	New content
SAGE Open

Open access

Research article

First published online May 10, 2020

A Meta-Analysis of Special Education Teachers’ Burnout

Eun-Young Park and Mikyung Shin https://orcid.org/0000-0001-7907-9193 [email protected]View all authors and affiliations

All Articles

https://doi.org/10.1177/2158244020918297

Abstract

This meta-analysis verifies associations between three dimensions of special education teachers’ (SET) burnout (emotional exhaustion, depersonalization, and personal accomplishment) and student-, teacher-, and school-related variables. Altogether, 28 peer-reviewed English articles and 13 dissertations (total sample of teachers = 6,623) published between 1983 and December 2018 were analyzed. The degree of correlation effect sizes between special education teachers’ burnout and its related variables was extensive. Results revealed distinct relations by each burnout dimension: emotional exhaustion, depersonalization, and personal accomplishment and student-, teacher-, and school-related variables. Student age (Fisher’s Z = .316) was significantly associated with SET depersonalization. Self-efficacy (Fisher’s Z = −0.390, emotional exhaustion; −0.321, depersonalization; 0.633, personal accomplishment), stress (0.366, emotional exhaustion; 0.340, depersonalization; −0.110 personal accomplishment), and support from school personnel (−0.119, emotional exhaustion; −0.140, depersonalization; 0.172, personal accomplishment) were also significantly related to each burnout dimension. Support programs to relieve SET burnout must consider these variables.

The problem of teacher’s burnout has been the steady subject of research. Special education teachers have been reported to be particularly vulnerable to burnout (Emery & Vandenberg, 2010; Y. L. Lee et al., 2011). Burnout was identified as a major contributor to attrition in previous research (Wong et al., 2017). Even in America, this problem has produced a sustained reduction in special education teachers (Emery & Vandenberg, 2010; Fore et al., 2002; Leko & Smith, 2010). The U.S. Department of Education, Office of Postsecondary Education (2017) reported that 46 states were drastically short of special education teachers. Burnout is closely related to the physical and psychological health and work performance of teachers, as well as attrition, and it may affect teaching and the way teachers interact with students in educational settings (Hall, 2013). Teachers’ burnout gradually develops over an extended period, becoming chronic and worse (Fernet et al., 2012). Teachers’ burnout is reported to be related to student motivation; teachers’ burnout can undermine student motivation (Shen et al., 2015). Therefore, research on special education teachers’ burnout has become a major concern and is extensively studied (Brunsting et al., 2014; Leung & Lee, 2006); however, effective intervention strategies for burnout do not yet exist (Hastings & Brown, 2002).

Burnout has a long-term relationship with individuals who are constantly exposed to fatigue, hostility, discouragement, maladjustment, discomfort, and restraint. It is defined as a psychological syndrome involving three dimensions: emotional exhaustion, depersonalization, and a diminished sense of personal accomplishment (Maslach, 1982). Those experiencing burnout have frustration, cynicism, and other negative emotions toward themselves and their work, which causes emotional depletion, impersonality, and an overall diminishment of their sense of achievement (J. H. J. Lee & Ok, 2012). Teacher burnout is a serious concern, because it threatens psychological well-being. Unlike stress, it is difficult for those afflicted to manage burnout in the work context where it manifests. It is therefore important to identify preventive measures and promotional variables.

Key Variables Associated With Special Education Teachers’ Burnout

Various student-, teacher-, and school-related variables related to the onset of special education teachers’ burnout exist. These variables closely align with the realities of special education teachers’ daily work experiences; therefore, it is critical to provide them with the means to alleviate burnout before it produces negative outcomes (Brunsting et al., 2014).

Student-Related Variables

Student-related variables include age, disability type, behavior problems, grade, student numbers, socioeconomic status (SES), and setting. Special education teachers demonstrated different responses based on students’ disability type, behavior problems, and the severity of students’ challenges. Banks and Necco (1990) discovered that teachers teaching students with emotional disorders demonstrated higher burnout than those teaching students with intellectual disabilities. Nichols and Sosnowsky (2002) reported that the proportion of students with emotional disorders who exhibited challenging behaviors was associated with higher intensity of teacher burnout in self-contained classrooms. Student measures of severity (e.g., intelligence quotient [IQ], language, adaptive behavior) negatively affected teacher satisfaction and required greater support in schools (Hastings & Brown, 2002; Ruble & McGrew, 2013). Moreover, students’ SES and service delivery models could affect teachers’ teaching experience and burnout levels. Kvande et al. (2018) identified the mediated effects of students’ SES status on special education services. Specifically, students with low SES were more likely to receive special education services, and teachers reported increased feelings of helplessness when these students demonstrated low academic achievement.

Teacher-Related Variables

Teacher-related variables include teachers’ age, gender, teaching experience, education level, self-efficacy, satisfaction, coping, and stress. Gong et al. (2013) noted that special education teachers aged between 26 and 30 years displayed greater emotional exhaustion than those above 30 years. Brewer and McMahan (2004) also reported that women were more likely to experience job stress and burnout. Gong et al. (2013) stated that female teachers experienced higher personal accomplishment, yet male teachers were more prone to depersonalization. Years of teaching experience were also associated with teacher burnout. Williams and Dikes (2015) found that the number of years of teaching was positively correlated with burnout; thus, the greater the number of years at school, the higher the stress and burnout level. Carlson and Thompson (1995), however, indicated that teacher experience did not significantly contribute to degree of teacher burnout. Teachers’ education levels were also connected to burnout issues. As Embich (2001), in particular, indicated, younger, less experienced, and less educated team-teaching teachers experienced higher burnout levels than other teachers. Studies also reported that teachers’ self-efficacy (Sariçam & Sakiz, 2014), job satisfaction (Platsidou, 2010), and coping capacities (Boujut et al., 2016) supported special education teachers’ experiences in schools, reducing teacher burnout. However, teacher stress was positively associated with teacher burnout variables of emotional exhaustion and depersonalization, yet negatively related to personal accomplishment (Ruble & McGrew, 2013).

