Predicting Special and General Educators’ Intent to Continue Teaching Using Conservation of Resources Theory

Abstract

Retaining teachers is an important priority for school leaders, especially in special education, a field with chronic shortages. We analyzed a nationally representative survey using conservation of resources theory to examine how job demands and resources interacted with one another and with teachers’ assignments (i.e., as special and general educators) to predict intentions to continue teaching. We found that teachers were more inclined to stay in their schools when they experienced stronger school administrative and collegial support, had more access to instructional materials, and were more experienced. Teachers indicated weaker intent to stay when they experienced more problems with students, spent more time working, were less experienced, and served in higher-poverty schools. School administrative support moderated relationships between intent and (a) school poverty and (b) experience. We found no differences between special and general educators’ intent to stay, though we did find differences in the conditions predicting special versus general educators’ intent.

Teacher attrition is academically and financially costly. Studies have consistently found that attrition negatively predicts student achievement (Ronfeldt, Loeb, & Wyckoff, 2013) and money invested in preparing, hiring, and inducting teachers is lost when they leave (Milanowski & Odden, 2007). These costs disproportionately fall on high-poverty districts, which spend more than twice as much, per school, per year, to replace teachers as low-poverty districts (Barnes, Crowe, & Schafer, 2007). Special education teacher (SET) attrition is especially concerning, given chronic, severe, and growing SET shortages (Billingsley & Bettini, 2017; McLeskey & Billingsley, 2008). In this context, retaining SETs is essential to ensure that students have equitable access to experienced SETs (Billingsley & Bettini, 2017).

Research consistently finds that working conditions are an important predictor of attrition and intent to leave (Billingsley & Bettini, 2019; we use “attrition” to include all forms of attrition, e.g., leaving the profession, moving schools, or transferring educational fields). For example, Johnson, Kraft, and Papay (2012) examined Massachusetts data and found that working conditions predicted teachers’ decisions to stay in or leave their schools (for reviews on general education teacher [GET] attrition, see Achinstein, Ogawa, Sexton, & Freitas, 2010; Borman & Dowling, 2008; Guarino, Santibañez, & Daley, 2006; Simon & Johnson, 2015).

Research consistently finds that working conditions are an important predictor of attrition and intent to leave.

Similarly, in special education, researchers have found that working conditions support SETs’ intent to stay (often used as a proxy for attrition; Billingsley & Bettini, 2019). For example, Conley and You (2017) analyzed the nationally representative Schools and Staffing Survey Teacher Questionnaire (SASS-TQ; 2007–2008) and found that secondary SETs’ ratings of administrative support and team efficacy predicted intent to stay. These findings are consistent with a strong body of prior research examining predictors of SETs’ intentions to stay (e.g., Cross & Billingsley, 1994; Gersten, Keating, Yovanoff, & Harniss, 2001; Miller, Brownell, & Smith, 1999; Singh & Billingsley, 1996). These studies are robust, reflecting different samples, measures, and model specifications, though many are now dated, and few relied on strong theoretical foundations (for a review of SET attrition research, see Billingsley, 2004; Billingsley & Bettini, 2019).

Only a few studies have examined whether working conditions predict SETs’ intent to stay differently from how they predict GETs’ intent (e.g., Billingsley & Cross, 1992; Bettini, Jones, et al., 2017; Gilmour & Wehby, 2019; Jones, Youngs, & Frank, 2013). Furthermore, these studies all relied on state or regional data sets that do not generalize nationally. Understanding how working conditions differentially predict intent among SETs versus GETs is important for differentiating supports. Furthermore, some working conditions (e.g., content areas, instructional materials) have received scant attention in prior quantitative studies, despite qualitative evidence that they may contribute to intent (Billingsley & Bettini, 2019). As such, the purpose of this study is to examine how working conditions interact with one another and with teachers’ assignments (SET vs. GET) to predict intent to stay, using nationally representative SASS-TQ data.

Theoretical Framework: Conservation of Resources Theory

We use conservation of resources (COR) theory to conceptualize how working conditions interact with one another to predict intent to stay in one’s school. COR is a motivational theory, which posits that individuals strategically draw on limited resources to fulfill job demands (Alarcon, 2011; Halbesleben, Neveu, Paustian-Underdahl, & Westman, 2014). Resources include anything that employees feel helps them to successfully fulfill job demands, including social supports (e.g., from administrators, colleagues), material resources (e.g., instructional materials), and internal resources (e.g., knowledge, skill; Halbesleben et al., 2014). When demands and resources are balanced, employees feel able to manage responsibilities, feel motivated by their job, and thus are more likely to intend to stay. In contrast, high demands and low resources may reduce motivation and thereby increase intent to leave. Meta-analyses support the utility of COR theory for predicting employees’ commitment and retention (Alarcon, 2011; Halbesleben et al., 2014).

When demands and resources are balanced, employees feel able to manage responsibilities, feel motivated by their job, and thus are more likely to intend to stay.

In a seminal study of why teachers leave, Johnson and Birkeland (2003, p. 581) proposed that teachers are motivated to leave when they do not experience “a sense of success” and particularly when conditions of their work do not support their capacity to effectively serve students. They did not reference COR theory, but their conception of how the conditions shaping effectiveness may also shape motivation to stay in teaching is consistent with COR theory. Subsequent studies of teachers have relied on COR theory, obtaining results consistent with its major tenets (e.g., Bettini et al., in press; Bettini, Jones, et al., 2017). In the following sections, we first describe research on demands and resources that teachers report influence their capacity to effectively serve students and that are associated with their intent or attrition; we then describe evidence that the importance of these conditions may vary for SETs versus GETs.

Demands

Prior studies indicate that teachers intend to leave when they experience higher demands, including more instructional responsibilities and more student problems (note that although some of these studies used strong methods, none used methods that permit causal inferences).

Instructional responsibilities

Providing instruction is teachers’ core demand, and SETs report struggling to plan and deliver instruction in more subjects to more grades (e.g., Bettini, Wang, Cumming, Kimerling, & Schutz, 2018). Yet, no quantitative research has examined whether more instructional responsibilities predict intent (Billingsley & Bettini, 2019).

Student problems

Students themselves present teachers with challenges, and teachers often report that student problems challenge their efforts to teach (National Center for Education Statistics, 2015). Teachers who report that students have more substantial behavior problems (e.g., absenteeism, tardies, disruptive behavior) are significantly less likely to intend to stay or to actually remain (e.g., Boyd et al., 2011; Conley & You, 2017; Feng, 2009; Gilmour & Wehby, 2019). For example, analyzing Florida’s data set, Feng (2009) found that teachers were more likely to leave when they were assigned to teach students who had more disciplinary incidents. Working in a school with more behavior challenges may also make other forms of assistance more important for ensuring that teachers have support to manage challenging behavior (Bettini, Cumming, Merrill, Brunsting, & Liaupsin, 2017; Gilmour & Wehby, 2019).

School poverty

Prior studies have consistently found higher attrition in high-poverty schools (e.g., Borman & Dowling, 2008; Carver-Thomas & Darling-Hammond, 2017). Some scholars posit that this may indicate that teachers are less willing to teach students living in poverty or that students living in poverty may be more challenging to teach (e.g., Hanushek, Kain, & Rivkin, 2004). However, subsequent research has refuted this possibility, finding that high attrition from high-poverty schools is largely accounted for by disparities in supportive working conditions across high- and low-poverty schools (Johnson et al., 2012; Simon & Johnson, 2015). These studies have documented a range of disparities, including more demanding workloads (Fall & Billingsley, 2011), reduced autonomy (Diamond & Spillane, 2004), and less stable leadership (Beteille, Kalogrides, & Loeb, 2012; Grissom, 2011). Thus, school poverty may capture other unmeasured demands on teachers (Simon & Johnson, 2015).

