Beginning English Teacher Attrition, Mobility, and Retention
By Hahs-Vaughn, Debbie L Scherff, Lisa
ABSTRACT. Although much research on teacher attrition and mobility exists, few researchers have addressed English teachers specifically. The present authors, using the 1999-2000 Schools and Staffing Survey (SASS) and the Teacher Follow-Up Survey (TFS; National Center for Education Statistics, 2005) examined individual and school characteristics and mentoring and induction activities that affect beginning English teachers’ attrition, mobility, and retention. The results indicated that only salary was statistically significantly related to increased odds of beginning English teachers’ leaving the profession. No factors related to decreased attrition. In terms of mobility, no teacher or school characteristics were associated with migration (i.e., changing schools). Reviewing combined effects of mentoring and induction activities when controlling for teacher and school characteristics, the authors found that the results suggested none of the activities were related to attrition and migration. Keywords: beginning English teachers, induction activities, mentoring, teacher attrition, teacher migration, teacher retention
ACCORDING TO THE U.S. Department of Education, 8% of all teachers (public and private) changed schools between 1999-2000 and 2000- 2001 (Luekens, Lyter, & Fox, 2004). During the same time period, 7% of public school teachers and 13% of private school teachers left teaching (Luekens et al.). Thus, one of the biggest challenges facing schools today comprises attracting teachers and then retaining them (Imazeki, 2005). Attrition and mobility create teacher shortages that result in a poor return on investment because of the money spent to recruit, hire, train, and support new teachers (Haselkorn & Fideler, 1999). Hare and Heap (2001) estimated that the cost to replace one teacher is 25%-35% of his or her annual salary and benefits. A study of the California New Teacher Project-which focuses on new teacher support, including mentoring and professional development workshops-found that districts saved an average of 31 cents after the 1st year for every dollar spent on the program and 68 cents after the 2nd year because of decreased recruitment and hiring costs (California Commission on Teacher Credentialing and California Department of Education, 1992; Curran & Goldrick, 2002). Citing a Texas study showing that teacher turnover costs the state around $329 million per year, Darling-Hammond (2003) reiterated, “early attrition bears enormous costs” (p. 8). Poor teacher retention rates (due to either attrition or mobility) can ultimately lead to substantial consequences on the quality of education provided to students (Imazeki).
Although researchers have consistently demonstrated high teacher- attrition rates (Darling-Hammond, 2003; Gold, 1996; Ingersoll, 2003; Smith & Ingersoll, 2004), little is known about the social, institutional, and personal factors that contribute to teacher attrition; researchers and educators have called for investigation of teaching conditions, mentoring, and induction programs as they relate to attrition and mobility (Curran & Goldrick, 2002; Imazeki, 2005). Although Smith and Ingersoll recently used data from a national survey to examine the relationship between (a) induction and mentoring of beginning teachers (with a focus on teachers of math, science, special education, and English as a second language) and (b) attrition, migration, and retention, little is known about English teachers specifically. Although some small-scale studies have examined the experiences of novice teachers (Bullough, 1989; Johnson & Birkeland, 2004; Rogers & Babinski, 2002), with some specifically examining English teachers (McCann, Johannessen, & Ricca, 2005; Scherff, 2006; Smagorinsky, Cook, Moore, Jackson, & Fry, 2004), and although those studies’ results have suggested reasons for dissatisfaction, the profession has had little success in lowering the attrition rate. For example, McCann et al.’s case study of novice English teachers showed that the ability to define a teacher persona and narrow the mismatch between expectations and job reality might reduce the attrition rate. They found that two of six teachers whom they studied quit by their 5th year. A similar follow- up study of 12 novice English teachers found that antagonistic colleagues, the demands of meeting special education requirements, and a lack of administrative support contributed to their unhappiness and their decision to change schools or leave the profession (Scherff).
Because of the predominance of high-stakes reading and writing assessments for middle and high school students, researchers’ study of individuals who teach English or language arts is critical. Burns (2007) concurred, claiming, “teachers, as individuals responsible for increasing student achievement, are directly implicated as a primary source of school failure. And as literacy achievement is a central agent for testing in current accountability mandates, literacy teachers and English teachers are particular targets for scrutiny” (p. 123). Educators who prepare and administrators who employ English teachers must understand not only who is entering the profession but also what factors influence novice teachers’ determination about whether to leave teaching or change schools early in their careers.
Leavers and Movers
Teacher turnover can exist in the form of either (a) attrition (teachers leaving the profession, or leavers) or (b) mobility or migration (teachers changing schools, or movers). The two alternatives have approximately the same percentage of annual turnover (Ingersoll, 2004). Most researchers on teacher attrition have not examined teacher mobility. This lack is because teachers who change schools, although yielding an attrition statistic at one school, become a new hire at another school. Thus, it may seem that teacher mobility is not a problem related to staffing schools and teacher shortages (Ingersoll). However, teacher attrition and teacher mobility have the same impact from the perspective of the organization: They create a shortage of faculty that must be replaced, so that examining movers in addition to leavers is critical. If movers are not considered and examined along with leavers, total teacher turnover appears far less problematic than it is for those viewing this issue from a school management perspective (Ingersoll). For this reason, in the present study, we examined teacher attrition and teacher mobility.