School-Related Variables

School-related variables include work hindrances, emotional experiences, number of support personnel, available resources, and support from school personnel. Recently, with the emphasis on teacher accountability, special education teachers’ increased workloads (e.g., managing classrooms, working as case managers for Individualized Education Program meetings, student progress monitoring) are positively associated with their burnout levels (Embich, 2001; Ruble et al., 2011). Studies also support that work hindrances (e.g., role conflict, role ambiguity) are positively associated with special education teachers’ burnout levels (Adera & Bullock, 2010; Embich, 2001). However, emotional experiences in schools such as feeling secure, autonomous (Carlson & Thompson, 1995), and supported by other teachers and administrators (Bataineh & Alsagheer, 2012; Boujut et al., 2016) were negatively associated with special education teachers’ burnout. Zarafshan et al. (2013) also revealed that positive relations with coworkers and opportunities for promotion at schools decreased the level of emotional exhaustion and depersonalization. Unfortunately, however, Kaff (2004) noted that numerous special education teachers did not feel they received sufficient support from their principals and resources in schools to manage their academic responsibilities.

Exploring Special Education Teachers’ Burnout in Previous Reviews and Syntheses

Literature has focused on issues such as stress and burnout among special education teachers for more than three decades (Banks & Necco, 1990; Embich, 2001; R. H. Zabel & Zabel, 2002). Given, especially, the increasing demands on special education teachers through education quality and educational accountability policies (e.g., the Every Student Succeeds Act of 2015 in America and the Special Education Act for Individuals with Disabilities and Others of 2008 in South Korea), they are prone to increased burnout, a high risk factor associated with teacher attrition (Hagaman & Casey, 2018; Ryan et al., 2017). Special education teachers’ burnout is associated with their personal and classroom issues, as well as their increasing educational accountability. Despite these ongoing concerns, in previous years, only two studies (Brunsting et al., 2014; Wisniewski & Gargiulo, 1997) conducted a synthesis and a review, respectively, investigating the associations between special education teachers’ burnout and related variables. Other two recent studies (Aloe et al., 2014; Iancu et al., 2018) targeted school teachers in general and examined the intervention programs that aim to reduce teacher burnout levels.

Wisniewski and Gargiulo (1997) conducted a review of studies of occupational stress and burnout among special education teachers. They reviewed mediating variables that affect teacher stress. For example, they stated that role conflict and role ambiguity were the primary sources of teacher stress. Specifically, special education teachers’ stress increased when organizations provided information about their roles and responsibilities that conflicted with their jobs and when they had insufficient information to perform their teaching responsibilities. Furthermore, Wisniewski and Gargiulo also indicated that professional interactions with other school personnel (e.g., teachers, administrators, parents) and instructional assignments (e.g., the highly stressful condition of classrooms of students with emotional and behavioral disorders) influenced special education teachers’ burnout.

Brunsting et al. (2014) conducted a synthesis of special education teacher burnout from 1979 to 2013. They reviewed 23 studies, altogether, which measured teacher burnout in terms of their emotional exhaustion, depersonalization, and lack of personal accomplishment. In addition, they analyzed teacher burnout at individual, classroom, school, district, and state levels. Findings revealed that stress management and emotional regulation affected a decrease in burnout. Students’ challenging behavior (e.g., percentage of students with emotional disorders) was identified as a variable predominantly affecting teachers’ degree of burnout. Student age was another risk factor demonstrating that special education teachers in secondary schools were exposed to higher levels of burnout. Such school variables as role conflict, role ambiguity, and administrative support were strong factors in teacher burnout.

Although Aloe et al. (2014) also quantitatively analyzed teacher burnout, they only included one study (Sedgwick, 1998) that combined general and special education teachers, thus lacking disaggregated findings for special education teachers’ issues. More recently, Iancu et al. (2018) conducted a meta-analysis on the effects of interventions that aim to reduce teacher burnout in general. Only two out of 23 studies focused on special education teacher as their participants (Breeman et al., 2016; Cooley & Yovanoff, 1996). Therefore, exclusive focus on special education teacher burnout is required. Furthermore, as Brunsting et al. (2014) were the last to review special education teacher burnout, targeted studies were conducted only in the United States. However, the issue of special education teacher burnout is gaining research attention in various countries around the world, such as the United Kingdom (e.g., Hastings & Brown, 2002), the Netherlands (e.g., Hopman et al., 2018), Greece (e.g., Platsidou, 2010), Iran (e.g., Zarafshan et al., 2013), Turkey (e.g., Sariçam & Sakiz, 2014), and the United States (e.g., Ruble & McGrew, 2013). These studies commonly reported high special education teacher burnout levels compared with that of general education teachers, and related factors included age, gender, number of teaching years, degree of self-efficacy, stress level, and type of disabilities. Therefore, to generalize the findings across nations and update the synthesis to cover the global context, we targeted studies published in English and extended our research scope to include countries beyond the United States. Furthermore, Brunsting et al. (2014) descriptively reviewed issues related to special education teachers’ burnout, while this study utilized a meta-analytic method to quantitatively examine variables associated with special education teachers’ burnout. Meta-analysis serves as a statistical method for synthesizing the findings of primary studies; it possesses a more advanced statistical capacity to detect effects than primary studies. Through meta-analysis, we can compare effect sizes across studies that differ in experimental rigor and other methodological factors (Lipsey, 2003). Thus, we conducted a meta-analysis to reach an overall conclusion regarding special education teachers’ burnout associated with teacher-, student-, and school-related variables. It provides the magnitude and direction of effect of related variables on special education teachers’ burnout. The research questions are as follows: (a) Is there a relation between burnout score and student-related variables? (b) Is there a relation between burnout score and teacher-related variables? (c) Is there a relation between burnout score and school-related variables?

Method

Literature Search

A comprehensive search of the literature was performed through a three-step process (Cooper, 1998). First, an online database search of EBSCOhost (n = 912), PsycINFO (n = 989), Education Source (n = 72), and Education Resources Information Center (ERIC) (n = 98) was conducted to locate a total of 2,071 studies.

Online search terms and their combinations

In Field 1, terms related to disabilities such as mental retardation, intellectual disability, autism, emotional and/or behavioral disorder, emotional disturbance, health impairment, or disorder were used. Terms related to special education teachers were inputted in Field 2. Terms related to burnout such as emotional exhaustion, depersonalization, personal accomplishment, and burnout were inserted in Field 3. A total 2,071 studies were identified in the initial search. Of these, 114 duplicated studies were excluded.