Resources

Teachers have resources that help them fulfill their responsibilities and that may thereby improve their desire to stay. These include social resources, physical resources, and internal resources (Billingsley & Bettini, 2019; Bettini et al., in press; as before, note that, although some of these studies used strong methods, none used methods that permit causal inferences).

Social resources

Social resources, including collegial and administrative support, help teachers meet the demands of their jobs, by providing opportunities to learn about curriculum, instruction, and other resources (e.g., Grossman & Thompson, 2008; C. K. Jackson & Bruegmann, 2009). Early career teachers, in particular, report that collegial and administrative relationships are essential for their capacity to fulfill their responsibilities (Grossman & Thompson, 2004; Johnson & Birkeland, 2003); colleagues often provide early career teachers with clear expectations about what they should be doing (Youngs, Holdgreve-Rezendez, & Qian, 2011). Additionally, teachers who experience stronger administrative and collegial support are more likely to stay or intend to stay (e.g., Boyd et al., 2011; Conley & You, 2017).

Social resources are not independent of one another (e.g., Bettini, Jones, Brownell, Conroy, & Leite, 2018; Gersten et al., 2001). Administrators foster collegial interactions among teachers (Gersten et al., 2001; Singh & Billingsley, 1996), and teachers’ reliance on each social resource may vary depending on the strength of others. For example, teachers may rely more on administrators when collegial support is weak (Billingsley & Bettini, 2019; Youngs, Jones, & Low, 2011).

Physical resources

Physical resources, especially instructional materials, can support teachers to learn and enact pedagogical content knowledge (Grossman & Thompson, 2008; K. Jackson & Makarin, 2016), and teachers report struggling to determine what and how to teach when they have insufficient instructional materials (e.g., Grossman & Thompson, 2008; Otis-Wilborn, Winn, Griffin, & Kilgore, 2005; Siuty, Leko, & Knackstedt, 2016). Teachers with stronger instructional materials feel better able to meet demands (e.g., Johnson & Birkeland, 2003), and having adequate resources may be associated with intent to stay (Billingsley & Bettini, 2019).

Internal resources

Teachers’ internal resources—knowledge and skills—may support them to manage job demands (Boe, Shin, & Cook, 2007; Brownell et al., 2014; Ronfeldt, Schwartz, & Jacob, 2014). Measuring teachers’ knowledge has proven challenging (Brownell et al., 2014), but qualifications (i.e., experience, certification) often serve as proxies for knowledge (Billingsley & Bettini, 2017). Scholars have found experience is resource that supports teachers to provide effective instruction (e.g., Henry, Bastian, & Fortner, 2011; Ladd & Sorenson, 2014) and, until they reach retirement, to continue teaching (e.g., Boyd et al., 2011; Connelly & Graham, 2009). Certification is an indicator of whether teachers have demonstrated, through an exam or preparation program, adequate knowledge and skill, and some scholars have found that certified SETs provide more effective instruction (Feng & Sass, 2013) and are more likely to intend to stay (Conley & You, 2017; though studies of GETs generally do not find effects of certification except in math; Wayne & Youngs, 2003).

Differential Effects on Teachers in Special vs. General Education

Several studies suggest that schools’ social resources may shape SETs’ experiences differently from how they shape GETs’ experiences, likely because of substantial differences in the demands of their roles in schools. Jones et al. (2013) analyzed surveys of teachers in their first 3 years in 11 urban districts. They found that perceptions of collegial support significantly predicted SETs’ intent but not GETs’ intent. In another analysis of the same data set, Bettini, Jones, et al. (2018) found that beginning SETs felt less overwhelmed when they had frequent instructional interactions with colleagues, whereas GETs felt more overwhelmed when they had frequent instructional interactions with colleagues.

Purpose and Hypotheses

Prior studies indicate that SETs likely respond to resources and demands differently than GETs, but prior studies’ samples were limited to specific regions and did not include the full range of resources and demands likely related to intent. Thus, the purpose of this study is to extend prior research to examine how teachers’ resources and demands interact with one another and with teachers’ assignments (SET vs. GET) to predict intent to continue teaching. We used regression to test the following hypotheses, derived from COR theory and from extant research:

Hypothesis 1: SETs will express weaker intent to stay than GETs.

Hypothesis 2: Teachers who experience heavier demands (i.e., number of grade levels served, number of subjects taught, student behavior problems, administrative duties, school poverty) will express weaker intent to stay.

Hypothesis 3: Teachers who experience stronger social resources (i.e., collegial support, administrative support), physical resources (i.e., materials), and internal resources (i.e., experience teaching, qualifications) will express stronger intent to stay.

Hypothesis 4: Teachers’ experience, school poverty, and student problems will interact with resources such that resources will predict intent more strongly for less experienced teachers, teachers in higher-poverty schools, and teachers who report more student problems.

Hypothesis 5: Administrative support will predict intent to stay more strongly among teachers who report weaker collegial support.

Hypothesis 6: Resources and demands will predict intent more strongly among SETs than GETs.

Methods

Sample

We tested hypotheses using the National Center for Education Statistics’ 2011–2012 SASS-TQ. We included all full-time teachers who taught elementary, special education, English language arts, math, science, social studies, or English as a second language, excluding physical education, music, art, and other specials teachers, as these roles may not be comparable. We also excluded charter school teachers. We used SASS-TQ sample weights, resulting in 2,536,540 weighted cases (25,460 unweighted): 385,470 SETs and 2,067,250 GETs.

Table 1 displays characteristics of the weighted full sample, the weighted sample of SETs, and the weighted sample of GETs. Teachers in our analytic sample averaged 13.6 years’ experience (20.8% were early career teachers with <5 years’ experience, 64.7% were midcareer teachers with 5 to 24 years’ experience, and 14.5% were late career teachers with ≥25 years’ experience), and 92% were fully certified. Nearly 16% were SETs. SETs and GETs did not differ substantively in their years’ experience or in their reports of many resource variables. In the full sample, teachers taught an average of two grade levels, ranging from a single grade to 15 (including prekindergarten and ungraded). SETs taught more grade levels than GETs (3.26 vs. 1.83). GETs and SETs reported moderate overall levels of student behavior problems. Teachers reported working an average of 52.4 hours/week, with SETs reporting that they worked about 2 hours/week less than GETs. Fewer SETs were fully certified than GETs (88.7% of SETs vs. 92.7% of GETs). Teachers’ schools served an average of 50% students qualifying for free or reduced lunch (FRL). GETs more frequently worked in elementary schools than SETs.

Table 1.

Sample Characteristics.