Although many factors have contributed to overall teacher attrition and mobility, most novices leave the profession because of low salaries, student discipline problems, lack of support, poor working conditions, inadequate preparation, and little opportunity to participate in decision making (Andrews & Martin, 2003; Cochran- Smith, 2004; Darling-Hammond, 2003; Hirsch, 2006; Ingersoll, 2003, 2004; Kent, 2000; Liu & Meyer, 2005). The inability to staff schools with qualified teachers is “directly connected to the social stratification processes” (Ingersoll, 2004, p. 3). High-poverty urban and rural districts are more likely to have higher teacher turnover (Hare & Heap, 2001; Haselkorn & Fideler, 1999; Imazeki, 2005). Imazeki found that higher salaries are related to decreased attrition; however, “higher salaries relative [emphasis added] to surrounding districts appear to increase attrition out of the profession” (p. 438). Moreover, increases in salary of up to 20% were necessary to decrease attrition, although “new female teachers consider future wages in at least their exit decisions” (Imazeki, p. 447). School districts have perceived the effectiveness of salaries on retention (yielding stayers) as moderately successful in decreasing attrition (Hare & Heap, 2001).
Others leave the profession because they never intended to stay in it over the long term. Johnson and Birkeland (2004) reported that potential teachers do not necessarily choose to stay in teaching because today’s work environment offers a plethora of career possibilities, including jobs with high status and pay, productive work environments, and chances for promotion to high levels. In addition, “serial careers are the norm, and short-term employment is common” (Johnson & Birkeland, 2004, p. 585).
As compared with public school teachers, private school teachers are more likely to leave teaching, although teachers from both sectors are equally likely to change schools (movers; Luekens et al., 2004). Likewise, the effect of gender on teacher attrition and mobility differs. Some researchers suggest that women are more likely to leave teaching (Stinebrickner, 1998), whereas other researchers suggest that women are less likely to leave (Imazeki, 2005). Recent researchers have identified another cause for teacher attrition numbers: the accountability associated with the No Child Left Behind Act (NCLB; 2002). In a recent study, over 55% of those surveyed indicated an “overemphasis on testing” as an influence on their decision to leave the profession (Hirsch, 2006).
Mentoring and Induction
One way to reduce the attrition rate is through supportive induction and mentoring programs (Darling-Hammond, 2003; Feiman- Nemser, 1996; Gold, 1996). Teacher induction programs are not necessarily extra training but do provide activities for teachers who have already completed basic preparation. Such programs are considered a bridge, enabling the student of teaching to become a teacher of students (Smith & Ingersoll, 2004). Teacher induction programs offer activities for beginning teachers (those with 3 or fewer years of teaching experience) that support, train, and assess them (Recruiting New Teachers, 2000), and they are tools to increase teacher retention. Induction activities may be mentoring programs, professional development, new teacher orientations, networking opportunities, or similar activities (Recruiting New Teachers, 2000). For example, an induction component to combat stress and foster encouragement among novice teachers is membership in network (i.e., support) groups. Research shows that such groups lower stress and feelings of isolation while promoting teacher enthusiasm, competency, and reflection (Chubbuck, Clift, Allard, & Quinlan, 2001; DeWert, Babinski, & Jones, 2003). Aspects of formal induction, such as implementing team teaching and providing common planning time, are also perceived as effective strategies for retention (Hare & Heap, 2001). However, the literature suggests a wide discrepancy among induction programs and their levels of support, with many programs providing generic information and strategies and lacking in breadth and depth that beginning teachers need (Darling-Hammond, Berry, Haselkorn, & Fideler, 1999; Odell, 1986). Although some states have comprehensive state-level policies that require or strongly encourage districts to provide teacher induction programs (Hare & Heap, 2001), the structure and content of new teacher induction programs vary widely by state and district (Curran & Goldrick, 2002). Induction can range from a single orientation held before the start of the school year to a series of activities and sessions lasting several years. Some programs are school based, whereas others are state adopted. In addition, although some programs have few requirements, others require novices to complete workshops (e.g., technology, discipline), create portfolios, take part in online discussions, and attend district-based meetings while trying to navigate their 1st year in the classroom. Too little or too much can be equally ineffective (Kaplan, Scherff, & Faulkner, 2005).
Research that the North Central Regional Educational Laboratory (NCREL; Hare & Heap, 2001) conducted indicates that districts perceive induction programs to be effective for retaining and recruiting new teachers, with mentoring as one of the most important attributes. Mentoring, personal support for novices from experts, is one aspect of induction. As with induction programs, mentoring can look decidedly different from school to school. At minimum, some principals simply assign mentors to novice teachers with no guidelines for their responsibilities. Sometimes mentors are assigned but are unwilling to take on that role. Good mentoring requires “strong collaboration, critical reflection, and excellent interpersonal skills” (Rogers & Babinski, 2002), and this relationship cannot take place when mentors do not want the role, are not comfortable in the role, or are not provided with professional development to take on the role. Some recommendations for effective mentoring include structured programs, mentor training, release time for mentors (e.g., allowing teachers paid time away from their classroom to mentor other teachers), and common planning time with novices (Darling-Hammond, 2003; Feiman-Nemser, 1996; Holloway, 2001). Effective mentoring combines the professional (observing, evaluating, and advising) and the personal (befriending and counseling; Rippon & Martin, 2006). These conditions are essential for programs to have the desired positive impact. The “mentoring of new teachers will never reach its potential unless it is guided by a deeper conceptualization that treats it as central to the task of transforming the teaching profession itself” (Hargreaves & Fullan, 2000).
Smith and Ingersoll (2004) conducted a study of induction program impact on beginning teachers’ retention and migration by using the restricted-use 1999-2000 Schools and Staffing Survey (SASS) and Teacher Follow-Up Survey (TFS) data sets (National Center for Education Statistics, 2005). The present study builds on their work and adds to the research base by examining the individual and school characteristics of beginning English teachers that affect teacher attrition, migration, and retention. Although Smith and Ingersoll studied teachers in all content areas, research focusing specifically on English teachers is important for a number of reasons. For example, despite significant research related to overall teacher attrition and mobility, little research specifically related to English teachers has been conducted. As we indicated earlier, high-stakes reading and writing assessments place English teachers under an especially critical lens (Burns, 2007) not applied equally to teachers in other areas, such as history or science. Understanding those who become English teachers and the factors that may influence novice English teachers to leave teaching or change schools early in their careers is critical for people who educate these teachers and those educators who employ them. Consequently, we did not include a comparison group of teachers for examination in the present study. Instead, limiting the study to examine only English teachers was appropriate to isolate this group of teachers.