Second, the two authors screened the titles and abstracts of the 1,957 studies across four inclusion criteria: (a) Participants should be special education teachers, (b) studies should be published in English peer-reviewed journals and dissertations, (c) studies should measure and report disaggregated data on special education teachers’ burnout (i.e., emotional exhaustion, depersonalization, and personal accomplishment) in schools, and (d) studies should provide the correlation data. According to recommendations of meta-analysis article searching procedures, studies which employed similar methodology were included in this meta-analysis (Gopalakrishnan & Ganeshkumar, 2013). We included studies with correlation, regression, and longitudinal research designs and excluded intervention studies. After assessing the titles, 1,712 studies were excluded. Of the 245 screened studies, 133 were excluded after reviewing the abstracts.

Third, the full-text copies of 112 studies were retrieved and two authors reviewed them to meet the inclusion criteria; 89 noneligible studies were excluded for the following reasons: (a) 28 (31.5%) did not target special education teachers, (b) 15 (16.9%) were not related to burnout, (c) 43 (48.3%) did not meet the research design criteria, (d) one (1.1%) did not provide sufficient information for data analysis, and (e) two studies (2.2%) did not report subcategory measures. Twenty-three articles were selected through this review. In December 2018, we updated the included articles through database and manual searches. Statistically nonsignificant results are less likely to be published and included in meta-analyses. To prevent this file-drawer effect, two authors conducted an inverse search on the reference pages of the 23 included articles and found five additional articles. As a last step, a search for dissertations was completed. In ProQuest and PsycINFO, 56 dissertations were initially searched; 13 dissertations meeting the four inclusion criteria (see above) were finally included. In these research procedures, 28 articles and 13 dissertations were selected for this meta-analysis.

Out of 41 included studies in the current meta-analysis, 11 studies (26.8%) overlapped with Brunsting et al. (2014); although Brunsting et al. (2014) included 23 studies in their synthesis, 12 of the 23 studies (e.g., Frank & McKenzie, 1993; M. K. Zabel et al., 1984) did not meet our criteria. Figure 1 describes the screening process.

Figure 1. A flowchart describing the selection of the 41 studies after multiple phases of the screening process.
*Note.* ERIC = Education Resources Information Center.

Coding Procedure and Inter-Rater Reliability

Using Microsoft Excel, we coded the following variables: design, number of special education teachers, age, gender, years of teaching, burnout measures, related variables, and correlation coefficients between their burnout and related variables. Regarding burnout measures, we focused on emotional exhaustion, depersonalization, and personal accomplishment following the three dimensions of the Maslach Burnout Inventory (MBI; Maslach & Jackson, 1996): depersonalization, that is, development of negative and cynical attitudes toward service users; emotional exhaustion, that is, staff feeling that they have little to offer at a psychological level to their work; and personal accomplishment, that is, staff evaluating themselves and their accomplishments positively. Given the meaning discrepancy underlying these three dimensions, the positively higher scales in depersonalization and emotional exhaustion denote negative attitudes, while the positively higher scales in personal accomplishment denote positive attitudes. We reversely coded eight studies (Chatlos, 2016; Crane & Iwanicki, 1986; Dickerson, 2017; Embich, 2001; Freed, 1994; Goodall, 1986; Hastings & Brown, 2002; LaMonica, 1983) that analyzed personal accomplishment as reduced personal accomplishment in their original survey. When coding the data from the longitudinal research design, we independently considered each correlation effect size at each different time point. Furthermore, for burnout-related variables, we referred to Bronfenbrenner’s (1977) ecological model and organized variables as student-, teacher-, and school-related (see Brunsting et al., 2014). For example, we included student age, disability, and setting as student-related variables; teacher age, gender, and years of teaching as teacher-related variables; and resources, support from school personnel, and the number of support personnel as school-related variables. Variables that could not be categorized as specific variables were coded as others. Others in teacher-related variables included instructional strategies, teaching quality, teacher race, and general health. Others in student-related variables included language, IQ, and student engagement.

We received training before initiating the coding. Four (14.8%) articles were coded independently. We initially achieved an inter-rater reliability of 97.3%, using a percent agreement method: “agreements/(agreements + disagreements) × 100% = P%” (Araujo & Born, 1985, p. 208). The disagreement was resolved through discussion of the definition of variables. In the second coding, there was 100% agreement. After the training, two coders independently coded all articles and cross-checked the completed coding. Between the two coders, the average inter-rater reliability was 98% for emotional exhaustion (range = 97%–100% for each variable), 99% for depersonalization (range = 99%–100% for each variable), and 99% for personal accomplishment (range = 99%–100% for each variable). Two coders reviewed the dissimilar coding items by rechecking the full text together to ensure 100% agreement on all the coding variables.

Data Analysis

Calculation of effect sizes

We employed Pearson’s r as the effect size index by examining the correlation between special education teachers’ burnout and related variables. When studies reported regression data results, we extracted r with the following conditions: The regression should be linear, indicating one predictor variable (df = 1); or we could extract R (or R²) from the regression table, and the correlation direction should be the same as the nonstandardized B value. In certain cases, studies reported more than one related measure; here, we coded more than one category whenever applicable and displayed orthogonal contrasts to capture the relation between two variables (Gersten et al., 2009). Following Borenstein et al.’s (2009) recommendations, we converted the correlation coefficients to Fisher’s Z scale using the following transformation formula: Fisher’s Z = 0.5 * Log (1 + Corr)/(1 − Corr). Subsequently, we performed data analysis using the transformed Fisher’s Z values.

Statistical analyses

For statistical analyses such as publication bias, overall effect sizes, and Q statistics, data were analyzed employing Comprehensive Meta-Analysis (CMA) version 3.0 (Biostat, 2017) and aggregated effect sizes across studies. First, before applying the meta-analytic approach, we examined the publication bias using Duval and Tweedie’s Trim and Fill (Egger et al., 1997). The possibility of publication bias can indicate an overestimation of meta-analysis results (Rosenberg, 2005; Rosenthal, 1979).

