The Impact of Teacher Burnout on Student’s Non-Academic Outcomes

• In Event: Mental health and school safety: Current situations and implied solutions

Proposal

Teaching is a stressful job with a high risk of burnout (Hakanen et al. 2006; Maslach et al. 2001). The impact of burnout is consequential, directly relating to the students’ learning environment. However, many studies have been focused on factors associated with burnout, rather than its consequences. In particular, the impact of burnout on students’ non-academic skills (i.e., social-emotional skills, executive function, and social competence) is rarely studied. Given that burnout is a known phenomenon amongst teachers, especially early childhood educators and in low-income countries (UNESCO, 2015; Salifu, 2014; Akyeampong et al., 2007), more attention is warranted to understand its consequences for student outcomes. This study examines the relationship between teacher burnout and students’ development in non-academic skills, controlling for other confounding variables, in the Ghanaian primary school context using both propensity score matching methods and generalized propensity score (GPS) models.
The samples for this study are drawn from the 3-year follow-up survey of a randomized control trial, called Quality Preschool for Ghana (QP4G). The intervention was designed to improve ECE in Ghana by providing rigorous teacher training programs and a parental awareness campaign (Wolf et al., 2019). The follow-up data was collected in May-June 2018, from students who had participated in the intervention (n = 2,777), and their primary school teacher (n = 949). Students and teachers in 520 schools were recruited, and each teacher had 1 to 18 students in the intervention.
Maslach Burnout Inventory (MBI-ES) (Maslach et al, 1986) was used to measure teacher burnout, which was embedded in the follow-up survey. Teachers responded to their frequency of feelings related to each MBI-ES scale, which was a seven-point Likert scale ranging from 1 (“never”) to 7 (“every day”), with higher scores representing higher levels of burnout. Exploratory and confirmatory factor analyses were conducted. Consistent with previous studies on teacher burnout on Ghanaian kindergarten teachers (Lee & Wolf, 2019), we found that there were two distinct dimensions for Ghanaian primary school teachers, which were emotional exhaustion (EE) and lack of personal accomplishment (LOPA) (Table 1). We could not find the third dimension, depersonalization, unlike other teacher burnout studies (Maslach et al., 1997).
Social-emotional skills (SE), executive function (EF) and social competence were measured as a student outcome in non-academic skills. SE and EF were measured through direct assessments using the International Development and Early Learning Assessment (IDELA) subscales, which were modified for the Ghanaian context (Pisani et al., 2018). Social competence was measured using the Social Competence Scale (Conduct Problems Prevention Research Group, 1990) and reported by teachers.
The covariates selected in this study, such as teacher characteristics, their perception of job support, school environment, and students’ characteristics were based on previous studies of teacher burnout and student outcomes (Table 2).
We first examine t-tests to determine the extent to which confounding covariates, the burnout variable, and the outcome variables were related to each other. The results support the need to balance the two groups (high-burnout level vs. low-burnout level (top and bottom 25% in burnout measures) before estimating the burnout effect (Table 3). We matched two groups on a rank-based Mahalanobis distance with a propensity score caliper, modeling burnout as a binary “treatment”. Then, with the paired group, we estimate the burnout effect using multiple regression, adjusting for clustered standard errors. Because of the nested nature in this dataset, we accounted for the nesting of students within classrooms and school (highest cluster level ICC range from 0.13 to 0.28) using the cluster robust standard error from the sandwich estimator in addition to the 95% confidence interval from 1,000 bootstrap samples.
The full matching yielded a balance of covariates between students in the high-level and low-level teacher burnout classrooms in both dimensions (standard mean difference (SMD) < 0.2, Table 3). The results identified a negative, statistically significant effect of teachers with higher burnout in the LOPA dimension on students’ social competence skill sets (Table 5). Teachers’ emotional exhaustion also had a marginally statistically significant effect on students’ executive functioning. We did not find any other evidence of burnout’s effect on students.
Although a binary approach may give an easier interpretation, it may reduce the power of detecting the treatment effect and underestimate the extent of variation in the outcomes, due to the continuous nature of burnout (Altman, 2006). Therefore, we also estimate the burnout effect on students using a generalized propensity score model (GPS) and inverse probability weighting (IPW), considering burnout as a continuous treatment (Hirano & Imbens, 2004).
The balance was successful (Pearson correlations after weighting < 0.1, Table 4) for GPS model. Although the magnitude of the treatment effects was small, we found evidence of a significant negative effect of higher teacher burnout in both dimensions. For teacher burnout in emotional exhaustion, there was a negative effect on students’ executive functioning (τ=-0.02) and social competence skills (τ=-0.0012). LOPA burnout also showed a negative effect on social competence skills (τ=-0.0014) (Table 5).
It is important to consider some limitations of the study. First, we note that although we used the propensity score approach to eliminate any possible confounding effects, it is quite hard to claim that what we found is a causal relationship. Secondly, although there is a certain relationship between the teacher burnout and student’s social competence, the effect of this area was stronger because the measure of the area was dependent on the teachers’ observation.
Despite these limitations, this study made an innovative attempt to disentangle the relationship between teacher burnout and its consequences. As far as we know, this kind of approach has not been performed, and this is one of the first empirical studies that attempt to reveal any burnout effect related to students’ non-academic outcomes, especially in a low-income country’s context. Regardless of the definition of burnout, whether it is considered as binary or continuous, we could find that teachers’ perceived LOPA could be associated with students’ social competence skills, and EE could relate to the development in students’ executive function.

Authors

• Jimin Oh, University of Pennsylvania

• Sharon Wolf, University of Pennsylvania