	Full sample			Special education teachers			General education teachers
Variable	Mean or %	SD	Range	Mean or %	SD	Range	Mean or %	SD	Range
Years of experience	13.61	9.19	1–51	13.11	9.30	1–47	13.70	9.16	1–51
Full certification	92.03			88.69			92.66
Administrative support	8.75	2.92	0–12	8.95	2.85	0–12	8.71	2.93	0–12
Collegial support	6.25	1.98	0–9	6.20	1.94	0–9	6.26	1.99	0–9
Instructional materials	2.11	0.88	0–3	2.08	0.87	0–3	2.11	0.89	0–3
Grades taught	2.05	1.49	1–15	3.26	1.67	1–15	1.83	1.34	1–15
Subjects taught	3.65	0.82	0–4	2.96	1.15	0–4	3.75	0.66	0–4
Student problems	6.30	3.64	0–16	6.31	3.70	0–16	6.30	3.62	0–16
Hours of work	52.39	8.91	30–80	50.61	8.47	30–80	52.72	8.95	30–80
FRL at school, %	50.37	28.59	0–100	51.21	28.47	0–100	50.22	28.61	0–100
Intent	6.70	2.21	0–9	6.84	2.10	0–9	6.67	2.22	0–9
Elementary	66.67			61.72			67.60
Secondary	25.96			28.61			25.46
Combined	7.37			9.67			6.94
n (weighted)	2,452,720			385,470			2,067,250

Note. FRL = free and reduced-price lunch.

Measures

Teacher assignment

We categorized teachers as SETs if they reported that their main assignment was special education (see Table 2 for all measures).

Table 2.

Measures.

Measure	Items	Response options
Teacher assignment	This school year, what is your main teaching assignment field at this school? (T0090)	110 indicates the teachers’ main assignment is special education
Demands
Number of grades	Do you currently teach students in any of these grades at THIS school? Please mark (X) Yes or No for each grade level. (T0070–T0084)	For each grade level (preK–12), participants can indicate Yes or No
Number of subjects	During your most recent FULL WEEK of teaching, approximately how many hours did YOU spend teaching each of the following subjects at THIS school? (T0096, T0098, T0099, T0100, T0110, T0111, T0112, T0113, T0114, T0115, T0116, T0117, T0118, T0119)	For each subject area (English/language arts, math, social students, or science, and subfields), participants indicate None or a two-digit number. We dummy coded each subject they taught summed these to obtain a total number of subjects taught.
Hours worked	Including hours spent during the school day, before and after school, and on the weekends, how many hours do you spend on ALL teaching and other school-related activities during a typical FULL WEEK at THIS school? (T0392)	Teacher reports total weekly hours spent on all teaching- and school-related activities.
Student problems	To what extent do you agree or disagree with each of the following statements? The level of student misbehavior in this school (such as noise, horseplay or fighting in the halls, cafeteria, or student lounge) interferes with my teaching. (T0437)	Strongly agree, somewhat agree, somewhat disagree, strongly disagree
	Item 64: To what extent is each of the following a problem in this school?	Serious problem, moderate problem, minor problem, not a problem
	Item a: Student tardiness (T0455)
	Item b: Student absenteeism (T0456)
	Item c: Student class cutting (T0457)
	Item f: Student apathy (T0460)
	Item 67c: In the past 12 months, how many times has a student FROM THIS SCHOOL threatened to injure you? (T0475)	Yes or No
School poverty	Percentage of K–12 enrollment approved for free or reduced-price lunches (NSLAPP_S)	Continuous variable reported on the school survey
Social resources
Administrative support	To what extent do you agree or disagree with each of the following statements?	Strongly agree, somewhat agree, somewhat disagree, strongly disagree
	The school administration’s behavior toward the staff is supportive and encouraging. (T0435)
	My principal enforces school rules for student conduct and backs me up when I need it. (T0441)
	The principal knows what kind of school he or she wants and has communicated it to the staff. (T0444)
	In this school, staff members are recognized for a job well done. (T0446)
Collegial support	To what extent do you agree or disagree with each of the following statements?	Strongly agree, somewhat agree, somewhat disagree, strongly disagree
	Rules for student behavior are consistently enforced by teachers in this school, even for students who are not in their classes. (T0442)
	Most of my colleagues share my beliefs and values about what the central mission of the school should be. (T0443)
	There is a great deal of cooperative effort among the staff members. (T0445)
Physical resource
Instructional materials	To what extent do you agree or disagree with each of the following statements?	Strongly agree, somewhat agree, somewhat disagree, strongly disagree
	Necessary materials such as textbooks, supplies, and copy machines are available as needed by the staff. (T0439)
Internal resources
Experience	Excluding time spent on maternity/paternity leave or sabbatical, how many school years have you worked as an elementary- or secondary-level teacher in public, public charter, or private schools? (T0042)	Participants can enter a two-digit number.
Full certification	Which of the following describes the teaching certificate you currently hold that certified you to teach in this state?	Options include varied descriptions of certification status
Control
School level	Three-category level of school based on grade levels offered as reported by the school (SHCLEVEL)	1 = elementary, 2 = secondary, 3 = combined
Intent	To what extent do you agree or disagree with each of the following statements?	Strongly agree, somewhat agree, somewhat disagree, strongly disagree
	I am generally satisfied with being a teacher at this school (T0451)
	The stress and disappointments involved in teaching at this school (T0465)
	I think about transferring to another school. (T0469)

Demands

We conceptualized five variables as demands. Instructional demands included the number of different grades and subjects taught and the hours/week that teachers reported working. We calculated the number of grades taught by dummy coding each grade and summing the number of grades. We calculated the number of subjects taught by dummy coding the four content areas (math, English language arts, social studies, science) and summing the number of subjects taught. Due to limitations with how SASS-TQ reports subject areas, we were limited to only examining these four subjects. We mean centered these variables for analyses but did not standardize variables. Results reflect the change in intent associated with a one grade or subject change.

Our conceptualization of demands also included student problems. We calculated a composite student problems score based on teachers’ responses to six items (Table 2). We calculated this composite score and all other composite scores with the weighted data. We recoded these variables so that higher values indicated more of a problem, and we summed the values of the six items. The internal consistency of the composite student problems variable was .78 in the full sample and in the GET and SET sample. We standardized this variable with a mean of 0 and a standard deviation of 1 to aid in the interpretation. Higher values indicate more student problems.

Finally, we examined the percentage of students receiving FRL (reported in the SASS-TQ school survey) at teachers’ schools as an indicator of school poverty. We did not standardize this variable; thus, the coefficients reported for this variable are interpreted as the association between a percentage-point change in FRL and intent.

Resources

We focused on social resources, physical resources, and internal resources. Social resources included school administrative support and collegial support. We constructed the school administrative support variable by combining teacher responses from four items (Table 2). Internal consistency was .85 in the full sample of teachers and the GET and SET subsamples. The second social resource was collegial support. Internal consistency was .75 in the full sample and the GET sample; internal consistency was .74 in the SET sample. We included a single indicator of physical resources. For each scale, higher values reflect stronger resources. We standardized all variables with a mean of 0 and a standard deviation of 1.

We examined two internal resources. First, we examined teachers’ years of experience. We converted this continuous variable to a categorical variable by defining early career teachers as teachers with <5 years of experience, midcareer teachers as teachers with ≥5 years of experience but <25 years of experience, and late career teachers as teachers with ≥25 years of experience. We dummy coded these categories and used midcareer teachers as the comparison group. Second, we examined if a teacher reported being fully certified. We dummy coded the variable so that 1 indicated fully certified and 0 indicated that the teacher was not fully certified.