Second, we disaggregated effect sizes by emotional exhaustion, depersonalization, and personal accomplishment. Evaluating heterogeneity of variance using the Q statistic for each effect size, we applied a random-effects model for the current study when p value was less than .05, indicating significant between-studies variance; subsequently, we applied a fixed-effects model when p value was equal or greater than .05, indicating nonsignificant between-studies variance (Borenstein et al., 2009; Lipsey & Wilson, 2001).

Results

Table 1 displays the summary of study features of 41 included studies (28 articles, 13 dissertations) that examined special education teachers’ burnout issues related to student-, teacher-, and school-related variables. The publication years of 41 studies ranged from 1983 to 2018: there were seven between 1983 and 1989, 12 between 1990 and 2009, and 22 between 2010 and 2018. Of the 41 studies, 29 (70.7%) were published in the United States. A total of 6,623 teachers were included in the meta-analysis. Sample sizes ranged from 21 to 443 special education teachers. Out of 41 studies, 32 reported information regarding the samples’ gender: out of 4,070 reported samples, 881 (21.6%) were male and 3,189 (78.4%) were female special education teachers. In 24 studies reporting samples’ age, participants’ ages ranged from 21 to 71 years. In 30 studies reporting samples’ teaching experience, participants possessed an extensive range of years of teaching from less than 1 year to at least 30 years.

Table 1. Summary of Study Characteristics.

Author (year)	N	Age years (n)	Gender	Years of teaching (n)	Related variables	Country	Type
Ambrosini (2013)	94	29–42 (25), 43–55 (46), 56–69 (23)	19 M, 75 F	1–8 (50), 9–12 (24), 13–33 (20)	T: Age, EL, stress, TE	United States	D
Banks and Necco (1990)	181	NR	NR	9.36	T: Age, TE	United States	A
Bataineh (2009)	83	24–48	43 M, 40 F	3–15	Sc: SSP	Jordan	A
Bataineh and Alsagheer (2012)	300	22–45	50 M, 250 F	4–17	Sc: SSP	UAE	A
Biglan et al. (2013)	42	NR	NR	NR	T: Coping	United States	A
Bloom (1992)	33	37.6	2 M, 31 F	12.7	T: Age, TE; Sc: EE, NSP, WH	United States	D
Boujut et al. (2016)	103	41.58	15 M, 88 F	14.84	T: Coping; Sc: EE, R, satisfaction, stress, SSP	France	A
Caputo and Langher (2015)	276	30–62	52 M, 224 F	17.6	Sc: SSP	Italy	A
Chatlos (2016)	31	NR	5 M, 23 F, 3 NR	0–10 (12), 11–20 (14), >20 (5)	T: Self-efficacy, O	United States	D
Coman et al. (2013)	53	NR	1 M, 52 F	9.83	St: Disability, SN; T: E; Sc: SSP	United States	A
Crane and Iwanicki (1986)	443	NR	NR	NR	T: O; Sc: WH	United States	A
Dickerson (2017)	46	NR	8 M, 37 F, 1 NR	NR	Sc: Satisfaction	United States	D
Eichinger et al. (1991)	78	NR	78 F	NR	T: Coping	United States	A
Embich (2001)	300	40.19	NR	11.18	T: Age, EL, TE; Sc: SSP, WH	United States	A
Fimian and Blanton (1986)	415	NR	NR	NR	St: SN, O; T: Age, gender, stress, O; Sc: WH	United States	A
Freed (1994)	342	20–39 (219), 40–49 (88), >50 (35)	63 M, 279 F	11.04	St: Disability, grade, setting, SN; T: Age, gender, TE; Sc: Stress, SSP, WH	United States	D
Garwood et al. (2018)	64	41.16	4 M, 60 F	14.94	T: Self-efficacy; Sc: WH	United States	A
Goetzinger (2006)	224	NR	12 M, 211 F, 1 NR	13.4–19.9	St: SN; T: TE; Sc: WH	United States	D
Gong et al. (2013)	256	30 (108), 31–50 (102),51 (46)	72 M, 184 F	0–9 (174), 10–29 (67), 30 (15)	T: Age, gender, TE, O; Sc: SSP	United States	A
Goodall (1986)	212	20–69	66 M, 146 F	1–45	T: Age, EL, gender, TE	United States	D
Hassan and Mohamed (2015)	81	32.38	81 F	1–5 (19), 6–10 (33), 10 (29)	T: Stress, TE; Sc: SSP	Oman	A
Hastings and Brown (2002)	55	NR	14 M, 41 F	7.34	St: Behavior problem; T: Coping	United Kingdom	A
Hill (2011)	71	34.69	20 M, 51 F	6.05	T: Age, EL, TE; Sc: SSP	United States	D
Hopman et al. (2018)	141	38.7	55 M, 86 F	5.1	St: Behavior problem, O; T: O; Sc: SSP, WH	NL	A
LaMonica (1983)	99	NR	39 M, 60 F	12.86	T: TE; Sc: EE, stress	United States	D
Martin (2010)	105	21–38 (48), 39–55 (39), 56 (18)	12 M, 93 F	1–15 (61), 16–25 (16), 25 (28)	T: Self-efficacy, TE St: Disability, grade, SN; Sc: Stress, SSP, WH	United States	D
McDow (1993)	426	NR	NR	NR	St: SN	United States	D
McIntyre (1984)	399	NR	NR	NR	T: Coping	United States	A
Moss (2014)	72	NR	1 M, 71 F	10.79	Sc: WH	United States	D
Platsidou (2010)	123	39.6	47 M, 79 F	6.2	T: Age, satisfaction, TE; Sc: EE	Greece	A
Platsidou and Agaliotis (2008)	127	39.6	49 M, 78 F	4 (68), 5–9 (32), 10–23 (27)	T: Satisfaction, stress	Greece	A
Ruble and McGrew (2013)	47	NR	NR	NR	St: Age, behavior problem, O; T: TE, stress, O; Sc: SSP	United States	A
Ruble et al. (2011)	35	NR	2 M, 33 F	10.6	T: TE; Sc: SSP, WH	United States	A
Sari (2005)	295	34.67	141 M, 154 F	NR	T: Age, coping, gender, satisfaction	Turkey	A
Sariçam and Sakiz (2014)	118	33.9	58 M, 60 F	NR	T: Self-efficacy	Turkey	A
Shea (1984)	215	20–69	215 F	1–35	St: Disability, SN; Sc: WH	United States	D
Shyman (2010)	100	21–71	11 M, 89 F	NR (3), 1 (20), 1–2 (20), 2–3 (25), 3 (32)	T: Self-efficacy; Sc: SSP, WH	United States	A
Weber and Toffler (1989)	68	34.5	17 M, 51 F	4.36	St: Age, SN, SES, O; T: Age, EL, TE; Sc: R, SSP	United States	A
Wong et al. (2017)	79	46	3 M, 76 F	11.02	T: Stress, other; St: O; Sc: WH	United States	A
R. H. Zabel and Zabel (2002)	298	NR	NR	NR	Sc: SSP	United States	A
Zarafshan et al. (2013)	93	36.81–40.29	93 F	8.53–19.32	St: Disability; T: Satisfaction, O; Sc: EE, SSP, WH	Iran	A

