance (0.24; p = 0.02) and quality of mentor feedback (0.26; p = 0.05) to be associated with lower odds of leaving teaching among AC teachers. Among AC teachers who receive extra classroom assistance, the predicted probability of leaving teaching is 0.026. The predicted probability of leaving for AC teachers who do not receive this support is 0.061. Among AC teachers who report receiving helpful feedback from their mentor, the predicted probability of leaving teaching is 0.047. The predicted probability of leaving for AC teachers who do not receive this support is 0.10. Table A2 reports the corresponding estimates model in terms of moving schools. We find no significant evidence of a relationship between in-service supports and moving schools for AC teachers. We find marginally significant evidence that AC teachers assigned to a mentor are predicted to move schools at lower rates and AC teachers who participated in shared planning time have greater odds of moving schools.

In Table 4b we estimate a series of models predicting the relationship between participation in professional development and leaving teaching. We find marginally significant evidence that early career AC teachers who report receiving content-based professional development had lower odds of leaving teaching than those who did not receive such training. Although the interaction between AC and content PD is not significant, when equalizing teacher background and school characteristics, this difference translates to a marginal effect of 0.054 (p = 0.04). Although not statistically significant, the direction of the odds ratio on classroom management suggests that AC teachers who receive PD related to classroom management may be more likely to leave teaching. We find the opposite relationship in terms of moving schools. For AC teachers who received classroom management PD, the predicted probability of moving was 0.09. AC teachers who did not receive classroom management PD had a much larger predicted probability of moving0.17. This difference translates to a marginal effect of 0.08 (p = 0.08).

THE TRANSITION OUT OF THE CLASSROOM

Given our goal of better understanding the early career experience of AC teachers, we also examine descriptive evidence of the occupational status of teachers who left teaching. If teachers exit from the classroom is driven by a transition into non-instructional positions within education, ACPs may help springboard talented individuals into administrative or instructional coaching positions. On the other hand, the popular perception of selective ACPs suggests that participation is more akin to the two- to three-year commitment in programs such as Peace Corps or AmeriCorps. Once this commitment of service is complete, the teacher would transition into graduate school or an occupation outside the field of education. To distinguish these possible outcomes, we compare AC and TC teachers response to the question: What is your current MAIN occupational status? For this analysis, we focus on the subsample of teachers who left teaching at any point in their first four years and completed the former teacher questionnaire. In Table 5, we present teachers occupation for the year after they left teaching.

Table 5. Teacher Occupational Status After Leaving Teaching
	Traditional Certification	Alternative Certification
Working in education not as a K12 classroom teacher	0.26	0.32
Working in pre-K or postsecondary education	0.07	0.05
Working in an occupation outside the field of education	0.18	0.11
Student at a college or university	0.07	0.27+
Caring for family members	0.09	0.02*
Unemployed and seeking work	0.14	0.17
Other (Retired, disabled, etc.)	0.20	0.06*
Observations	180	60
Notes. Estimates adjusted using BTLS survey weights. Statistical significance based on a t-test, adjusting for clustering at the school level with robust standard errors. + p < 0.10 * p < 0.05 ** p < 0.01 *** p<0.001. Sample sizes rounded in accordance with NCES nondisclosure rules.

We find that after exiting the classroom, AC and TC teachers worked in positions inside education, either in K12, pre-K, or postsecondary education, at similar rates. Thirty-seven percent of AC teachers 33% of TC teachers worked in education but not as a K12 classroom teacher. Counter to the perception that ACPs are used primarily as an interlude to beginning a formal career, only 11% of AC teachers who left teaching are working in an occupation outside the field of education the year following their departure, a smaller but not significant difference compared to TC teachers (18%). Yet, we do find that AC teachers are more likely to enter graduate school after leaving teaching (27% versus 7%). AC teachers were less likely to be caring for a family member (2% versus 9%).