Intent to stay/leave school

We examined teachers’ intent to stay in or leave their schools as the dependent variable. Like other recent researchers (e.g., Bettini, Jones, et al., 2017), we focused on intent to stay in one’s school, not intent to stay in the profession or in the field (e.g., special education), for several reasons. First, patterns of teacher movement between schools exacerbate inequities in access to skilled teachers, as more effective teachers are more likely to move from high- to low-poverty schools (e.g., Boyd, Grossman, Lankford, Loeb, & Wyckoff, 2008). Second, movement out of the profession is less common than movement between schools (Boe, Cook, & Sunderland, 2008; DeAngelis & Pressley, 2011). Third, teachers who plan to leave the profession also implicitly intend to leave their school; thus, this variable includes both teachers who intend to move to another school and teachers who intend to leave the profession entirely. Finally, because intent to stay in one’s school includes all possible destinations to which a teacher could be going, prior studies have found that it is a more sensitive measure than intent to stay in the profession (Johnson et al., 2012).

Note that intent to leave does not perfectly predict actual attrition. Researchers have found a moderate correlation between intent to leave the profession and actual attrition in the following 15 months (Gersten et al., 2001); note, however, that no researchers have tested the correlation between intent to leave one’s school and actual attrition, a limitation to this variable. However, we also concur with other scholars (e.g., Bettini, Jones, et al., 2017; Jones et al., 2013) that teachers’ intent is concerning even when they are not able to actually follow through on that intent by leaving, because intent to leave indicates disinvestment in schools, and it is strongly associated with burnout (e.g., Brunsting, Sreckovic, & Lane, 2014), which is in turn associated with weaker-quality instruction (Irvin, Hume, Boyd, McBee, & Odom, 2013; Ruble & McGrew, 2013; Wong, Ruble, Yu, & McGrew, 2017).

The intent scale includes three items (Table 2). Past research has found that these items load highly onto a single factor (Bettini, Mason-Williams, et al., 2018) and are distinct from related items in the SASS-TQ. We standardized this variable; higher value reflects stronger intent (a = .75 in the full sample, GET sample, and SET sample).

School level

We included dummy variables indicating school level (elementary, secondary, or combined grades; elementary is the comparison group). In elementary, teachers typically provide instruction in multiple subjects; in secondary, teachers specialize in a single subject. These differences could be associated with predictors (e.g., number of grades taught) and with intent, so we accounted for this by controlling for school level.

Analyses

We built a series of regression models to test hypotheses. In Model 1, we predicted intent as a function of teacher assignment and demands (Hypotheses 1 and 2). We added resources in Model 2 (Hypothesis 3). We then added interactions (Hypothesis 4) between resources and (a) experience (Model 3), (b) school poverty (Model 4), and (c) student problems (Model 5). In Model 6, we added an interaction between collegial and administrative support (Hypothesis 5). We then split the sample, by teacher assignment, into a sample of GETs and a sample of SETs, and fit the same series of models (Hypothesis 6). All models included school level as a control.

We analyzed data according to Institute of Education Sciences protocols, using Stata 14. We weighted data using the teacher final sampling weight variable and the SASS-TQ-supplied 88 replicate weight variables using the balanced repeated replication procedure as recommended by the institute. We rounded all n’s and degrees of freedom to the nearest 10 to ensure anonymity, per National Center for Education Statistics requirements.

Results

We examined pairwise correlations between the demands, resources, teacher assignment, school poverty, and teachers’ intent to stay (the full correlation table is available in Appendix B online). The social resources, school administrative support and collegial support, were moderately correlated (r = .59). Social resources were positively correlated with reported access to instructional materials (r = .29–.34). Collegial support and access to instructional materials had small positive correlations with internal resources (r = .02–.05), and experience had a small negative correlation with administrative support. Of the demands, the highest correlations were between the number of subjects and number of grades taught (r = −.23) and student problems and the number of grades taught (r = .16). Being a SET was positively correlated with the number of grades taught (r = .35) and negatively correlated with the number of subjects taught (r = −.36). School poverty was positively correlated with student problems (r = .21) and negatively correlated with the resource variables. Intent was positively correlated with the resource variables and negatively correlated with the demand variables.

Assignments and Demands

The first model included SET assignment and demands as predictors of intent. Contrary to our hypothesis, being a SET was not associated greater intent to leave (Appendix B). The strongest association in this model was between student problems and intent. A 1-SD increase in student problems was associated with a 0.45-SD decrease in intent (p < .001). Working more hours was negatively associated with intent (b = −0.006, p < .001); a 1-hour increase in hours worked was associated with a 0.006-SD decrease in intent. A teacher who reported an average number of hours working each week (52.39 hours) had an average level of intent of −0.05; a teacher reporting that she or he worked 1 SD more hours per week (about 9 hours) had an average level of intent of −0.10. After accounting for other demands and teachers’ assignment, a 1–percentage point increase in the percentage of students qualifying for FRL was associated with a 0.002-SD decrease in intent (p < .01). A teacher in a school in which none of the students qualified for FRL had an average level of intent of −0.05, whereas a teacher in which 100% of students qualified for FRL had an average level of intent of −0.25, a change in intent of about 9% of a standard deviation. An increase of one grade taught was associated with a 0.02 change in intent, and an increase in one subject taught was associated with a −0.01 change in intent; however, these small associations were not statistically significant. This model accounted for 18.5% of variance in intent.

Contrary to our hypothesis, being a SET was not associated greater intent to leave.

Differences between SETs and GETs

Appendix B also reports results of the model examining the association between demands and intent for the subsample of SETs and GETs. Student problems were consistently strong negative predictors of teachers’ intent for both samples (b = −0.36 for SETs and −0.47 for GETs, both p < .001). The number of hours spent working was negatively associated with intent for SETs and GETs (b = −0.010 for SETs and −0.006 for GETs). The number of grades that teachers instructed was not associated with SETs’ intent, whereas it was associated with GETs’ intent. Despite similar coefficients for SETs and GETs, the association between school FRL and intent was statistically significant only for GETs.

Teacher Resources

We then added resources to the model. In the full sample, the association between student problems and intent decreased in magnitude (Appendix B). The majority of resources were positively associated with intent. A 1-SD increase in teachers’ reported administrative support was associated with a 0.44-SD increase in intent, after accounting for other resources, demands, and teacher assignment (p < .001). Collegial support and instructional materials had weaker associations with intent. A 1-SD increase in collegial support or instructional materials was associated with a 0.08- or 0.07-SD increase, respectively, in intent (p < .001). Being an early career teacher was associated with a 0.07-SD decrease in intent (p < .05), and being a late career teacher was associated with a 0.14-SD increase in intent (p < .001). Full certification was associated with a small change in intent, −0.01 SD, that was not statistically significant. This model accounted for 40% of variation in intent.

Differences between SETs and GETs

Associations between some resources and intent to stay differed for SETs and GETs in statistical significance, despite similarly sized coefficients across the samples. Collegial support did not have a statistically significant association with SETs’ intent (b = 0.07, p > .05), but it did have a statistically significant association with GETs’ intent (b = 0.09, p < .001), despite similar coefficients. Early career SETs reported lower average intent than midcareer SETs (b = −0.03), but this difference was not statistically significant (p > .05), whereas early career GETs reported statistically significant lower intent than midcareer GETs (b = −0.08, p < .05). Access to instructional materials was associated with increases in GETs’ intent (b = 0.07, p < .001) but not SETs’ (b = 0.04, p > .05). In contrast, full certification was associated with a substantive and statistically significant increase in SETs’ intent but not GETs’, indicating that certification was associated with a 0.20-SD increase SETs’ intent to stay (p < .05).

Interactions

In the last four columns of Appendix B, we report the results of models that included interactions between teacher experience and resources (Model 3), school poverty and resources (Model 4), student problems and resources (Model 5), and administrative and collegial support (Model 6). In the full sample, we identified a statistically significant interaction between early career status and administrative support; administrative support was more strongly associated with intent to stay for early career teachers (b = 0.11, p < .05; see Figure 1). Other resources did not moderate the association between being an early career teacher and intent.