Note. M = male; F = female; T = teacher-related variable; EL = education level; TE = teaching experience; D = dissertation; NR = not reported; A = article; Sc = school-related variable; SSP = support from school personnel; EE = emotional experience; NSP = number of support professional; WH = work hindrance; R = resource; O = other; SN = student number; St = student-related variable; SES = socioeconomic status.

Figures 2 to 4 showed the forest plots of weighted effect sizes on each dimension of burnout. The overall effects as well as specific confidence intervals of correlation effects of each variable were described. Each figure demonstrates different ranges of effect sizes for each student-, teacher-, and school-related variables.

Figure 2. A forest plot of emotional exhaustion.
*Note.* CI = confidence interval; SES = socioeconomic status.

Figure 3. A forest plot of depersonalization.
*Note.* CI = confidence interval; SES = socioeconomic status.

Figure 4. A forest plot of personal accomplishment.
*Note.* CI = confidence interval; SES = socioeconomic status.

Publication Bias

As effect size calculation was performed by dividing the subcategories of emotional exhaustion, depersonalization, and personal accomplishment, publication bias was also divided into three. There was no evidence of publication bias in this study in terms of emotional exhaustion and depersonalization. There were zero studies in the Trim and Fill, and adjusted values were identical to the random effect size of observed values. There was also no evidence of personal accomplishment. The difference of effect sizes between observed value and an adjusted value was trivial, and only three values were trimmed.

Correlation Effects of Student-Related Variables

The effect size of student-related variables is displayed in Table 2. There were no variables related to significant levels of emotional exhaustion. Regarding depersonalization, only student’s age was significant with a small-to-medium effect size, 0.316 (95% CI = [0.127, 0.465]). Higher age was related to a significantly higher level of depersonalization. Other variables showed significant relations with level of personal accomplishment (Fisher’s Z = 0.107, 95% CI = [0.023, 0.191]).

Table 2. Effect Size of Student-Related Variables.

Burnout	Related variables	K	Fisher’s Z	SE	Q statistics (p)	Lower	Upper
						95% CI
Emotional exhaustion	Age	2	−0.048	0.096	.918	−0.236	0.140
	Behavior problem	5	0.087	0.107	.009	−0.124	0.297
	Disability	7	0.005	0.083	.000	−0.158	0.169
	Grade	2	−0.021	0.048	.119	−0.114	0.073
	Student number	21	0.015	0.047	.000	−0.078	0.107
	Other	9	−0.106	0.080	.000	−0.264	0.051
	SES	1	−0.030	0.124	1.000	−0.273	0.213
	Setting	2	−0.010	0.038	.297	−0.085	0.065
Depersonalization	Age	2	0.316	0.096	.094	0.127	0.465
	Behavior problem	4	0.121	0.074	.161	−0.023	0.259
	Disability	5	0.012	0.109	.000	−0.2201	0.225
	Grade	1	0.110	0.054	1.000	−0.306	0.491
	Student number	8	0.035	0.024	.285	−0.011	0.081
	Other	6	−0.075	0.043	.274	−0.159	0.009
	SES	1	−0.030	0.124	1.000	−0.273	0.213
	Setting	2	0.055	0.038	.240	−0.020	0.130
Personal accomplishment	Age	2	−0.104	0.096	.325	−0.292	0.083
	Behavior problem	4	−0.080	0.074	.349	−0.224	0.065
	Disability	5	0.016	0.083	.000	−0.147	0.178
	Grade	2	−0.060	0.048	.125	−0.154	0.033
	Student number	9	−0.004	0.023	.381	−0.049	0.041
	Other	6	0.107	0.043	.643	0.023	0.191
	SES	1	−0.192	0.124	1.000	−0.045	0.051
	Setting	2	−0.045	0.038	.696	−0.120	0.030

Note. Bold font indicates statistically significant results, where 95% CI does not include zero. K = number of effect size; CI = confidence interval; SES = socioeconomic status.

Correlation Effects of Teacher-Related Variables

The effect of teacher-related variables on burnout is presented in Table 3. Certain variables had a small effect on emotional exhaustion among teacher-related variables. The significant effect size was shown in other (Fisher’s Z = 0.068, 95% CI = [0.025, 0.110]), satisfaction (Fisher’s Z = −0.223, 95% CI = [−0.264, −0.182]), self-efficacy (Fisher’s Z = −.390, 95% CI = [−0.562, −0.218]), and stress (Fisher’s Z = 0.366, 95% CI = [0.270, 0.461]). Higher levels of satisfaction and self-efficacy were related to significantly lower emotional exhaustion, while a higher level of stress was related to significantly higher emotional exhaustion.

Table 3. Effect Size of Teacher-Related Variables.