For the purposes of the present study and as outlined in the SASS and TFS (Tourkin et al., 2004), we operationally defined attrition as leaving the teaching profession during the year between administrations of the SASS and TFS; migration as changing schools between administrations of the SASS and TFS; and retention as remaining at the same school between administrations of the SASS and TFS. In addition to examining only teachers who were in their first few years of teaching (i.e., began teaching between 1996 and 2000 or had been teaching for 4 years or less), we limited teachers to those who indicated that English or language arts was their primary teaching assignment during 1999-2000 (n = 186). We addressed the following research questions:
1. What is the profile of novice English teachers?
2. In what types of mentoring and induction activities do novice English teachers participate?
3. How does teacher retention, migration, or attrition relate to mentoring and induction activities when the researcher controls for gender, salary, and school characteristics?
We drew data from the publicly available 1999-2000 SASS and the 2000-2001 TFS (National Center for Education Statistics, 2005). The rationale for the creation and administration of the SASS was to “inform policymakers about the status of teaching and education, identify the areas that most need improvement, and clarify conflicting reports on issues related to policy initiatives, such as teacher shortages” (Tourkin et al., 2001, p. 1). We collected data on U.S. school districts, schools, principals, library media centers, and teachers, and from multiple school types, including public, private, charter, and Bureau of Indian Affairs (BIA). The SASS, conducted by the U.S. Department of Education’s National Center for Education Statistics (NCES), is the largest and most comprehensive survey related to U.S. K-12 education. Data collected from school districts included enrollment information, recruitment and hiring of teachers, compensation, school and student performance, and migrant education. Data gathered from schools included such categories as parent involvement and school safety, staffing, and technology. Data collected from principals included, for example, experience and training, attitudes and opinions about education and the school at which the principal is employed, and professional development activities in which the principal participates. The teacher questionnaires had items related to certification and training, professional development, student resources and assessment, working conditions, and decision making. Data collected in relation to library media centers included facilities, staffing, technology, and collaborations. The TFS was administered 1 year after the SASS to selected teachers (see Sampling Method in the following for additional information on teacher selection) who completed the SASS and was designed specifically to collect information on teacher retention, mobility, attrition, attitudes of teachers toward the profession, and job satisfaction. The data from the TFS are linked to the data from the SASS to allow researchers to examine relationships between those entities that were surveyed (Tourkin et al.). For additional general information about the SASS and TFS, see Tourkin et al.
Although the SASS and TFS provide researchers with an abundance of data on the K-12 environment and its employees, there are some limitations to the use of this secondary data. For example, only 1 year passed between administration of the SASS and that of the TFS. Although data is collected in the TFS to determine teacher attrition, mobility, and retention in this 1-year time frame, a complete teacher career history assessment is not collected. Thus, it is possible that teachers identified as stayers at the time of TFS administration had previously in their careers changed schools or temporarily left the teaching profession. Likewise, although the SASS collects information on the types of support teachers received during their 1st year of teaching (e.g., seminars or classes for beginning teachers), it provides no operational definitions to ensure that teachers have common perceptions of what this item includes. In addition, teachers who completed the SASS and who were not in their 1st year of teaching completed the items on mentoring and induction retrospectively. Thus, it is possible that teachers may have forgotten the types of activities in which they participated. Aside from the limitations of secondary data, the SASS and TFS provide the most comprehensive assessment of K-12 teachers to date, and the results of the present study contribute greatly to what is known about English or language arts teachers and their preparation, attrition, retention, and mobility. As mentioned earlier, the present study builds on Smith and Ingersoll’s (2004) work, limiting participants to only English teachers. However, limitations that prevent exact duplication of Smith and Ingersoll’s study exist either because of the researchers’ use of the restricted- use data (as compared with the public-use data that we used in the present study) or their examination of novice teachers in all teaching fields (as compared with the limitation to English or language arts teachers in the present study). For example, to prevent possible identification of respondents, some variables that Smith and Ingersoll used are available in the restricted-data file but are not accessible in the public-use data (e.g., detailed sector classification of school including charter and BIA), and categories in some variables have been collapsed in the public-use data as compared with the restricted file’s data (e.g., age, earnings, school enrollment). In addition, although Smith and Ingersoll examined all 1st-year teachers in all teaching fields, we limited the present study to include beginning teachers (SASS variable = T0122) and only those teaching English or language arts (i.e., whose main teaching assignment was English or language arts; SASS variable = ASSIGN_S). Thus, variables that Smith and Ingersoll included that were related to type of assignment (including math, science, special education, English as a second language) are excluded as predictors in the present study because they are not applicable. For purposes of the present study, beginning teachers are defined as those teachers who were in their 1st, 2nd, 3rd, or 4th year of teaching (i.e., began teaching between 1996 and 2000). In addition, we excluded a few variables that Smith and Ingersoll included from the models in the present study because of little variation. These included full-time status (4% of teachers in the present study did not have regular fulltime status), race (only 11% of teachers were non-White), and age (only 14% of teachers were 30 years of age or older). Table 1 presents the variables included in the models that we tested and the scales of the variables.