Figure 1.

Illustration of Statistically Significant Interactions for the Full Sample of Teachers

Administrative support moderated the association between school FRL and intent (b = 0.002, p < .01). As shown in Figure 1, the association between school FRL and intent decreased when teachers reported higher administrative support; in other words, when administrative support was strong, school FRL was a weaker predictor of intent, whereas when administrative support was weak, school FRL was a stronger predictor of intent. No other interactions between FRL and resources were statistically significant. Resources also did not moderate the association between student problems and intent. In the final column in Appendix B, we added an interaction between administrative and collegial support. We did not find a significant interaction between administrative and collegial support, after accounting for other variables in the model.

Differences between SETs and GETs

The last four columns in the bottom portions of Appendix B report the moderator results for SETs and GETs. Though the association between administrative support and early career SETs’ intent was stronger than that for GETs’ intent (b = 0.17 for SETs and b = 0.10 for GETs) as compared with midcareer teachers, this association was statistically significant only in the sample of GETs. The association between school FRL and intent was moderated by administrative support for both SETs and GETs (b = 0.002, p < .05). Across both samples, the other resources did not moderate the demands associated with experience, school FRL, or student problems.

Discussion

We examined how teachers’ resources and demands interacted with one another and with their assignments in schools (SET vs. GET) to predict their intent to remain in their schools, using nationally representative data. This is the first study to compare predictors of SETs’ and GETs’ intent using a nationally representative sample. Although prior research identified how demands (e.g., ratings of workload, paperwork) and resources (e.g., administrative and colleague support) relate to teachers’ intent, this study considers these conditions through the lens of COR theory, which supported us in hypothesizing how these conditions may interact with one another. Our results were largely consistent with COR theory: teachers’ demands and resources predicted their intent, and some resources moderated relationships between demands and intent. This study also incorporates a wider range of working conditions as predictors, including specific aspects of demands (e.g., subjects taught, instructional materials), thus allowing us to identify which working conditions were more or less important after controlling for one another.

Our results were largely consistent with COR theory: teachers’ demands and resources predicted their intent, and some resources moderated relationships between demands and intent.

Consistent with COR theory, in the full model, we found that teachers who reported experiencing stronger resources (i.e., administrative support, collegial support, instructional materials, and experience) had greater intentions to stay in their schools, whereas teachers who reported experiencing more demands (i.e., more time spent working, more student problems, and higher school poverty) had lower intentions to stay in their schools. Teachers’ perceptions of student problems were particularly powerful in predicting intent. Interestingly, although student problems were significant for both SETs and GETs, the coefficient for SETs was substantively smaller than for GETs, indicating that SETs’ intent may be less related to student problems than GETs’ intent.

Although prior research clearly identifies the importance of administrative support for retention (e.g., Billingsley & Cross 1992; Gersten et al., 2001; Johnson et al., 2012), this study extends earlier findings, demonstrating that administrative support was more important for early career teachers than for late career teachers. In addition, administrative support fully moderated the association between school FRL and intent such that there was no association between FRL and intent when teachers reported stronger administrative support, indicating that administrative support may be a key leverage point for increasing retention in high-poverty schools. These findings align with prior research (e.g., Johnson et al., 2012) and COR theory and provide insights into the circumstances under which administrative support is especially important.

Our findings generally aligned with COR theory, but some of our hypotheses about which demands and resources would be important for which teachers were not upheld. First, based on prior research showing that SETs rate their workloads less manageable (Bettini, Jones, et al., 2017) and have higher attrition rates (e.g., Billingsley & Bettini, 2019), we posited that SETs would be less likely to intend to stay than GETs, but this was not the case. There were no significant differences in SETs’ versus GETs’ intent, after controlling for other variables in the model. It is possible that teachers’ intent to stay in their school does not explain the differential attrition rates that prior studies found and that some other factor (e.g., involuntary turnover, family/life reasons for leaving) accounts for higher SET attrition. Another possibility is that other variables in our model (e.g., time spent working) might explain high SET attrition; however, if this possibility were accurate, we would expect descriptive statistics to show SETs reporting higher demands and weaker resources than GETs, and this was not the case.

Our findings regarding school FRL also differ slightly from prior research (e.g., Johnson et al., 2012), which has typically found that the relationship between school FRL and intent is attenuated or eliminated by accounting for disparities in working conditions (Simon & Johnson, 2015). Although administrative support moderated the effect of school FRL (i.e., predicting intent more strongly in high-poverty schools: Model 4), disparities in administrative support did not attenuate or eliminate the association between school FRL and intent; the effect of FRL in Model 1 did not decline or disappear when working conditions were added in Models 2 and 3.

In addition, the factors that we posited as being more strongly associated with SETs’ intent (e.g., collegial support) were in fact less strongly associated with SETs’ than GETs’ intent, and some factors that had been significantly related to intent in the full model (e.g., school poverty, early career status) were not significantly related to SETs’ intent, though many of the coefficients were substantively similar. The finding regarding collegial support was particularly surprising. Because SETs’ roles are often collaborative, we posited that collegial support would be more important for SETs than GETs; indeed, prior studies have found to be the case (e.g., Bettini, Jones, et al., 2018; Jones et al., 2013). However, prior studies did not control for administrative support, as we did in this analysis, and these two variables were correlated at .6. It is possible that because administrators set the tone for collegial interactions and support SETs’ collaborative relationships (e.g., Billingsley, McLeskey, & Crockett, 2017; Youngs, Jones, & Low, 2011), collegial support could partially mediate relationships between administrative support and intent, as some scholars found in the past (e.g., Singh & Billingsley, 1996). Additionally, prior studies’ participants were primarily early career SETs, and it could be that collegial support operates differently for beginning SETs than for more experienced SETs. Another possibility is that administrative support is simply much more important than collegial support for SETs; because GETs often fulfill the same roles as grade-level or subject matter colleagues, they may rely more on their colleagues than SETs who fulfill unique roles within schools. Another possibility is that the effects of collegial support may vary depending on SETs’ service delivery models (i.e., inclusive, self-contained, resource), explaining more variability in SETs’ intent in inclusive settings where they rely more on collaboration with colleagues. The SASS-TQ does not collect data on service delivery models, so we cannot test this possibility, but this may be an important area for future research.

Interestingly, certification was significantly associated with intent only among SETs, suggesting that SET certification may support SETs’ intent, whereas GETs’ certification may not support their intent. This coefficient was especially large for SETs, indicating that certification is an important predictor of SETs’ intent. It is possible that SET certification is a better indicator of internal resources (i.e., knowledge and skill to teach) than GET certification. Alternatively, it is possible that the demands of SETs’ jobs require more knowledge and skill and, thus, that certification more strongly predicts their intent. Another possibility is that certification may be a better proxy for commitment to the profession for SETs than for GETs (i.e., uncertified SETs may be less committed than uncertified GETs). Finally, another possibility is that GET certified teachers may experience a substantially different labor market than SET certified teachers; for example, they could have a wider range of attractive alternative positions (in or outside of teaching), resulting in an insignificant relationship between their certification and their retention.