Burnout	Related variables	K	Fisher’s Z	SE	Q statistics (p)	Lower	Upper
						95% CI
Emotional exhaustion	Other	12	0.068	0.022	.000	0.025	0.110
	Age	17	−0.034	0.055	.000	−0.142	0.074
	Coping	14	0.060	0.063	.000	−0.063	0.183
	Education level	9	−0.095	0.078	.009	−0.248	0.058
	Gender	5	0.011	0.095	.000	−0.175	0.196
	Teaching experience	33	−0.006	0.040	.023	−0.084	0.073
	Satisfaction	24	–0.223	0.021	.169	–0.264	–0.182
	Self-efficacy	7	–0.390	0.088	.000	–0.562	–0.218
	Stress	21	0.366	0.049	.000	0.270	0.461
Depersonalization	Other	8	0.207	0.093	.000	0.025	0.388
	Age	15	−0.019	0.064	.000	−0.144	0.107
	Coping	13	−0.036	0.071	.000	−0.176	0.104
	Education level	9	–0.110	0.037	.171	–0.183	–0.037
	Gender	5	0.015	0.105	.000	−0.191	0.221
	Teaching experience	31	−0.006	0.016	.134	−0.037	0.026
	Satisfaction	24	–0.263	0.053	.015	–0.368	–0.159
	Self-efficacy	5	–0.321	0.048	.346	–0.415	–0.227
	Stress	21	0.240	0.054	.000	0.135	0.345
Personal accomplishment	Other	8	0.091	0.070	.000	−0.047	0.229
	Age	16	0.007	0.048	.000	−0.087	0.101
	Coping	13	0.338	0.056	.000	0.229	0.448
	Education level	9	0.040	0.037	.826	−0.033	0.113
	Gender	5	−0.055	0.078	.000	−0.209	0.099
	Teaching experience	33	0.075	0.016	.257	0.044	0.106
	Satisfaction	25	0.027	0.014	.000	−0.053	0.107
	Self-efficacy	6	0.633	0.081	.003	0.474	0.793
	Stress	21	–0.110	0.041	.000	–0.191	–0.030

Note. Bold font indicates statistically significant results, where 95% CI does not include zero. K = number of effect size; CI = confidence interval.

Education level, satisfaction, self-efficacy, stress, and other displayed significant effect size on depersonalization. Specific significant effect sizes were: −0.110 of education level (95% CI = [−0.183, −0.037]), −.263 of satisfaction (95% CI = [−0.368, −0.159]), −0.321 of self-efficacy (95% CI = [−0.145, −0.227]), 0.240 of stress (95% CI = [0.135, 0.345]), and 0.207 of other (95% CI = [0.025, 0.388]). Higher levels of education, satisfaction, and self-efficacy were related to a significantly lower level of depersonalization, while a higher level of stress was related to a significantly lower level of depersonalization.

Regarding personal accomplishment, higher level of coping, self-efficacy, and long duration of teaching experience were related to significantly higher personal accomplishment. Moreover, a higher level of stress was related to a significantly lower level of personal accomplishment. Detailed effect sizes were 0.338 in coping (95% CI = [0.229, 0.448]), 0.075 in teaching experience (95% CI = [0.044, 0.106]), 0.633 in self-efficacy (95% CI = [0.474, 0.793]), and −0.110 in stress (95% CI = [−0.191, 0.030]).

Correlation Effects by School-Related Variables

The effect of school-related variables on burnout is presented in Table 4. The number of support personnel, support from school personnel, and work hindrances were significantly correlated with emotional exhaustion. A higher number of support personnel and level of work hindrances was related to significantly higher levels of emotional experience. The effect sizes were 0.036 (95% CI = [0.104, 0.507]) and 0.241 (95% CI = [0.184, 0.298]), respectively. A higher level of support from school personnel was related to a significantly lower level of emotional exhaustion (Fisher’s Z = −0.119, 95% CI = [−0.172, −0.065]). Regarding depersonalization, higher levels of emotional experience and support from school personnel were related to a significantly lower level of burnout. The effect sizes were −0.107 (95% CI = [−0.175, −0.038]) and −0.140 (95% CI = [−0.157, −0.122]), respectively. Higher levels of work hindrances were related to a significantly higher level of depersonalization (Fisher’s Z = 0.382, 95% CI = [0.353, 0.412]). Higher levels of emotional experience and support from school personnel were related to a significantly higher level of personal accomplishment. The effect sizes were 0.245 (95% CI = [0.126, 0.364]) and 0.172 (95% CI = [0.127, 0.217]), respectively.

Table 4. Effect Size of School-Related Variables.

Burnout	Related variables	K	Fisher’s Z	SE	Q statistics (p)	Lower	Upper
						95% CI
Emotional exhaustion	Emotional experience	12	−0.008	0.070	.000	−0.145	0.128
	Number of support personnel	2	0.306	0.103	.921	0.104	0.507
	Resources	3	−0.014	0.133	.003	−0.274	0.246
	Support from school personnel	66	–0.119	0.027	.000	–0.172	–0.065
	Work hindrances	60	0.241	0.029	.000	0.184	0.298
Depersonalization	Emotional experience	11	–0.107	0.035	.000	–0.175	–0.038
	Number of support personnel	2	0.112	0.103	.378	−0.090	0.313
	Resources	1	0.234	0.124	1.000	−0.009	0.477
	Support from school personnel	58	–0.140	0.009	.000	–0.157	–0.122
	Work hindrances	33	0.382	0.015	.000	0.353	0.412
Personal accomplishment	Emotional experience	11	0.245	0.061	.000	0.126	0.364
	Number of support personnel	2	−0.058	0.152	.005	−0.357	0.240
	Resources	1	0.010	0.124	1.000	−0.233	0.253
	Support from school personnel	59	0.172	0.023	.000	0.127	0.217
	Work hindrances	38	−0.005	0.031	.000	−0.065	0.055

Note. Bold font indicates statistically significant results, where 95% CI does not include zero. K = number of effect size; CI = confidence interval.

Discussion

This study aimed to verify the effect size of variables related to special education teachers’ burnout through meta-analysis. To achieve this, quantitative relations between burnout and its associated variables reported in previous studies were analyzed. Specifically, the relation of student-, teacher-, and school-related variables with three dimensions of burnout (emotional exhaustion, personal accomplishment, and depersonalization) were analyzed.

Correlation Effects by Student-Related Variables

There were a few variables such as age, disability type, and other that showed significant relations with level of burnout. Although grade showed a significant effect size in relation to overall burnout, the size was trivial. Furthermore, student age was not significantly associated with the emotional exhaustion or personal accomplishment of special education teacher; yet, their age showed a significant correlation with depersonalization. The results of this study differed from the three studies previously reviewed in this article, where the effects of student age on burnout were consistent (Brunsting et al., 2014). Carlson and Thompson (1995) found that teachers of older students experienced higher levels of depersonalization and lower levels of personal accomplishment, and Weber and Toffler (1989) reported that student age was a predictor of the increase in the emotional exhaustion.