The SASS used a stratified probability sampling design. We selected schools in the first stage of sampling. The SASS public school sample was a three-stage stratified sample. The first stratification level was school type (Native American schools; schools in Delaware, Nevada, West Virginia; all remaining schools). Based on school type, the second level of stratification was by state (for Native American schools), by state and then district (for schools in Delaware, Nevada, West Virginia), or by state (for all remaining schools). The third stratification level was by grade (elementary, secondary, or combined). We oversampled schools with a Native American student enrollment of 19.5% or more to improve estimates from these schools.
We selected the sample of private schools by using a dual-frame approach. The private school list frame sample was a three-stage stratified sample: (a) school affiliation (20 affiliations such as military, Catholic, Hebrew day), (b) school level (elementary, secondary, combined), defined the same as that of public schools, and (c) census region (Northeast, Midwest, South, West). All area frame private schools were selected for the sample, and thus no stratification was necessary for area frame schools. A total of 11,139 schools were selected. The school districts associated with the selected schools created the school district sample and included 5,465 school districts.
From each school that was sampled, we drew a sample of teachers. The teacher-sampling frame was created from data that the selected schools reported on each teacher in the school (e.g., grade range taught, race, new or experienced teacher). In each school, teachers were stratified into the following categories: (a) Asian or Pacific Islander, (b) American Indian or Alaskan Native, (c) taught Limited English Proficiency students, (d) new (i.e., teachers with 3 years or less of experience), and (e) none of the above groups. New teachers were oversampled in private schools to ensure an adequate sample size for the SASS and TFS. A total of 72,058 teachers were selected. The overall SASS weighted response rate for public school teachers was 83.1%. The weight for private school teachers was 77.2%. The weight for BIA teachers was 87.4%. The weight for public charter school teachers was 78.6%. Researchers can get additional information on the SASS from the SASS user’s manual (Tourkin et al., 2004).
The TFS was a follow-up survey of 8,400 teachers who originally participated in the SASS approximately 1 year earlier. Luekens et al. (2004) designed the TFS to collect information on teacher attrition (leavers) and mobility (movers) and on those who remained in teaching (stayers). Of all the teachers who participated in the TFS, about 39% were stayers, about 26% were movers, and about 33% were leavers (Luekens et al.).
Adjustment for Complex Sampling Design
In complex sample designs, such as those in the SASS and TFS (i.e., stratifying the school sample, which resulted in a nonsimple random sample; and oversampling new school teachers, which resulted in sampling with differential probabilities), direct estimates of the sampling errors underestimate the variability because a simple random sample was not used. The preferred method of calculating sampling errors to address the complex sampling issues of the SASS and TFS is the replication method. The present analysis applied a full sample teacher weight to the data (TFSFINWT) and estimated variances by using the balanced repeated replication (BRR) method (replicate weights TFRPWT1-TFRPWT88) with AM software (AM Statistical Software, 2008). Researchers can find additional information on calculation of the weights, replicate weight, and other technical issues in the SASS technical manual (Tourkin et al., 2004). AM is a free, downloadable statistical software program that researchers at the American Institutes for Research created, are currently beta testing, and designed specifically to accommodate data sets that use complex sampling designs such as the SASS and TFS.
We present the results that emerged regarding our research questions in the following. We provide descriptive statistics to answer research questions 1 and 2. We provide multinomial logistic regression models to answer research question 3.
On Research Question 1: What is the Profile of Novice English Teachers?
Most beginning English teachers in our sample were White (89.1%; SE = .03), and the largest minority classification was Hispanic (5.8%; SE = .02), preceding Black (3.3%; SE = .01). The percentage of teachers who identified themselves as Asian (1.5%; SE = .01), American Indian or Alaska Native (3.0%; SE = .02), or Native Hawaiian or Other Pacific Islander (1.5%; SE = .01), fell below these figures. Most teachers were less than 30 years of age (86.2%; SE = .03), followed by 30-39 years of age (8.0%; SE = .03), 40-49 years of age (4.5%; SE = .02), and 50 years or older (1.2%; SE = .01). About 60% of beginning English teachers earned $25,000 or less. Nearly all beginning English teachers included in the present study held a bachelor’s degree (98.1%; SE = .01), but only 20% received a bachelor’s degree in English education (19.9%; SE = .05). The subject of English or language arts was not the only subject that these beginning teachers taught. When asked to identify the subjects taught during their most recent full week of teaching, 72% (average SE = .04) indicated that their first course was related to English or language arts (including literature; composition, journalism, or creative writing; English as a second language; reading or other English or language arts courses). The most frequently occurring non-English course was vocational family and consumer science (5%, SE = .04). Examining patterns for additional courses taught revealed similar patterns in terms of the percentage of teachers teaching English as opposed to other subjects. Although all the English teachers included in this study began teaching between 1996 and 2000, 77% had taught for 3 years or less (76.6%; SE = .07).
Of the schools in which the beginning English teachers taught, over 85% (86.6%; SE = .04) were regular schools, 8.3% (SE = .04) were schools that had a special program emphasis, 1.4% (SE = .01) were alternative schools, and 0.5% (SE = .004) were schools that served primarily special education students. Less than 1% of beginning English teachers taught at more than one school (0.2%; SE = .002). About one half of beginning English teachers taught at schools in urban fringes, and over 70% taught at schools in which 19% of students or less were eligible for free or reduced lunch. About one half of beginning English teachers taught at secondary schools. Over 60% of beginning English teachers taught at schools that had a total student enrollment of more than 500 students.
In terms of leavers and stayers, nearly 80% of beginning English teachers remained at the same school, and only 8% left teaching, yielding lower numbers as compared with all content area teachers (Smith & Ingersoll, 2004). In the present study, Table 1 shows descriptive statistics on beginning English teachers for the variables that we included in the multinomial logistic regression analyses. On Research Question 2: In What Types of Mentoring and Induction Activities Do Novice English Teachers Participate?