We were also surprised by findings regarding SETs’ experience; research has consistently shown that early career SETs leave at higher rates than experienced SETs (Billingsley & Bettini, 2019), and we hypothesized that they would also differ in their intent, but our results did not support this hypothesis. In a post hoc analysis, we considered the possibility that certification status might be capturing effects of experience on intent, but this was not the case. The most likely explanation is that other working conditions may explain apparent associations between experience and intent/attrition. In other words, early career teachers might be more likely to leave because they tend to work in settings that they perceive as less supportive or more demanding; thus, including a range of working conditions in the model may have eliminated any association between early career status and intent. It is also possible that new teacher attrition was depressed in 2011–2012 due to the economic recession (Dewey et al., 2017); new teachers’ responses to the intent items in those years may have captured reduced expectations for other employment options, though further research is needed to understand how teachers’ intent to stay changes in response to changes in economic conditions. Another possibility (albeit, an unlikely one) is that historic trends of higher attrition among early career SETs has changed (i.e., as a result of changes in policy, teacher preparation). This is an area in need of future research.

Finally, it is noteworthy that the models that we fit to the SET sample explained less variance in intent than the models fit to the GET sample. This could suggest that the associations between our predictors and intent were less consistent across SETs; this also may explain why similar sized coefficients across the SET and GET models differed in statistical significance despite large sample sizes. One possible explanation relates to the measures. Some prior research indicates that because SETs and GETs are situated in their schools’ social systems quite differently, it may be important to measure their social supports with slightly different items and, in particular, to measure SETs’ social systems with some items that focus on supports for students with disabilities (Bettini, Jones, et al., 2018). SASS-TQ does not include items that would allow us to address nuances in the ways that SETs’ support needs differ from GETs’ support needs, and this may also have reduced our models’ sensitivity to predictors of SETs’ intent.

Limitations

Several limitations are worth noting. First, we were limited to the items included in the SASS-TQ. We do not have any measure of SETs’ service delivery model (e.g., co-teaching, self-contained), but some variables could operate differently within different service delivery models (e.g., collegial support could be more strongly associated with intent among SETs in inclusive settings). As such, excluding service delivery model could limit the potential for detecting some effects for SETs. Similarly, FRL is a common indicator of school poverty, but it misses other important dimensions of socioeconomic status (education, profession; Kincaid & Sullivan, 2017). We were also unable to measure all demands (e.g., caseload size) and resources (planning time, whether the culture of the school is supportive of inclusion, experience specifically in one’s current role) likely to predict intent. Second, as previously noted, no SASS-TQ items focus on unique aspects of special education teaching; as a result, our measures may have been less sensitive to SETs’ experiences than to GETs’ experiences. Finally, we used intent to stay in one’s school as a proxy for attrition, but the relationship between intent and attrition requires further empirical investigation.

Implications for Future Research, Policy, and Practice

The present study illustrates the utility of COR theory for conceptualizing factors that contribute to teachers’ intent to leave their schools. Future research should continue using COR theory, investigating what resources and demands are most important to teachers and to SETs specifically. For example, researchers could examine resources (e.g., planning time) and demands (e.g., caseload size, disability categories served) that we could not include in this analysis. Future research should also use COR theory to investigate how resources and demands relate to other related outcomes (e.g., attrition, commitment, burnout). COR theory would be especially useful for conceptualizing potential mediators between demands/resources and attrition or intent; for example, researchers could use structural equation modeling to test if this relationship is mediated by teachers’ perceptions of how well they are able to do their job. Structural equation modeling would also allow researchers to test potentially complex relationships among demands and resources; for example, because administrators are responsible for fostering positive school cultures and coordinating the work of all teachers in a school, it is possible that the effects of administrative support might be partially mediated by teachers’ ratings of other working conditions (Billingsley et al., 2017; Gersten et al., 2001).

We found that predictors of intent varied for SETs versus GETs. We recommend that researchers disaggregate data for these groups; understanding how predictors of intent vary by teacher role has implications for differentiating supports. Careful measurement work should undergird this work, as teachers in different roles may require slightly different measures (Bettini, Jones, et al., 2018).

Consistent with prior research (e.g., Feng, 2009; Gilmour & Wehby, 2019) and our hypotheses, student problems predicted intent to leave, a concern as students with greater challenges are in greatest need of experienced teachers (Conroy, Alter, Boyd, & Bettini, 2014). Furthermore, resources did not moderate effects of student problems, a finding also aligned with prior work (Billingsley & Bettini, 2019). However, systems to prevent and address challenging behavior may have potential to reduce intent to leave. Principals should consider using schoolwide positive behavioral interventions and supports, which is associated with reduced discipline referrals, improved student achievement, and reduced burnout (Oakes, Lane, Jenkins, & Booker, 2013; Ross, Romer, & Horner, 2012). Teacher preparation programs also need to provide strong preparation in behavior management, so teachers are well prepared to address student behavior. We recommend that future research test whether the relationship between student problems and intent is moderated by these other factors, such as coursework in classroom/behavior management, practicum experiences with mentors skilled in behavior management, and positive behavioral interventions and supports.

This study bolsters prior research indicating that administrative support predicts teachers’ intent to stay (e.g., Boyd et al., 2011). School leaders should self-evaluate their performance on the specific items associated with intent, with a focus on supporting early career teachers and those in high-poverty schools. Unfortunately, to date, no studies have tested interventions to help leaders improve their supports for SETs (Billingsley & Bettini, 2019). Intervention research would contribute value, first, by permitting causal inferences and, second, by providing practical tools that leaders could use to better support SETs and reduce attrition. Thus, we recommend that scholars consider developing interventions to improve administrative support, testing their effects on attrition.

This study also bolsters prior research indicating that SET certification may support SETs in remaining in the profession (Billingsley & Bettini, 2019; Conley & You, 2017; Gilmour & Wehby, 2019). These results imply that policies aimed at increasing the proportion of fully certified SETs could yield dividends in terms of SET retention. However, further research is needed to understand this finding. Several possible explanations could be tested. First, it is possible that SET certification is a better indicator of knowledge and skill to teach. Second, SET certification could possibly be a better indicator of commitment to the profession. Third, GET certified teachers may experience a more favorable different labor market than SET certified teachers. Each of these possibilities would have different implications. For example, the first possibility would imply that GET preparation should adopt methods and content addressed in SET preparation, whereas the third possibility would imply that GET certification confers a stronger benefit to those who hold it. Understanding implications of this finding will require testing these explanations. Researchers could compare fully certified beginning SETs and GETs along a number of measures, including their knowledge and instructional skills, their commitment, and their other job opportunities.

Finally, intent to leave is a widely used measure in attrition research (Billingsley & Bettini, 2019). It is relatively inexpensive to measure, and it is associated with other negative outcomes (e.g., burnout; Bettini, Jones, et al., 2017; Brunsting et al., 2014). However, measurement research is needed to understand how it varies over the year and how it relates to actual attrition.

Supplemental Material

Appendix_A – Supplemental material for Predicting Special and General Educators’ Intent to Continue Teaching Using Conservation of Resources Theory

Supplemental material, Appendix_A for Predicting Special and General Educators’ Intent to Continue Teaching Using Conservation of Resources Theory by Elizabeth Bettini, Allison F. Gilmour, Thomas O. Williams and Bonnie Billingsley in Exceptional Children

Supplemental Material

Appendix_B – Supplemental material for Predicting Special and General Educators’ Intent to Continue Teaching Using Conservation of Resources Theory

Supplemental material, Appendix_B for Predicting Special and General Educators’ Intent to Continue Teaching Using Conservation of Resources Theory by Elizabeth Bettini, Allison F. Gilmour, Thomas O. Williams and Bonnie Billingsley in Exceptional Children

Footnotes

Open Practices

For making content publicly available, this article earned a badge for open materials. The public content for open materials may be retrieved from .