DISCUSSION AND LIMITATIONS

Our results show that even though alternatively certified teachers tend to receive few additional in-service supports, when they do, supports such as extra classroom assistance, mentor feedback, and, to a lesser degree, content professional development are associated with lower odds of leaving teaching. Yet, it is worth noting that the in-service supports that are asked about in the BTLS give little indication of their quality. One example of the importance of data on the quality of supports is evident in our own findings. We find no evidence that assignment to a mentor is linked with reduced odds of leaving teaching for AC teachers. However, when AC teachers report receiving helpful feedback from their mentor, they are predicted to leave teaching at much lower rates. Our results could also be biased to the extent to which the unobserved quality of organizational supports are correlated with the school placements of AC and TC teachers. Furthermore, that the in-service induction supports are only measured for teachers first year limits our understanding of the extent to which AC teachers differentially receive or benefit from in-service supports into their second and third years of teaching. Qualitative research on this topic or better measures of induction activities could strengthen further research that aims to understand the ways in which teacher background and in-service support predict early career teacher retention.

An additional limitation of this study relates to the scope. A tradeoff of studying national patterns of alternative certification teacher turnover is the lack of detail regarding specific characteristics of alternative certification programs. Previous research describes much in diversity in alternative certification programs (Humphrey & Wechsler, 2007). In supplementary analysis, we used selectivity of the teachers undergraduate institution as a proxy for the selectivity of the ACP. This analysis yielded few significant differences from our main analysis, other than to drive home the point that the high rates of leaving teaching at the end of the third year are driven by AC teachers who attended less selective college.

Finally, the findings should be interpreted in the broader context of the Great Recession, which hit the United States in the middle of the 20072008 school year and resulted in over 100,000 teacher layoffs nationwide, layoffs that were more likely to impact early career teachers (Goldhaber, Strunk, Brown, & Knight, 2016). Voluntary turnover during this time may have been suppressed while involuntary turnover (including layoffs) may have affected AC and TC teachers differently.

CONCLUSION

With new teachers increasingly likely to have entered teaching through an alternative certification program, the central debate remains of the extent to which ACPs are helping to recruit and retain a qualified teachers, particularly in high-needs schools (Johnson, Birkeland, & Peske, 2009). In this study, we show that AC teachers enter teaching feeling less prepared, receive few additional organizational supports to facilitate their transition into teaching, and are more likely to turn over than traditionally prepared teachers. The 10-percentage point gap in the turnover between early career AC and TC teachers is, in large part, explained by observable teacher and school characteristics. While most of this difference is driven by the higher rates with which early career AC teachers move schools, we also find notable differences in the patterns with which AC and TC teachers leave during their first four years. While TC teachers are slightly more likely to leave teaching following their first year, conditional on remaining in the same school, AC teachers are at much greater risk of leaving teaching following their third year. As many states require teachers to complete the coursework and testing by the end of their third year, the high levels of exiting teaching at this point in their career may indicate that AC teachers are not meeting these prerequisites for full certification. Alternatively, when teachers enter an alternative certification program, they may commit to teach for three years, only to leave teaching once they have fulfilled this programmatic commitment.

Previous research suggests that the higher turnover rates of alternatively certified teachers are linked to less of a commitment to remain in teaching, the lack of pre-service training, and lower levels of preparedness (Darling-Hammond, Chung, & Frelow, 2002; Kee, 2012). In this paper, we argue that in-service organizational supports have the potential to play an important role in the retention of AC teachers, as these teachers receive the bulk of their training once they are already teacher of record in a classroom. We show some evidence that AC teachers differentially benefit from extra classroom assistance, quality of mentor feedback, and content professional development, which were all associated with lower odds of leaving teaching. Yet, counter to our hypothesis that AC teachers would also receive additional induction supports during their first year, we only found AC teachers to be assigned to a mentor at higher rates than teachers who entered teaching through a traditional university-based program.

With little evidence of differences in the in-service supports received by AC and TC teachers and some evidence that AC teachers have higher retention rates with some organizational supports, our findings suggest that schools and districts could do more to target induction supports for novice AC teachers. Given that AC teachers receive most of their training once they begin teaching, schools and districts can customize supports for AC teachers to fit the needs of their local context. Teacher residency programs are one model of how districts can structure teacher preparation to meet both the needs of the district and the novice teacher (Papay et al., 2012; Silva et al., 2014). This approach also conforms to the argument made by Youngs, Frank, Thum, &and Low (2012), that induction programs are most influential when they are customized to the local school context where new teachers receive the majority of their mentoring from department chairs, other teachers in grade- or subject-level teams, and administrators. It also conforms to long-standing calls by teacher educators to transform the preparation of teachers into a shared responsibility of districts and teacher preparation programs (Goodlad, 1991; Robinson & Darling-Hammond, 1994).