Student disability type had a significant effect size on overall burnout and depersonalization. There were no significant variables related to emotional exhaustion. Examining the implications behind disability types affecting teacher burnout indicates that the problem behaviors of students, which differ according to disability type, affect burnout. Aloe et al. (2014) reported effects between misbehavior and emotional exhaustion (d = 0.44), depersonalization (d = 0.36), and personal accomplishment (d = −0.31). Special education teachers of students with emotional disorders experienced higher burnout than those of students with intellectual disabilities (Banks & Necco, 1990). Nichols and Sosnowsky (2002) also argued that the proportion of students with emotional disorders who exhibit challenging behaviors was associated with a higher intensity of teacher burnout. Special educators who teach disability types with a high frequency of problem behaviors are more likely to have difficulty controlling the classroom, and these difficulties may lead to burnout. This result can be interpreted based on previous research that teachers who have less classroom control report more burnout (d = 0.62; Fitchett et al., 2018).

Correlation Effects by Teacher-Related Variables

The fact that teacher-related variables influence burnout is an accepted result of previous studies (Brewer & McMahan, 2004; Coman et al., 2013; Gong et al., 2013; Sari, 2005). However, these studies do not consistently report the relation of burnout with teacher-related variables. For example, Fisher (2011) concluded that there was a significant difference in burnout between novice and experienced teachers among a sample of 400, with novice teachers displaying high levels of burnout. Carlson and Thompson (1995) reported that teaching experience did not significantly affect the degree of burnout. These inconsistent results may be attributed to the fact that they were analyzed without considering the dimensions of burnout.

In this meta-analysis, the correlation of each teacher-related variable was different across emotional exhaustion, personal accomplishment, and depersonalization. Special education teachers’ higher satisfaction and self-efficacy indicated significant relations to lower emotional exhaustion and depersonalization, and the direction of stress with emotional exhaustion was opposite. The level of satisfaction was not related to the level of personal accomplishment. Self-efficacy and stress among teacher-related variables demonstrated that the effect size was significant on all three dimensions. Given this discrepant correlation effect among the three dimensions of burnout, each component should be interpreted alone (Maslach et al., 2001).

Subsequent discussion focusing on self-efficacy and stress displayed medium-to-large effect size. Self-efficacy is known as an explained variable of teacher burnout. Teacher self-efficacy correlated negatively with both emotional exhaustion (.29) and depersonalization (.41) in structural equation modeling (Skaalvik & Skaalvik, 2010). The correlation study results reported a significant, medium, and negative correlation between teacher self-efficacy and burnout levels of 163 secondary school teachers (Savaş et al., 2014). The negative relation between self-efficacy and burnout was also confirmed in special education teachers. Ruble et al. (2011) reported an inverse correlation of special education teachers’ self-efficacy for students with autism spectrum disorder (ASD) with burnout. Increasing teachers’ self-efficacy (Savaş et al., 2014) is considered important for ensuring schools’ effectiveness. Based on the results of this meta-analysis, an increase in teachers’ self-efficacy is expected to have the maximum influence on enhancing personal accomplishment. These results are consistent with another meta-analysis analyzing associations between job burnout and self-efficacy in various professions (Shoji et al., 2016) and focusing on individual protective factors (Li et al., 2013). The association between special education teachers’ self-efficacy and burnout implies that self-efficacy may be a protective factor against burnout. These findings are in line with that of Jennett et al. (2003), in which teachers of students with ASD who had higher professional self-efficacy with greater commitment to their theoretical orientation of teaching approach is related to lower burnout dimension. Intervention such as global theoretical education can be a very important buffer against low teaching efficacy and high burnouts (Jennett et al., 2003).

Stress itself, a sudden failure of the mediation coping mechanism, or an inefficiency of the mediation coping mechanism over time can generate individual burnout (Brackett et al., 2010; Guglielmi & Tatrow, 1998). Montgomery and Rupp (2005) conducted a meta-analysis of the causes and effects of job-related stress on English-speaking teachers in North America. The relation of external stress and psychological burnout experienced by teachers indicated that active coping with stress and psychological burnout were of a 0.26 and 0.39 effect size, respectively. Teachers’ stress and burnout are positively correlated; teachers experiencing high stress levels are more likely to be exposed to greater rates of burnout. Teachers classified at risk for stress reported additional burnout symptoms (d = 1.48; Fitchett et al., 2018). Stress is a risk factor for teacher burnout, and stress reduction through protection such as self-efficacy improvement is necessary because the possibility of exhaustion increases when stress recurs. In particular, intervention programs such as stress management skills were helpful in reducing the emotional exhaustion of special education teacher (Cooley & Yovanoff, 1996).

Correlation Effects by School-Related Variables

In school-related variables, work hindrances, number of support personnel, and support from school personnel displayed significant small-to-medium effect size on emotional exhaustion; various effect sizes were reported about work hindrances (k = 60) and support from school personnel (k = 66). Significant meaningful variables were work hindrances. Work hindrances had a significant small-to-medium effect size on emotional exhaustion and depersonalization. Work hindrances increased emotional exhaustion, and depersonalization.

A higher level of support from school personnel was related to a decrease in burnout level. Although a small effect size was shown, the significance of support from school personnel was shown in all three dimensions. Recent meta-analysis by Iancu et al. (2018) highlighted the effects of social support such as peer collaboration groups on individual personal accomplishment at schools. Support from school personnel such as the principal and administrators were also frequently reported to be related to teachers’ burnout (Embich, 2001). A noticeable result with respect to support was the number of support personnel. The effect direction of number of support personnel was opposite to the support of school personnel in emotional exhaustion. This suggests that it is necessary to pay attention to qualitative aspects rather than quantitative aspects of support to reduce the level of special education teachers’ burnout.