Supplemental Material

Supplemental material for this article is available online.

References

Achinstein

Ogawa

R. T.

Sexton

Freitas

(2010). Retaining teachers of color: A pressing problem and a potential strategy for “hard-to-staff” schools. Review of Educational Research, 80, 71–107. doi:10.3102/0034654309355994

Alarcon

G. M.

(2011). A meta-analysis of burnout with job demands, resources, and attitudes. Journal of Vocational Behavior, 79, 549–562. doi:10.1016/j.jvb.2011.03.007

Barnes

Crowe

Schaefer

(2007). The cost of teacher turnover in five school districts. Washington, DC: National Commission on Teaching and America’s Future.

Beteille

Kalogrides

Loeb

(2012). Stepping stones: Principal career paths and school outcomes. Social Science Research, 41, 904–919. doi:10.1016/j.ssresearch.2012.03.003

Bettini

Cumming

Merrill

Brunsting

Liaupsin

C. J.

(2017). Working conditions in self-contained settings for students with emotional disorders. Journal of Special Education, 51, 83–94. doi:10.1177/0022466916674195

Bettini

Cumming

O’Brien

K. M.

Brunsting

Ragunanthan

Sutton

Chopra

(in press). Working conditions and special educators’ intent to continue teaching students with emotional/behavioral disorders in self-contained classes. Exceptional Children.

Bettini

Jones

N. D.

Brownell

M. T.

Conroy

Leite

(2018). Relationships between novices’ social resources and workload manageability. Journal of Special Education, 52, 113–126. doi:10.1177/0022466918775432

Bettini

Jones

Brownell

Conroy

Park

Leite

Crockett

Benedict

(2017). Workload manageability among novice special and general educators: Relationships with emotional exhaustion and career intentions. Remedial and Special Education, 38, 246–256. doi:10.1177/0741932517708327

Bettini

Mason-Williams

Mathews

Boveda

Rodgers

Zhao

Zeng

(2018, April). Comparing novice teachers’ access to supportive social contexts across shortage versus non-shortage areas. Presentation at the American Education Research Association Conference, New York City, NY.

10.

Bettini

Wang

Cumming

Kimerling

Schutz

(2018). Special educators’ experiences of roles and responsibilities in self-contained classes for students with emotional/behavioral disorders. Remedial and Special Education, 31, 126–142.

11.

Billingsley

(2004). Special education teacher retention and attrition: A critical analysis of the research literature. The Journal of Special Education, 38, 39–55. doi:10.1177/00224669040380010401

12.

Billingsley

Bettini

(2017). Factors influencing the quality of the special education teacher workforce. In Kauffman

Hallahan

Pullen

(Eds.), Handbook of special education. Abingdon, UK: Routledge.

13.

Billingsley

B. S.

Bettini

(2019). Special education teacher attrition research: A systematic literature review. Review of Educational Research. doi:10.3102/0034654319862495

14.

Billingsley

Cross

(1992). Predictors of commitment, job satisfaction, and intent to stay in teaching: A comparison of general and special educators. The Journal of Special Education, 25, 453–471. doi:10.1177/002246699202500404

15.

Billingsley

McLeskey

Crockett

J. B.

(2017). Principal leadership: Moving toward inclusive and high-achieving schools for students with disabilities (Document No. IC-8). Gainesville, FL: University of Florida, Collaboration for Effective Educator, Development, Accountability, and Reform Center.

16.

Boe

E. E.

Cook

L. H.

Sunderland

R. J.

(2008). Teacher turnover: Examining exit attrition, teaching area transfer, and school migration. Exceptional Children, 75, 7–31. doi:10.1177/001440290807500101

17.

Boe

E. E.

Shin

Cook

L. H.

(2007). Does teacher preparation matter for beginning teachers in either special or general education? Journal of Special Education, 41, 158–170. doi:10.1177/00224669070410030201

18.

Borman

G. D.

Dowling

N. M.

(2008). Teacher attrition and retention: A meta-analytic and narrative review of the research. Educational Research, 78, 367–409. doi:10.3102/0034654308321455

19.

Boyd

Grossman

Ing

Lankford

Loeb

Wyckoff

(2011). The influence of school administrators on teacher retention decisions. American Educational Research Journal, 48, 303–333. doi:10.3102/0002831210380788

20.

Boyd

Grossman

Lankford

Loeb

Wyckoff

(2008). Who leaves? Teacher attrition and student achievement (Working Paper No. 14022). Cambridge, MA: National Bureau of Economic Research.

21.

Brownell

M. T.

Steinbrecher

Kimerling

Park

Bae

Benedict

(2014). Dimensions of teacher quality in general and special education. In Sindelar

P. T.

McCray

E. D.

Brownell

M. T.

Lignugaris-Kraft

(Eds.), Handbook of research on special education teacher preparation (pp. 423–444). New York: Routledge, Taylor & Francis.

22.

Brunsting

N. C.

Sreckovic

M. A.

Lane

K. L.

(2014). Special education teacher burnout: A synthesis of research from 1979 to 2013. Education and Treatment of Children, 37, 681–711. doi:10.1353/etc.2014.0032

23.

Carver-Thomas

Darling-Hammond

(2017). Teacher turnover: Why it matters and what we can do about It. Palo Alto, CA: Learning Policy Institute.

24.

Conley

You

(2017). Key influences on special education teachers’ intention to leave: The effects of administrative support and teacher team efficacy in a mediational model. Educational Management, Administration, and Leadership, 45, 521–540. doi:10.1177/1741143215608859

25.

Connelly

Graham

(2009). Student teaching and teacher attrition in special education. Teacher Education and Special Education, 32, 257–269. doi:10.1177/0888406409339472

26.

Conroy

M. A.

Alter

P. J.

Boyd

B. A.

Bettini

(2014). Teacher preparation for students who demonstrate challenging behaviors. In Sindelar

P. T.

McCray

E. D.

Brownell

M. T.

Lignugaris-Kraft

(Eds.) Handbook of research on special education teacher preparation (pp. 320–333). New York, NY: Routledge, Taylor & Francis.

27.

Cross

Billingsley

(1994). Testing a model of special educators’ intent to stay in teaching. Exceptional Children, 60, 411–421. doi:10.1177/001440299406000504

28.

DeAngelis

K. J.

Pressley

J. B.

(2011). Toward a more nuanced understanding of new teacher attrition. Education and Urban Society, 43, 598–626. doi:10.1177/0013124510380724

29.

Dewey

Sindelar

P. T.

Bettini

Boe

E. E.

Rosenberg

M. S.

Leko

(2017). Explaining the decline in special education teacher employment from 2005 to 2012. Exceptional Children, 83, 315–329. doi:10.1177/0014402916684620

30.

Diamond

J. B.

Spillane

J. P.

(2004). High-stakes accountability in urban elementary schools: Challenging or reproducing inequality? Teachers College Record, 106, 1145–1176. doi:10.1111/j.1467-9620.2004.00375.x

31.

Fall

A. M.

Billingsley

(2011). Disparities in work conditions among early career special educators in high- and low-poverty districts, Remedial and Special Education, 32, 64–78. doi:10.1177/0741932510361264

32.

Feng

(2009). Opportunity wages, classroom characteristics, and teacher mobility. Southern Economic Journal, 75, 1165–1190.

33.

