Teacher efficacy in student engagement, instructional management, student stressors, and burnout: A theoretical model using in-class variables to predict teachers' intent-to-leave
Highlights
► We examine measurement invariance for eight scales across gender and grade level. ► We tested for latent factor mean differences between gender and grade level. ► We present models that considered relationships between eight variables. ► We present a relatively invariant model that predicts teachers' intent-to-leave.
Introduction
Teacher turnover is detrimental to the profession in very real and quantifiable ways. As many as one quarter of novice educators in the United States (U.S.) leave the profession within three years. By the fifth year, attrition increases with almost 40% departing the field (Milner & Woolfolk Hoy, 2003; National Center for Education Statistics, 2004; National Commission on Teaching and America's Future, 2003; Smith & Ingersoll, 2004). High attrition has been conservatively estimated to cost thousands of dollars for each teacher who leaves the profession, with the cumulative expense totaling millions of dollars per year (Darling-Hammond, 2003). Concerns regarding teacher turnover appear to be worldwide as a plethora of international studies yields similar results: Africa (Schroeder, Akotia, & Apekey, 2001), Australia (Pillay, Goddard, & Wilss, 2005); Asia (Leung & Lee, 2006); Europe (Hakanen, Bakker, & Schaufeli, 2006), and North America (Perrachione, Rosser, & Petersen, 2008).
Teacher migration within the profession is also cause for concern as it fosters instability among the faculty and hinders on-going professional development and team building efforts (Guin, 2004). However, the greatest cost is the detrimental impact on student learning (Darling-Hammond, 2003; Guin, 2004). Understanding and identifying the variables contributing to this phenomenon is an essential step in creating effective teacher retention and job satisfaction programs. However, before such programs can be created it is important to develop theoretically sound predictive models of teacher intent-to-leave.
Intent-to-leave can result in decreased school effectiveness (Ingersoll, 2001; Shaw, Gupta, & Delery, 2005) since it is likely characterized by reduced work effort (Maertz & Campion, 1998) as well as decreased productivity, motivation, and overall school morale (Tett & Meyer, 1993). Though intent-to-leave is generally considered a precursor to actual voluntary resignation (Ladebo, 2005; Price & Mueller, 1986; Rosser & Townsend, 2006; Steel & Ovalle, 1984; Tett & Meyer, 1993), empirical evidence of the relationship between the two is rare. However, Johnsrud and Russer (2002) determined a significant statistical connection between intent-to-leave and actual turnover. Thus, the value of studying teachers' intent-to-leave cannot be overestimated.
A sizable body of research exists related to teacher turnover and variables outside the classroom or outside the teacher's control (e.g., social support from superiors and colleagues, Betoret, 2006; Rieg, Paquette, & Chen, 2007; salary or the school's socioeconomic status, Garcia, Slate, & Delgado, 2009; Hancock, 2008; Malow-Iroff, O'Connor & Bisland, 2004). However, teachers are largely isolated from on-campus interactions with peers and other adults and spend a majority of their time with students. Thus, events occurring inside the classroom likely have substantial impact on teachers and merit in-depth study. Pines' (2002) cross-cultural comparison of Israeli and American teachers supports the notion that the classroom is the central part of a teacher's work environment and that teachers enter the profession with the goal of making a difference in their students' lives. Therefore, it is likely that variables inside the classroom have a significant impact on teacher turnover.
Further, we cannot assume that all teachers are alike or that one solution fits all. For example, models that predict teachers' intent-to-leave might present different prediction rates for elementary school teachers than high school teachers due to a different student population and instructional techniques. Therefore it is important to consider teacher sub-populations such as gender and grade level.
While few studies have explored whether the relationships between variables differ based on gender and grade level, several studies have explored mean differences between these groups. In particular, research has discerned differences between teacher gender regarding levels of efficacy, controlling instructional approaches, stress, burnout, job satisfaction, and attrition. In both Canada and Belgium female teachers have been found to experience lower classroom management efficacy and greater stress from student behaviors, yet greater job satisfaction than their male counterparts (Klassen & Chiu, 2010; Sabbe & Aelterman, 2007). Regarding burnout, males have been found to experience significantly less emotional exhaustion and personal accomplishment, but higher levels of depersonalization in the U.S., Belgium, and Turkey, respectively (Grayson & Alvarez, 2008; Sabbe & Aelterman, 2007; Sari, 2004), and lower levels of retention (Marso & Pigge, 1997). Using a Hong Kong sample, males have also been found more controlling in their instructional approaches than females (Lam, Tse, Lam, & Loh, 2010). Due to previous variations associated with gender, this study also considered gender as a moderator variable to investigate whether it influenced the associations between variables explored in our model.
Research also suggests mean differences exist between Canadian and U.S. elementary and upper-level teachers' approaches to instruction and efficacy in student engagement (Klassen & Chiu, 2010; Martin & Baldwin, 1996; Wolters & Daugherty, 2007). For example, Martin and Baldwin's (1996) study revealed elementary teachers to be significantly less controlling than secondary-level teachers regarding classroom management (defined as including approaches to instruction). Wolters and Daugherty (2007) provided evidence that teachers' instructional foci shift from mastery learning at the elementary level to more competitive, performance-based approaches as grade level increases. Klassen and Chiu (2010) discerned elementary teachers were significantly higher regarding efficacy in student engagement than upper-grade teachers. Regarding the depersonalization component of burnout and grade level, students across the grade spectrum reportedly perceive teachers as becoming more rigid, and less caring and supportive of their emotional needs (Anderman, 2003; Wentzel & Wigfield, 1998). Further, teacher motivational beliefs and attitudes become less tailored to students as grade level increases (Anderman & Maehr, 1994; Anderman & Midgley, 1997; Wigfield, 1994; Wigfield, Eccles, & Rodriguez, 1998), indicating a more calloused attitude (and heightened depersonalization) toward students. However, to date, no research has tested whether grade level moderates the relationship between these in-class variables. Building on previous research, this study explored whether grade level moderated our theorized model and tested for mean differences.