Of the mentoring and induction variables included in the present study, beginning English teachers were most likely to report participation in beginning seminars (56%). Of those surveyed, 50% indicated their having been assigned a mentor during their 1st year of teaching. Only 14% of beginning English teachers reported receiving reduced preparation time during their 1st year of teaching, and less than one fourth (21%) reported participating in teacher networking. However, these percentages were higher than those of teachers of other subjects (Smith & Ingersoll, 2004). Slightly more than one third indicated that they had participated in common planning time during their 1st year of teaching. This number was much lower than that of other teachers (Smith & Ingersoll). Examining the total number of mentoring and induction activities in which beginning English teachers participated showed us that most teachers participated in one activity (38.6%, SE = .05), preceding those in two (37.2%, SE = .06), those in three (10.3%, SE = .03), those in four (7.9%, SE = .04), and those in zero (5.9%, SE = .02). See Table 1 for percentages and standard errors.
On Research Question 3: How Does Teacher Retention, Migration, or Attrition Relate to Mentoring and Induction Activities When the Researcher Controls for Gender, Salary, and School Characteristics?
Many studies of teacher attrition treat attrition as a binary choice (leave or stay), which ignores the possibility that different factors may influence teacher mobility (i.e., migrate, or change schools) as compared with the factors that influence teachers who leave teaching or remain in teaching. Thus, the outcome for the present study includes teacher attrition, retention, and migration. The dependent variable for each model was the teaching status of the respondent as reported in the TFS (data set variable name = STATUS). We coded teachers as staying (i.e., stayed at the same school), migrating (i.e., migrated to a different school), or leaving the teaching profession (i.e., left teaching; see Table 1). Table 1 shows the dependent and independent variables that we included in the models and their respective scales.
Our plan for data analysis and the selection of variables for the models built on that of Smith and Ingersoll (2004). The dependent variable for each model was teaching status (stayed, migrated, or left teaching; see Table 1). To the extent possible, the independent variables were those that Smith and Ingersoll used with the aforementioned limitations. We computed hierarchical multinomial logistic regression models and reviewed results by using an alpha level of .05. We examined the adjusted Wald test as a measure for overall goodness of fit of the models. The results of the Wald test for Models 1-7 provided evidence to suggest that the models were not a statistically significant improvement over the intercept-only models (p values range from .074 to .361). Because, to the extent possible, this study was a replication of Smith and Ingersoll’s study, for preliminary analyses we present and discuss the models that examined the variables that those researchers used and the statistically significant factors in each. We caution researchers that interpreting the relationship of these factors to the outcome may not be valid because of evidence of a lack of model fit. We conducted additional analysis to improve model fit as well as to examine how participation in increasing numbers of mentoring or induction activities was related to attrition, migration, and retention (Model 8).
Model 1 examined the relationship between teacher and school characteristics, the likelihood that a teacher will leave teaching as compared with staying, and the likelihood that a teacher will migrate to a different school as compared with staying. Adding to the teacher and school characteristics, Models 2-7 reflect inclusion of mentoring and induction variables modeled independently and then in aggregate. In addition to including a reduced and more parsimonious model of teacher and school characteristics, Model 8 allowed examination of the relationship between the number of mentoring or induction activities in which the teacher participated and attrition, migration, and retention. Tables 2 and 3 present the odds ratios and related probability values for all models.
As in Model 1 (Table 2), when we held other variables constant, English or language arts teachers who were male and earned less than $20,000 per academic year were more likely to leave teaching as opposed to staying. This finding mirrors that of Smith and Ingersoll (2004). In particular, the odds of a man leaving teaching were nearly 8 times greater than those odds for a woman. The same trend was observed in earnings. Teachers who earned less than $20,000 per academic year were nearly 6.5 times as likely to leave teaching as compared with teachers who earned more than $20,000. However, no variables were significantly associated with the risk that a beginning English or language arts teacher would migrate to a different school after his or her 1st year of teaching (Table 3). The full results for the multinomial logistic regression analyses for Model 1 are shown in Table 4.
Model 2 built on Model 1, adding the examination of the relationship between participation in mentoring activities, the likelihood of leaving teaching as opposed to staying (Table 2), and the likelihood of migrating to a different school as opposed to staying (Table 3). Mirroring the results found for Model 1, Model 2 again indicated that English or language arts teachers who were men and earned less than $20,000 per academic year were more likely to leave teaching as compared with staying. Reflecting the same results that we found in Model 1, none of the variables in Model 2 were associated with the risk that a beginning English or language arts teacher would migrate to a different school. In terms of participation in mentoring, having a mentor during the 1st year of teaching appears to have had little impact on the odds that an English or language arts teacher would either leave teaching or change schools, because these results were not statistically significant. Table 5 shows the full results for the multinomial logistic regression analyses for Model 2.
Models 3 Through 6
In the remaining models, we examined the relationships between (a) attrition, migration, and retention and (b) beginning teacher support activities. Model 3 added to the previous models by examining participation in beginning teacher seminars. As indicated in the previous models, Model 3 provided evidence that English or language arts teachers who were men and earned less than $20,000 per academic year were more likely to leave teaching (see Table 2). In addition, Model 3 suggested that teachers who taught at a private school were more likely to leave teaching (see Table 2). There were no factors in Model 3 that were associated with the risk that a beginning English or language arts teacher would migrate to a different school.