Feng

Sass

(2013). What makes special-education teachers special? Teacher training and achievement of students with disabilities. Economics of Education Review, 36, 122–134. doi:10.1016/j.econedurev.2013.06.006

34.

Gersten

Keating

Yovanoff

Harniss

M. K.

(2001). Working in special education: Factors that enhance special educators’ intent to stay. Exceptional Children, 67, 549–567. doi:10.1177/001440290106700408

35.

Gilmour

A. F.

Wehby

(2019). The association between teaching students with disabilities and teacher turnover. Journal of Educational Psychology. doi:10.1037/edu0000394

36.

Grissom

J. A.

(2011). Can good principals keep teachers in disadvantaged schools? Linking principal effectiveness to teacher satisfaction and turnover in hard-to-staff environments. Teachers College Record, 113, 2552–2585.

37.

Grossman

Thompson

(2004). District policy and beginning teachers: A lens on teacher learning. Educational Evaluation and Policy Analysis, 26, 281–301. doi:10.3102/01623737026004281

38.

Grossman

Thompson

(2008). Learning from curriculum materials: Scaffolds for new teachers? Teaching and Teacher Education, 24, 2014–2026. doi:10.1016/j.tate.2008.05.002

39.

Guarino

C. M.

Santibañez

Daley

G. A.

(2006). Teacher recruitment and retention: A review of the recent empirical literature. Review of Educational Research, 76, 173–208. doi:10.3102/00346543076002173

40.

Halbesleben

J. R. B.

Neveu

J.-P.

Paustian-Underdahl

S. C.

Westman

(2014). Getting to the “COR”: Understanding the role of resources in conservation of resources theory. Journal of Management, 40, 1334–1364. doi:10.1177/0149206314527130

41.

Hanushek

E. A.

Kain

J. F.

Rivkin

S. G.

(2004). Why public schools lose teachers. Journal of Human Resources, 39, 326–354. doi:10.2307/3559017

42.

Henry

G. T.

Bastian

K. C.

Fortner

C. K.

(2011). Stayers and leavers: Early-career teacher effectiveness and attrition. Educational Researcher, 40, 271–280. doi:10.3102/0013189X11419042

43.

Irvin

D. W.

Hume

Boyd

B. A.

McBee

M. T.

Odom

S. L.

(2013). Child and classroom characteristics associated with the adult language provided to preschoolers with autism spectrum disorder. Research in Autism Spectrum Disorders, 7, 947–955. doi:10.1016/j.rasd.2013.04.004

44.

Jackson

C. K.

Bruegmann

(2009). Teaching students and teaching each other: The importance of peer learning for teachers. American Economic Journal, 1, 85–108. doi:10.3386/w15202

45.

Jackson

Makarin

(2016). Simplifying teaching: A field experiment with online “off-the-shelf” lessons. Cambridge, MA: NBER. Retrieved from http://www.nber.org/papers/w22398

46.

Johnson

S. M.

Birkeland

S. E.

(2003). Pursuing a “sense of success”: New teachers explain their career decisions. American Educational Research Journal, 40, 581–617. doi:10.3102/00028312040003581

47.

Johnson

S. M.

Kraft

M. A.

Papay

J. P.

(2012). How context matters in high-need schools: The effects of teachers’ working conditions on their professional satisfaction and their students’ achievement. Teachers College Record, 114, 1–39.

48.

Jones

N. D.

Youngs

Frank

K. A.

(2013). The role of school-based colleagues in shaping the commitment of novice special and general educations. Exceptional Children, 79, 1–19. doi:10.1177/001440291307900303

49.

Kincaid

A. P.

Sullivan

A. L.

(2017). Parsing the relations of race and socioeconomic status in special education disproportionality. Remedial and Special Education, 38, 159–170. doi:10.1177/0741932516671199

50.

Ladd

H. F.

Sorenson

L. C.

(2014). Returns to teacher experience: Student achievement and motivation in middle school (Working Paper No. 112). Washington, DC: National Center for Analysis of Longitudinal Data in Education Research.

51.

McLeskey

Billingsley

B. S.

(2008). How does the quality and stability of the teaching force influence the research-to-practice gap? A perspective on the teacher shortage in special education. Remedial and Special Education, 29, 293–305. doi:10.1177/0741932507312010

52.

Milanowski

A. T.

Odden

A. R.

(2007). A new approach to the cost of teacher turnover (Working Paper No. 13). Retrieved from the School Finance Redesign Project website: http://www.schoolfinanceredesign.org

53.

Miller

M. D.

Brownell

M. T.

Smith

S.W.

(1999). Factors that predict teachers staying in, leaving, or transferring from the special education classroom. Exceptional Children, 65, 201–218.

54.

National Center for Education Statistics. (2015). Schools and Staffing Survey. Washington, DC: Institute of Education Sciences. Retrieved from https://nces.ed.gov/surveys/sass/

55.

Oakes

W. P.

Lane

K. L.

Jenkins

Booker

B. B.

(2013). Three-tiered models of prevention: Teacher efficacy and burnout. Education and Treatment of Children, 36, 95–126. doi:10.1353/etc.2013.0037

56.

Otis-Wilborn

Winn

Griffin

Kilgore

(2005). Beginning special educators’ forays into general education. Teacher Education and Special Education, 28, 143–152. doi:10.1177/088840640502800401

57.

Ronfeldt

Loeb

Wyckoff

(2013). How teacher turnover harms student achievement. American Educational Research Journal, 50, 4–36. doi:10.3102/0002831212463813

58.

Ronfeldt

Schwartz

Jacob

B. A.

(2014). Does preservice preparation matter? Examining an old question in new ways. Teachers College Record, 116, 1–46.

59.

Ross

S. W.

Romer

Horner

R. H.

(2012). Teacher well-being and the implementation of school-wide positive behavior interventions and supports. Journal of Positive Behavior Interventions, 14, 118–128. doi:10.1177/1098300711413820

60.

Ruble

L. A.

McGrew

J. H.

(2013). Teacher and child predictors of achieving IEP goals of children with autism. Journal of Autism and Developmental Disorders, 43, 2748–2763. doi:10.1007/s10803-013-1884-x

61.

Simon

N. S.

Johnson

S. M.

(2015). Teacher turnover in high-poverty schools: What we know and can do. Teachers College Record, 117, 1–37.

62.

Singh

Billingsley

B. S.

(1996). Intent to stay in teaching: Teachers of students with emotional disorders versus other special educators. Remedial and Special Education, 17, 37–47. doi:10.1177/074193259601700105

63.

Siuty

M. B.

Leko

M. M.

Knackstedt

K. M.

(2018). Unraveling the role of curriculum in teacher decision making. Teacher Education and Special Education, 41, 39–57. doi:10.1177/0888406416683230

64.

Wayne

A. J.

Youngs

(2003). Teacher characteristics and student achievement gains. Review of Educational Research, 73, 89–122. doi:10.3102/00346543073001089

65.

Wong

V. W.

Ruble

L. A.

McGrew

J. H.

(2017). Too stressed to teach? Teaching quality, student engagement, and IEP outcomes. Exceptional Children, 83, 412–427. doi:10.1177/0014402917690729

66.

Youngs

Holdgreve-Rezendez

R. T.

Qian

(2011). The role of instructional program coherence in beginning elementary teachers’ induction experiences. Elementary School Journal, 111, 455–476. doi:10.1086/657655

67.

Youngs

Jones

Low

(2011). How beginning special and general education elementary teachers negotiate role expectations and access to professional resources. Teachers College Record, 113, 1506–1540.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB

0.05 MB