In summary, this study provides national evidence related to the early career experiences of AC teachers. In particular, the focus on in-service supports adds to our understanding of how AC teachers may differentially benefit from induction supports given their lack of significant pre-service training. Given the role that schools play in this induction process, future research should explore the possible strain that new AC teachers place on their schools, particularly in schools serving high concentrations of children from traditionally underserved racial/ethnic groups. Future research should also focus on the ways in which organizational supports offered to novice teachers are linked not only higher retention rates but improvements in instruction as well. With many large urban districts signing expensive contracts with TFA and the TNTP to provide teachers in high-needs schools, it is critical to understand the in-service support needed to compensate for the lack of pre-service training. This focus could target in-service supports to help AC teachers develop their instructional practices and also induct them into their school community, both of which may foster greater retention.

Notes

1. In each of the remaining years of the survey, approximately 40 teachers returned to teaching each year (2%). In looking at differences in the rate of return for AC and TC teachers, we found no practically or statistically significant differences.

2. In supplementary analysis, we use selectivity of teachers undergraduate institution as a proxy for the selectivity of the alternative certification program. We link the teacher report of their undergraduate institution with NCESs Barrons data file. Barrons ranks institutions by seven levels of competitiveness. We classified individuals as having attended one of three institution types: most selective, which we coded to include Barrons rankings of most and highly competitive institutions, very selective, and all other levels of university competitiveness. We distinguish between alternatively certified teachers who attended a selective college or university and those who attended a non-selective institution. Although institutional selectivity has served as a primary recruiting mechanism for selective ACPs, this distinction still is an imperfect measure of how alternative certification programs select prospective teachers.

3. The wording for the question related to mentoring changed after the base year. In the base year, respondents were asked did you receive Ongoing guidance or feedback from a master or mentor teacher? In subsequent waves, teachers were asked, During the current school year, are you working with a master or mentor teacher who was assigned by your school or district?

4. In results not presented, when teacher and school controls are added to this model, this higher overall turnover rate among early career selective AC teachers becomes even smaller, suggesting that differences in AC and TC teachers backgrounds and the schools in which they begin their career. For example, we find Black teachers, teachers working part time, and teachers who have earned more than a masters degree or more likely to turnover. Of the school characteristics, principal effectiveness is associated with lower turnover rates.

References

Bastian, K. B., & Henry, G. T. (2014) Teachers without borders: Consequences of teacher labor force mobility. Educational Evaluation and Policy Analysis, 37(2), 163183.

Boyd, D., Grossman, P. L., Hammerness, K., Lankford, R. H., Loeb, S., McDonald, M., Reininger, M., Ronfeldt, M., & Wyckoff, J. (2008). Surveying the landscape of teacher education in New York City: Constrained variation and the challenge of innovation. Educational Evaluation and Policy Analysis, 30(4), 319343.

Brantlinger, A., & Smith, B. (2013). Alternative teacher certification and the new professionalism: The pre-service preparation of mathematics teachers in the New York City Teaching Fellows Program. Teachers College Record, 115(7), 144.

Clark, M. A., Chiang, H. S., Silva, T., McConnell, S., Sonnenfeld, K., Erbe, A., & Puma, M. (2013). The effectiveness of secondary math teachers from Teach for America and the teaching fellows programs (NCEE 2013-4015). Washington, DC: National Center for Education Evaluation and Regional Assistance, Institute of Education Sciences, U.S. Department of Education.

Clotfelter, C. T., Ladd, H. F., & Vigdor, J. L. (2010). Teacher credentials and student achievement in high school: A cross-subject analysis with student fixed effects. The Journal of Human Resources, 45(3), 655681.

Constantine, J., Player, D., Silva, T., Hallgren, K., Grider, M., Deke, J., & Warner, E. (2009). An evaluation of teachers trained through different routes to certification. (NCEE Publication No. 2009-4043). Washington, DC: Institute of Education Science.