The aforementioned variables' importance is not limited to mean differences, as this variation does not predict teachers' intent-to-leave or migrate. Therefore, the purpose of our studies is to explore the connections between variables that occur inside the classroom and impact teachers' intent-to-leave. To ensure an appropriate statistical comparison across gender and grade level, Study 1 investigated the psychometric properties and measurement invariance for each variable in the theoretical model and tested for latent factor mean differences. Study 2 investigated an inclusive model (see Fig. 1) that integrates international teacher research and considered the complex relationships between in-class variables and whether teacher gender and grade level moderated the structural coefficients. As cited in Fig. 1, independent, worldwide research supports the theoretical model; however, less is known regarding moderation effects. Both studies build on previous research by investigating these variables in an integrative model framework and, perhaps more importantly, consider whether these relationships are moderated by teacher gender and grade level.
Section snippets
Efficacy in student engagement, instructional management, personal accomplishment, and student behavior stressors
Teachers' general efficacy beliefs, defined as “…judgments about their abilities to promote students' learning” (Woolfolk Hoy, & Spero, 2005, p. 343), shape how teachers behave in the classroom and have consistently been found to impact the learning environment. Teacher efficacy is generally considered specific to a given context or task and has been further focused on a variety of areas, including student engagement (Tschannen-Moran & Woolfolk Hoy, 2001). This particular aspect of efficacy
Significance of the studies
While a large body of international research exists regarding the antecedents of teacher stress and burnout (Chang, 2009), validated theoretical models focused on in-class variables predicting teacher burnout and intent-to-leave are scarce. Moreover, few studies have tested the generalizability of such models across different groups (e.g., gender, grade level, etc.). Examining moderation effects across groups and mediating variables related to intent-to-leave will provide a valuable step toward
Participants
Data were collected from n = 631 certified teachers employed by three public school districts in the southwestern U.S. The majority of participants were females (n = 488, 78.8%) employed by urban school districts (92.7%). The participants were 2.7% African-American (n = 17), 0.8% Asian (n = 5), 57.1% Caucasian (n = 360), 35.4% Hispanic (n = 224), 0.6% Native American (n = 4), and 1.7% reported ethnicity as “other” (n = 11). Participants' age ranged from 22 to 78, with a mean age of 41.78 years (
Factorial validity, measurement invariance, and internal consistency reliability
Results revealed that the CFA model fit statistics provided strong evidence of factorial validity across gender (see Table 2) and grade levels (see Table 3), while possessing consistently large (λ > 0.35) and statistically (p > 0.0001) significant standardized factor loadings. Both the configural and measurement invariance models (see Table 2, Table 3) also fit well, confirming that latent factor mean comparisons and structural invariance tests were appropriate for statistical comparisons
Structural model across samples
Prior to invariance testing, the structural model was tested for the entire sample to validate the theoretical model of teacher-level variables (see Fig. 1). Results revealed reasonable support for the proposed model based on the model fit statistics, χ2 (1166) = 3725.23, p < 0.0001, CFI = 0.936, TLI = 0.933, RMSEA = 0.059, although a slight improvement was suggested based on the modification indices. Estimating the relationship from Personal Accomplishment to Depersonalization significantly
Summary and conclusion
From a measurement perspective, Study 1 results are promising as verification was obtained for measurement invariance across the eight factors and internal consistency reliability estimates were reasonably stable. Although mean differences emerged between the groups on the eight factors, the largest differences often occurred between elementary and middle school teachers, with fewer differences between elementary and high school teachers and teacher gender. Comparatively, elementary school
Conclusion
High teacher attrition has expensive and widespread implications (Darling-Hammond, 2003). Beyond the vast budgetary strain it creates, the high rate of teacher turnover undermines trust in the public schools and negatively impacts both teachers and students. Novice teachers who leave the profession sacrifice the considerable time, effort, and money invested to achieve their teaching credential. Moreover, experienced educators are continuously expected to mentor the influx of newcomers in
References (91)
- et al.
Changes in achievement goal orientations, perceived academic competence, and grades across the transition to middle-level school
Contemporary Educational Psychology
(1997) - et al.
School climate factors relating to teacher burnout: a mediator model
Teaching and Teacher Education
(2008) - et al.
Burnout and work engagement among teachers
Journal of School Psychology
(2006) - et al.
Exploring the validity of a teachers' self-efficacy scale in five countries
Contemporary Educational Psychology
(2009) - et al.
Teachers' job satisfaction: analyses of the teacher follow-up survey in the United States for 2000–2001
Teaching and Teacher Education
(2008) - et al.
Construct validation of the behavior and instructional management scale
Teaching & Teacher Education
(2010) - et al.
Teacher self-efficacy and retaining talented teachers: a case study of an African American teacher
Teaching and Teacher Education
(2003) Causes and effects of teacher conflict-inducing attitudes towards pupils: a path analysis model
Teaching and Teacher Education
(2002)- et al.
Does school context matter? Relations with teacher burnout and job satisfaction
Teaching and Teacher Education
(2009) - et al.
Teacher efficacy: capturing an elusive construct
Teaching and Teacher Education
(2001)