As did Models 2 and 3, Model 4 (Table 2) suggested that attrition may be predictable only by gender and salary in that English or language arts teachers who were men and earned less than $20,000 per academic year were more likely to leave teaching as opposed to remaining in the profession. As with the previous models, there were no variables associated with the risk that a beginning English or language arts teacher would migrate to a different school after his or her 1st year of teaching (see Table 3). In Models 5 and 6, as reflected in all previous models, English or language arts teachers who were male and earned less than $20,000 per academic year were more likely to leave teaching (see Table 2). The only additional factors in Models 5 and 6 suggesting a relationship to either attrition or mobility were in Model 5. In Model 5, teachers who taught in schools in which the percentage of students eligible for free and reduced lunch was greater than 20% were more likely to leave teaching or change schools (see Table 2). As in Models 2-6, there were no mentoring or induction activities that were statistically significant predictors of teacher attrition or migration. The full results for the multinomial logistic regression analyses for Models 3-6 are shown in Tables 6-9.
Model 7 estimated the impact of all the mentoring and induction activities concurrently on either beginning English or language arts teacher attrition or migration. Controlling for other variables, we found that the odds of an English or language arts teacher leaving the profession were higher for men who earned less than $20,000; those who taught in a private school; and those who taught in a school in which 20% or more of the students were eligible for free or reduced lunch (see Table 2). In comparison, there were fewer factors, indeed only one statistically significant factor, that affected teachers’ decisions to migrate to a different school as compared with remaining at the same school. The odds of a teacher migrating to a different school were greater for those who taught in a high-poverty school (i.e., 20% or more of the students were eligible for free or reduced lunch; see Table 3). The full results for the multinomial logistic regression analyses for Model 7 are shown in Table 10.
In their study, Smith and Ingersoll (2004) noted the following:
The attenuation of the size of the coefficients when modeled simultaneously, making a number of them statistically insignificant, suggests that teachers who participated in at least one of the programs were likely to have participated in others, making it difficult to isolate an individual effect. However, the fact that the impact of a number of these activities was not strong enough individually to be statistically significant does not necessarily mean that they are of no value as components in a comprehensive induction program. (p. 704) The results of these models are similar to those of Smith and Ingersoll and suggest that an examination of a combination of programs and their associated effects (rather than each in isolation) is necessary. Thus, we conducted additional analysis (Model 8).
Model 8 is a reduced model of teacher and school characteristics that also includes an examination of the combination of mentoring and induction programs and their associated effects (rather than each in isolation). Although we do not present it here, we conducted an expanded model that included all previously examined teacher and school characteristics, and it showed similar results but indicated a lack of model fit. Thus, we developed a more parsimonious model. We do present the results of this reduced model. In it, we included the number of mentoring or induction activities and teacher and school characteristics to determine the relationship to attrition, mobility, or retention. The adjusted Wald test, a measure of overall model evaluation, indicated a good overall model. Controlling for other variables, we found that the odds of an English or language arts teacher leaving the profession were higher only for teachers who earned less than $20,000 (see Table 2). Beginning English teachers who earned less than $20,000 were more than eight times more likely to leave teaching as opposed to remaining in the profession. There were no factors that were statistically significantly related to teachers’ decisions to migrate to a different school as compared with remaining at the same school (see Table 3). When controlling for teacher and school characteristics, we found that the results did not provide evidence to suggest that the number of mentoring or induction activities in which teachers participated is related to decreased attrition or migration. The full results for the multinomial logistic regression analyses for Model 8 are shown in Table 11.
Conclusion and Discussion
Although other researchers have examined teachers’ attrition and mobility, and some of those researchers have controlled for content areas such as mathematics and science (e.g., Smith & Ingersoll, 2004), the review of beginning English teachers independently of all other content areas is important because of the high-stakes reading and writing assessments that are in place as accountability measures and the resulting emphasis placed on increasing student test scores by administrators as well as external sources (e.g., media) on English or language arts teachers. A better understanding of the factors related to beginning English teachers’ attrition and mobility is important for their employers and their preservice education programs. Teachers’ attrition and migration exacerbates teacher shortages and results in poor return on investments for those schools that experience these conditions (Haselkorn & Fideler, 1999). A number of factors, such as schools offering competitive salaries, enhance teacher retention (Hare & Heap, 2001; Imazeki, 2005). Schools or districts can also provide a number of programmatic offerings to increase teacher retention. Researchers have recommended administrators’ encouragement of teachers’ participation in mentoring and induction activities as one strategy to decrease teacher turnover (Colley, 2002; Darling-Hammond, 2003; Feiman-Nemser, 1996; Gold, 1996). Other factors, such as gender (Imazeki, 2005; Stinebrickner, 1998) and school sector (i.e., public vs. private; Luekens et al., 2004), are related to attrition and migration but are beyond the schools’ ability to change.
Building on Smith and Ingersoll’s (2004) research, models that examined teacher characteristics (gender and salary), school characteristics (sector and percentage of students on free or reduced lunch), and the number of mentoring and induction activities in which teachers participated indicated that ultimately only salary was statistically significantly related to increased odds of beginning English teachers’ leaving the profession, and no teacher or school characteristics were associated with increased or decreased migration. In reviewing the mentoring and induction factors in the present study, we found that none provided evidence of either increasing or decreasing the odds of teachers’ leaving or changing schools. Comparing these results with those of Smith and Ingersoll, we found that salary was not a factor for either attrition or migration in those researchers’ examination of beginning teachers from all content areas. Smith and Ingersoll also found a number of mentoring and induction activities that were related to decreased attrition and migration, including having a mentor, receiving common planning time, and being part of a network. Smith and Ingersoll found that a reduced teaching schedule increased attrition and migration. In addition, and also in contrast to the present findings, Smith and Ingersoll found that participation in greater numbers of activities decreased the odds of attrition and migration. The present findings concur with neither Smith and Ingersoll’s findings nor those of retention researchers in general, which have suggested that a solid, well-planned mentoring and induction experience is integral to lowering attrition rates (Cochran-Smith, 2004; Darling-Hammond et al., 1999; McCann et al., 2005; Rogers & Babinski, 2002; Smith & Ingersoll, 2004). In the present study, teachers’ participation in mentoring or induction activities did not appear to decrease the odds of leaving or changing schools for beginning English teachers.