Consuegra, E., Engels, N., & Stryyven, K. (2014). Beginning teachers experiences of the workplace learning environment in alternative teacher certification programs: A mixed methods approach. Teaching and Teacher Education, 42, 7988.

Darling-Hammond, L., Chung, R., & Frelow, F. (2002). Variation in teacher preparation: How well do different pathways prepare teachers to teach? Journal of Teacher Education, 53(4), 286302.

DeAngelis, K. J., Wall, A. F., Che, J. (2013). The impact of preservice preparation and early career support on novice teachers career intentions and decisions. Journal of Teacher Education, 64(4), 338355.

Duke, L., Karson, A., & Wheeler, J. (2006). Do mentoring and induction programs have greater benefits for teachers who lack pre-service training? Journal of Public and International Affairs, 17(2), 6182.

Evans, E. (2010) Job queues, certification status, and the education labor market. Educational Policy, 25(2), 267298.

Feiman-Nemser, S. (2001). From preparation to practice: Designing a continuum to strengthen and sustain teaching. Teachers College Record, 103(6), 10131055.

Glazerman, S., Mayer, D., & Decker, P. (2006). Alternative routes to teaching: The impacts of Teach for America on student achievement and other outcomes. Journal of Policy Analysis and Management, 25(1), 7596.

Goldhaber, D., Strunk, K. O., Brown, N., & Knight, D. S. (2016). Lessons learned from the Great Recession: Layoffs and RIF-induced teacher shuffle. Educational Evaluation and Policy Analysis, 38(3), 517548.

Good, T. L., McCaslin, M., Tsang, H. Y., Zhang, J., Wiley, C. R. H., Bozack, A. R., & Hester, W. (2006). How well do 1st-year teachers teach: Does type of preparation make a difference? Journal of Teacher Education, 57(4), 410430.

Goodlad, J. I. (1991). School-university partnerships. Education Digest, 56(8), 5861.

Gray, L., Goldring, R., & Taie, S. (2015). Users Manual for the First Through Fifth Waves of the 200708 Restricted-use Beginning Teacher Longitudinal Study Data File (NCES 2015-338). U.S. Department of Education. Washington, DC: National Center for Education Statistics.

Hill, H. C., Rowan, B., & Ball, D. L. (2005). Effects of teachers mathematical knowledge for teaching on student achievement. American Educational Research Journal, 42(2), 371406.

Humphrey, D. C., & Wechsler, M. E. (2007). Insights into alternative certification: Initial findings from a national study. Teachers College Record, 109(3), 483530.

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

Ingersoll R. M., & Strong, M. (2011). The impact of induction and mentoring programs for beginning teachers: A critical review of the research. Review of Educational Research, 81(2), 201233

Johnson, S. M., Birkeland, S., & Peske, H. G. (2009). Is fast-track preparation enough? It depends. In P. Grossman & S. Loeb (Eds.) Alternative routes to teaching. Mapping the new landscape of teacher education (pp. 101128). Cambridge, MA: Harvard Education Press.

Johnson, S. M., Kardos, S. M., Kauffman, D., Liu, E., & Donaldson, M. L. (2004). The support gap: New teachers early experience in high-income and low-income schools. Education Policy Analysis Archives, 12(61), 123.

Kane, T. J., Rockoff, J. E., & Staiger, D. O. (2008). What does certification tell us about teacher effectiveness? Evidence from New York City. Economics of Education Review, 27(6), 615631.

Kee, A. N. (2011). Feelings of preparedness among alternatively certified teachers: What is the role of program features? Journal of Teacher Education, 63(1), 2338.

National Council for Accreditation of Teacher Education. (2010). Transforming teacher education through clinical practice: A national strategy to prepare effective teachers: Report of the Blue Ribbon Panel of Clinical Preparation and Partnerships for Improved Learning. Washington, DC: NCATE.

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.

Redding, C., & Smith, T. M. (2016). Easy in, easy outAre alternatively certified teachers turning over at increased rates? American Educational Research Journal, 53(4), 10861125.