Limitations and Recommendations for Future Research

There are some limitations to this meta-analysis. First, although we conducted a meta-analysis of special education teachers’ burnout and other related variables regarding students, teachers, and schools, these results only indicated how significantly these variables correlated with each other. We still could not estimate the causational relation between burnout and related variables. Thus, in future research, the effects of student-, teacher-, and school-related variables should be investigated by analyzing the experimental designs of intervention studies. Second, since the main interest of present meta-analysis was to assess special education teacher burnout associated with student-, teacher-, and school-related variables, we could not provide longitudinal evidence of why special education teachers refuse to keep their jobs. Future research should conduct a follow-up meta-analysis, considering how the correlation effect sizes from longitudinal studies could be mediated by time factors at multiple time points to further investigate whether certain types of variables aggravate or improve relations with special education teacher burnout rates. Third, we analyzed correlation effects by related variables (e.g., teaching experience, students’ disabilities, and support from school personnel) for each student-related, teacher-related, and school-related variable. However, these correlation effects are possibly mediated by different categorical variables (e.g., less than 5 years vs. greater than 5 years for teaching experience; ASD, learning disabilities, emotional and behavioral disorders for students’ disabilities; and principals, general education teachers, para-professionals for support from school personnel, cultural contexts). Thus, future research should conduct moderator analysis based on these categorical variables. Although the studies completed in various countries were included in this meta-analysis, we could not analyze the effect of cultural context including school contexts, school systems, regulations, and practices about special education. Fourth, when examining the 41 included studies, the number of correlation effect size (k) for each variable was less than 3 in several cases. Considering the limited published articles on related variables such as teacher resources and school personnel, generalizing the findings should be considered carefully. Thus, in the future, researchers should conduct studies focusing on burnout issues and related variables that are not being currently investigated.

Implications for Teacher Education and Training

The findings of this meta-analysis reveal four implications for teacher education and training. First, for teacher-related variables, special education teachers’ self-efficacy was significantly related to all three burnout scales (95% CI = [−0.562, −0.218] for emotional exhaustion, 95% CI = [−0.415, −0.227] for depersonalization, 95% CI = [0.474, 0.793] for personal accomplishment). Thus, when local school districts provide teacher training for special education teachers, the program content should emphasize the improvement of their self-efficacy level, which is closely related to their burnout degree. Second, student-related variables were found to be differently related to the three dimensions of burnout. A possible explanation is that student-related variables may differ more and more complex interaction according to the individual special education teacher’s contest than other variables. Individual and school-level intervention plans are needed to strengthen the capacity of teachers so that individual teachers can cope with their situation (Argentin et al., 2014; Caputo & Rastelli, 2014). Third, our results regarding school-related variables indicated that special education teachers’ support from school personnel was another variable that was significantly related to their burnout scale (95% CI = [−0.172, −0.065] for emotional exhaustion; 95% CI = [−0.157, −0.122] for depersonalization; 95% CI = [0.127, 0.217] for personnel accomplishment). Lack of support from school personnel has one of the most frequently cited relations with teachers’ burnout. Furthermore, the potential role of perceived support was also reported to be associated with reduced emotional exhaustion and improved personal accomplishment and with developing professional special education teachers (Langher et al., 2017). This can be seen by looking at the number of effect sizes analyzed in this meta-analysis. The number of effect sizes of support from school personnel was 66 in emotional exhaustion, 58 in depersonalization, and 59 in personnel accomplishment. Perceived lack of support may lead to feelings of burnout. Research has shown that special education teachers feel more supported by the school when they have administrators who value their opinions, provide effective feedback, and engage teachers in policy decisions (Bettini et al., 2015). When considering support from school personnel, schools should provide various types of support for special education teachers. Finally, the results revealed that in several cases, the correlation effects displayed various features for each burnout dimension: emotional exhaustion, depersonalization, and personal accomplishment. Given that these dimensions of emotional exhaustion, depersonalization, or personal accomplishment under the MBI (Maslach & Jackson, 1996) were constructed distinctively (Byrne, 1994; Denton et al., 2013), educators should avoid solely calculating total burnout that combines these three scores. The three burnout results should be considered individually when examining teachers’ burnout degrees.

Conclusion

The significant variables associated with special education teacher burnout rates were level of self-efficacy in teacher-related variables and support from school personnel in school-related variables. The effects of student-related variables vary depending on the extent of burnout, suggesting that the situation of individual teachers, including student characteristics, the evaluation of teacher burnout, and management, needs to be considered. The effect sizes of student-related variables were different across burnout dimensions. Teacher education and training programs that consider the significant variables, especially self-efficacy, might be effective in reducing special education teachers’ burnout levels. Regarding the results of school-related variables, effective support might be developed for special education teachers, including support from school personnel.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Mikyung Shin https://orcid.org/0000-0001-7907-9193

References

References marked with an asterisk (*) indicate studies included in the meta-analysis.

Adera B. A., Bullock L. M. (2010). Job stressors and teacher job satisfaction in programs serving students with emotional and behavioral disorders. Emotional and Behavioural Difficulties, 15, 5–14. https://doi.org/10.1080/13632750903512365

Abstract

Key Variables Associated With Special Education Teachers’ Burnout

Student-Related Variables

Teacher-Related Variables

School-Related Variables

Exploring Special Education Teachers’ Burnout in Previous Reviews and Syntheses

Method

Literature Search

Online search terms and their combinations

Coding Procedure and Inter-Rater Reliability

Data Analysis

Calculation of effect sizes

Statistical analyses

Results

Publication Bias

Correlation Effects of Student-Related Variables

Correlation Effects of Teacher-Related Variables

Correlation Effects by School-Related Variables

Discussion

Correlation Effects by Student-Related Variables

Correlation Effects by Teacher-Related Variables

Correlation Effects by School-Related Variables

Limitations and Recommendations for Future Research

Implications for Teacher Education and Training

Conclusion

Declaration of Conflicting Interests

Funding

ORCID iD

References

Cite article

Cite article

Download to reference manager

Share

Share this article

Share with email

Share on social media

Share access to this article

Information

Published In

Keywords

Rights and permissions

Authors

Affiliations

Notes

Metrics

Journals metrics

Article usage*

Altmetric

Articles citing this one

Figures & Media

Tables

View options

PDF/ePub

Get access

Access options

Sign in

Also from Sage

Article usage^*