There are several reasons mentoring and induction activities may not have been related to attrition or mobility of beginning English teachers. Perhaps beginning English teachers participated in various forms of professional development that were not specific to novice teachers, and these activities provided them with the tools that were necessary to navigate their early teaching careers beyond what they received or would have received by being mentored or by participating in other induction activities. In addition, as compared with teachers in other content areas, teachers assigned to English or language arts teaching positions perhaps received more mentoring and induction activities as support during their preservice programs, and thus participation in these same activities during their 1st year of teaching did little to affect their decisions to stay or leave the profession. It is also important for researchers to note that the data in the present study, although the most recently available, were administered before NCLB (NCLB Act of 2001, 2002). Consequently, there may have been less emphasis on accountability for English or language arts teachers. In turn, lesser accountability may have made individuals’ entry into teaching less stressful and the significance of mentoring and induction activities less of a factor for their retention in the profession. These reasons suggest that there are still a number of areas in relation to beginning English teachers’ retention, attrition, and mobility that need exploration.
Regardless, according to our knowledge, the present study is (a) the first study that has used data from the SASS and the TFS to examine beginning English teachers and (b) one of the few quantitative studies that have examined characteristics of and factors that may predict beginning English teachers’ attrition, retention, and mobility. Although a rich resource, the data collected from these surveys do not indicate the programmatic elements of the mentoring or induction activities (or clear operational definitions of the various mentoring activities), whether there were different types of mentoring in which the teacher participated (and if so, what those entailed), or the length of participation in them. Likewise, there are no data that will allow researchers to determine whether the mentoring and induction activities in which teachers participated were voluntary or required. In addition, the TFS provides only a 1-year follow-up on respondents. Researchers’ explaining retention by using a 1-year window may be assuming and inferring too much. Also, teachers may change schools for a variety of reasons other than those included in the models that we tested, such as reaching the end of short-term contracts or having their contracts terminated. We did not examine such factors, but we do suggest analyses that review other reasons for teacher migration. Additional research, specifically research that combines quantitative and qualitative methods, to explore these areas would add insight to the findings of the present study. We also suggest that future researchers compare the 1999-2000 SASS and 2000-2001 TFS data (National Center for Education Statistics, 2005) with data from the post-NCLB implementation (2003-2004, 2004-2005) to determine whether changes in educational policy affected teachers’ attrition and retention rates.
Despite the limited reasons that we suggested for teacher mobility and attrition, some additional points deserve discussion. First, in terms of the profile of novice teachers, although diversifying the teaching profession has been a strategic goal, our analysis showed that nearly 90% of English teachers were White. That proportion is actually worse than national averages that Darling- Hammond et al. (1999) reported. Second, teachers’ salaries were extremely low, lower than other salaries that we knew about during that same time period. The correlation between salary and attrition is no coincidence.
Nevertheless, the aforementioned limits of our findings and the fact that recent SASS and TFS data have not been released publicly for researcher use make it difficult not only to compare with current conditions but also to make absolute claims. Even so, the present analysis provides the profession with valuable insights. This type of research can inform teacher education, as part of a “diverse, coherent, systematic agenda for research, scholarship, and knowledge that can be used to establish positions beyond mere belief from which professionals can take well-informed public action” (Burns, 2007, p. 132). Furthermore, the present work adds to the knowledge base and can inform the profession as a whole. As Cochran- Smith (2004) maintained, We face enormous challenges as we rethink teacher recruitment, preparation, and retention. There are many new role and partnership possibilities for the universities, professional organizations, school districts, and communities with the vision to imagine them and the will to implement them. (p. 391)
AM Statistical Software. 2008. Free statistical software from the American Institutes for Research. Retrieved July 23, 2008, from http://am.air.org
Andrews, S. P., & Martin, E. (2003). No teacher left behind: Mentoring and supporting novice teachers. Paper presented at the Georgia Association of Colleges for Teacher Education/Georgia Association of Teacher Educators, St. Simons Island, GA.
Bullough, R. V. (1989). First-year teacher: A case study. New York: Teachers College Press.
Burns, L. D. (2007). On being unreasonable: NCTE, CEE, and political action. English Education, 39(2), 120-145.
California Commission on Teacher Credentialing and California Department of Education. (1992). Success for beginning teachers: The California New Teacher Project. Sacramento, CA: California Commission on Teacher Credentialing.
Chubbuck, S. M., Clift, R. T., Allard, J., & Quinlan, J. (2001). Playing it safe as a novice teacher: Implications for programs for new teachers. Journal of Teacher Education, 52(5), 365-376.
Cochran-Smith, C. (2004). Stayers, leavers, lovers, and dreamers: Insights about teacher retention. Journal of Teacher Education, 55(5), 387-392.
Colley, A. C. (2002). What can principals do about new teacher attrition? Principal, 81(4), 22-24.
Curran, B., & Goldrick, L. (2002). Mentoring and supporting new teachers. Issues brief. Washington, DC: National Governors’ Association Center for Best Practices.
Darling-Hammond, L. (2003). Keeping good teachers: Why it matters, what leaders can do. Educational Leadership, 60(8), 6-13.