Robinson, S. P., & Darling-Hammond, L. (1994). Change for collaboration and collaboration for change: Transforming teaching through school-university partnerships. In L. Darling-Hammond (Ed.), Professional development schools: Schools for developing a profession (pp. 20318219). New York: Teachers College Press.

Silva, T., McKie, A., Knechtel, V., Gleason, P., & Makowsky, L. (2014). Teaching residency programs: A multisite look at a new model to prepare teachers for high-need schools (NCEE 2015- 4002). Washington, DC: National Center for Education Evaluation and Regional Assistance, Institute of Education Sciences, U.S. Department of Education.

Singer, J. D. & Willett, J. B. (2003) Applied longitudinal data analysis: Modeling change and event occurrence. New York: Oxford Press.

Xu, Z., Hannaway, J., & Taylor, C. (2011). Making a difference? The effects of Teach for America in high school. Journal of Policy Analysis and Management, 30(3), 447469.

Youngs, P. (2007). District induction policy and new teachers experiences: An examination of local policy implementation in Connecticut. Teachers College Record, 109(4), 797837.

Youngs, P., Frank, K. A., Thum, Y. M., & Low, M. (2012). The motivation of teachers to produce human capital and conform to their social contexts. In T. Smith, L. Desimone, & A. C. Porter (Eds.), Yearbook of the National Society for the Study of Education: Vol. 110. Organization and effectiveness of high-intensity induction programs for new teachers (pp. 248272). Malden, MA: Blackwell Publishing.

Table A1. Discrete Time Hazard Models of the Odds that a Teacher Will Turn Over, Move Schools, or Leave Teaching

	Turnover vs. Stay				Move vs. Stay				Leave vs. Stay
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Non-selective AC	1.14	0.79	0.75	0.62	1.23	1.04	1.09	0.91	0.94	0.43	0.36	0.34*
	(0.53)	(-0.56)	(-0.60)	(-1.04)	(0.70)	(0.09)	(0.16)	(-0.18)	(-0.16)	(-1.57)	(-1.84)	(-1.99)
Selective AC	1.45	1.07	1.20	0.97	1.47	1.12	1.30	0.98	1.26	0.96	1.00	1.02
	(1.54)	(0.18)	(0.48)	(-0.08)	(1.16)	(0.26)	(0.59)	(-0.04)	(0.62)	(-0.07)	(-0.00)	(0.03)
2008-2009	0.52**	0.46***	0.51**	0.52**	0.59*	0.52*	0.59*	0.60*	0.48*	0.45*	0.49*	0.51*
	(-3.13)	(-3.48)	(-3.11)	(-3.04)	(-2.04)	(-2.42)	(-2.00)	(-1.98)	(-2.44)	(-2.22)	(-2.12)	(-1.97)
2009-2010	0.50**	0.40***	0.46**	0.48**	0.46*	0.44*	0.53*	0.56	0.68	0.42*	0.44*	0.46*
	(-2.86)	(-3.60)	(-3.16)	(-3.01)	(-2.55)	(-2.48)	(-2.01)	(-1.88)	(-1.09)	(-2.56)	(-2.48)	(-2.32)
2010-2011	0.35***	0.32***	0.37**	0.40**	0.36**	0.32**	0.40*	0.42*	0.42*	0.40	0.43	0.45
	(-3.70)	(-3.49)	(-3.02)	(-2.87)	(-2.92)	(-2.78)	(-2.28)	(-2.16)	(-1.97)	(-1.82)	(-1.73)	(-1.63)
Non-selective AC*20082009		1.84	1.64	1.76		1.68	1.51	1.70		1.81	1.57	1.43
		(1.09)	(0.81)	(0.95)		(0.78)	(0.59)	(0.76)		(0.80)	(0.60)	(0.45)
Non-selective AC*20092010		2.25	2.05	2.14		0.70	0.64	0.65		8.86*	8.57*	7.74**
		(1.12)	(0.95)	(0.98)		(-0.42)	(-0.53)	(-0.51)		(2.34)	(2.22)	(2.59)
Non-selective AC*20102011		1.47	1.48	1.40		1.74	1.64	1.49		0.46	0.50	0.50
		(0.44)	(0.42)	(0.36)		(0.56)	(0.48)	(0.38)		(-0.75)	(-0.66)	(-0.65)
Selective AC*20082009		1.41	1.32	1.43		1.59	1.48	1.50		0.88	0.83	0.91
		(0.35)	(0.28)	(0.37)		(0.40)	(0.34)	(0.36)		(-0.16)	(-0.25)	(-0.13)
Selective AC*20092010		2.53	2.33	2.54		2.22	2.13	2.16		2.55	2.21	2.52
		(1.08)	(0.96)	(1.08)		(0.74)	(0.68)	(0.72)		(0.91)	(0.76)	(0.91)
Selective AC*20102011		1.34	1.20	1.33		0.97	0.95	1.02		1.99	1.58	1.82
		(0.36)	(0.22)	(0.35)		(-0.03)	(-0.05)	(0.02)		(0.60)	(0.40)	(0.51)
Constant	0.30***	0.33***	0.19	0.53	0.18***	0.19***	0.00	0.00	0.08***	0.09***	24.61	29.44
	(-7.28)	(-6.64)	(-0.46)	(-0.18)	(-8.46)	(-7.85)	(-1.39)	(-1.29)	(-10.73)	(-10.39)	(0.72)	(0.76)
Teacher Controls			X	X			X	X			X	X
School Controls				X				X				X
Observations	2890	2890	2890	2890	2890	2890	2890	2890	2890	2890	2890	2890
Deviance	290121	289201	280539	273921	225600	225318	218428	211528	138934	137391	132469	127504
Notes. Estimates adjusted using BTLS longitudinal probability weights. Logistic regression coefficients reported as odds ratios. T statistics (in parentheses) are based on standard errors clustered at the teacher level. * p < 0.05 ** p < 0.01 *** p < 0.001.

Table A2. Logistic Regression Analysis of the Differential Effect of In-Service Induction Supports on Moving Schools for Alternatively Certified Teachers
Panel A. Induction Supports
	Mentor	Induction Program	Shared planning time	New teacher class	Extra assistance	Regular commun-	Mentor feedback
						ication
Alternatively certified (AC)	1.97	0.54	0.56	0.74	0.74	1.14	0.80
	(1.31)	(-1.08)	(-1.33)	(-0.61)	(-0.65)	(0.20)	(-0.34)
Organizational Support	1.59	0.74	0.81	0.54*	0.82	0.47	1.82
	(1.81)	(-0.98)	(-0.89)	(-2.21)	(-0.69)	(-1.88)	(1.77)
Alternatively certified (AC)*Support	0.44	2.03	2.20	1.35	2.33	0.81	1.22
	(-1.56)	(1.26)	(1.72)	(0.59)	(1.73)	(-0.33)	(0.31)
Observations	2860	2890	2810	2810	2810	2810	2810
Panel B. Professional Development
	Content	Technology	Reading	Classroom management	Students with disabilities	LEP students	PD matched needs
Alternatively certified (AC)	1.30	0.74	0.88	1.50	0.80	0.83	1.04
	(0.57)	(-0.67)	(-0.32)	(0.90)	(-0.56)	(-0.47)	(0.08)
Organizational Support	1.46	0.85	1.20	1.05	1.10	1.15	1.35
	(1.55)	(-0.65)	(0.77)	(0.22)	(0.36)	(0.40)	(1.13)
Alternatively certified (AC)*Support	0.64	1.64	1.18	0.44	1.42	1.43	0.90
	(-0.86)	(0.94)	(0.32)	(-1.68)	(0.70)	(0.57)	(-0.20)
Observations	2890	2890	2890	2880	2880	2880	2840
Notes. Models include full controls. Estimates adjusted using BTLS longitudinal probability weights. Logistic regression coefficients reported as odds ratios. T statistics (in parentheses) are based on standard errors clustered at the teacher level. * p < 0.05 ** p < 0.01 *** p < 0.001.

Cite This Article as: Teachers College Record Volume 121 Number 11, 2019, p. 1-32 https://www.tcrecord.org ID Number: 22818, Date Accessed: 10/21/2021 8:41:45 PM Purchase Reprint Rights for this article or review