Darling-Hammond, L., Berry, B. T., Haselkorn, D., & Fideler, E. (1999). Teacher recruitment, selection, and induction: Policy influences on the supply and quality of teachers. In L. Darling- Hammond & G. Sykes (Eds.), Teaching as the learning profession: Handbook of policy and practice (pp. 183-232). San Francisco: Jossey- Bass.
DeWert, M. H., Babinski, L. M., & Jones, B. D. (2003). Safe passages: Providing online support to beginning teachers. Journal of Teacher Education, 54(4), 311-320.
Feiman-Nemser, S. (1996). Teacher mentoring: A critical review. Washington, DC: Office of Educational Research and Improvement.
Gold, Y. (1996). Beginning teacher support: Attrition, mentoring, and induction. In J. Sikula, T. J. Butterly, & E. Guyton (Eds.), Handbook of research on teacher education (2nd ed., pp. 548-594). New York: Macmillan.
Hare, D., & Heap, J. L. (2001). Effective teacher recruitment and retention strategies in the Midwest: Who is making use of them? Naperville, IL: North Central Regional Educational Laboratory.
Hargreaves, A., & Fullan, M. (2000). Mentoring in the new millenium. Theory into Practice, 39(1), 50-56.
Haselkorn, D., & Fideler, E. (1999). Learning the ropes: Urban teacher induction programs and practies in the United States. Retrieved June 8, 2006, from http://www.rnt.org/publications/ ropes.html
Hirsch, E. (2006). Recruiting and retaining teachers in Alabama: Educators on what it will take to staff all classrooms with quality teachers. Chapel Hill, NC: Center for Teaching Quality.
Holloway, J. H. (2001). Who is teaching our children? Educational Leadership, 58, 1-3.
Imazeki, J. (2005). Teacher salaries and teacher attrition. Economics of Education Review, 24(4), 431-449.
Ingersoll, R. M. (2003). Is there really a teacher shortage? A report co-sponsored by the Center for the Study of Teaching and Policy and the Center for Policy Research in Education. Seattle: University of Washington Center for the Study of Teaching and Policy.
Ingersoll, R. M. (2004). Why do high-poverty schools have difficulty staffing their classrooms with qualified teachers? Washington, DC: Center for American Progress and the Institute for America’s Future.
Johnson, S. M., & Birkeland, S. E. (2004). Pursuing a “sense of success”: New teachers explain their career decisions. American Educational Research Journal, 40(3), 581-617.
Kaplan, J., Scherff, L., & Faulkner, L. (2005, November). First year teachers: Triumphs, tragedies, and realities. Paper presented at the annual meeting of the National Council of Teachers of English, Pittsburgh, PA.
Kent, S. I. (2000). Problems of beginning teachers: Comparing graduates of bachelor’s and master’s level teacher preparation programs. The Teacher Educator, 35(4), 83-96.
Liu, Y., & Meyer, J. P. (2005). Teachers’ perceptions of their jobs: A multilevel analysis of the Teacher Follow-Up Survey for 1994- 95. Teachers College Record, 107(5), 985-1003.
Luekens, M. T., Lyter, D. M., & Fox, E. E. (2004). Teacher attrition and mobility: Results from the Teacher Follow-Up Survey, 2000-01 (No. NCES 2004-301). Washington, DC: U.S. Government Printing Office.
McCann, T. M., Johannessen, L. R., & Ricca, B. P. (2005). Supporting beginning English teachers: Research and implications for teacher induction. Urbana, IL: National Council of Teachers of English.
National Center for Education Statistics. (2005). 1999-2000 Schools and Staffing Survey (SASS) and 2000-01 Teacher Follow-up Survey (TFS) CD-ROM: Public-use data with electronic codebook (No. NCES 2005335). Washington, DC: Author.
No Child Left Behind Act of 2001. (2002). Public law 107-110.
Odell, S. (1986). Induction support of new teachers: A functional approach. Journal of Teacher Education, 37(1), 26-29.
Recruiting New Teachers. (2000). A guide to developing teacher induction programs. Belmont, MA: Recruiting New Teachers.
Rippon, J. H., & Martin, M. (2006). What makes a good induction supporter? Teaching and Teacher Education: An International Journal of Research and Studies, 22(1), 84-99.
Rogers, D. L., & Babinski, L. M. (2002). From isolation to conversation: Supporting new teachers’ development. Albany: State University of New York Press.
Scherff, L. (2006). Reality check: A teacher educator returns home. Studying Teacher Education, 2(2), 155-167.
Smagorinsky, P., Cook, L. S., Moore, C., Jackson, A. Y., & Fry, P. G. (2004). Tensions in learning to teach: Accommodation and the development of a teaching identity. Journal of Teacher Education, 55(1), 8-24.
Smith, T. M., & Ingersoll, R. M. (2004). What are the effects of induction and mentoring on beginning teacher turnover? American Educational Research Journal, 41, 681-714.
Stinebrickner, T. R. (1998). An empirical investigation of teacher attrition. Economics of Education Review, 17(2), 127-136.
Tourkin, S. C., Pugh, K. W., Fondelier, S. E., Parmer, R. J., Cole, C., Jackson, B., et al. (2004). 1999-2000 Schools and Staffing Survey (SASS) data file user’s manual (No. NCES 2004-303). Washington, DC: U.S. Department of Education.
DEBBIE L. HAHS-VAUGHN
University of Central Florida
University of Alabama
Address correspondence to: Debbie L. Hahs-Vaughn, University of Central Florida, PO Box 161250, Orlando, FL 32816-1250, USA. E- mail: firstname.lastname@example.org
Copyright Heldref Publications Fall 2008
(c) 2008 Journal of Experimental Education, The. Provided by ProQuest LLC. All rights Reserved.