Gail Swingler, MA, The Graduate Center of the City University of New York; Irvin Sam Schonfeld, PhD, MPH, The City College and the Graduate Center of of the City University of New York; Jay Verkuilen, PhD, The Graduate Center of of the City University of New York; Renzo Bianchi, PhD, University of Neuch?tel
Burnout has been defined as a workplace problem comprising three components, emotional exhaustion (EE), depersonalization (DP), and a reduced sense of professional accomplishment (rPA)(Maslach et al., 2016). Researchers have found high rates of burnout in teachers (Quattrin et al., 2010; Li et al., 2020). Questions, however, have arisen regarding whether burnout is a separate construct or a depressive condition (Bianchi et al., 2015). Of particular concern is the relation of EE, considered burnout?s core (Halbesleben & Demerouti, 2005; Kristensen et al., 2005; Maslach et al., 2016), to depression (Schonfeld, 2019b). Teachers are the focal interest of this meta-analysis bearing on burnout?s relationship with depression. The meta-analysis centered on the Maslach Burnout Inventory (MBI) because it is the measure of reference in burnout research (Schonfeld, 2019a). The meta-analysis?s specific purpose is to answer the following questions. 1. Given research involving teachers, what is the average correlation of each MBI subscale with depression? 2. How does the correlation between EE and depression compare to the (a) correlation of DP and rPA with depression and (b) the intercorrelations among EE, DP, and rPA? 3. What do these findings tell us about the widely-held view that burnout has a three-factor structure as defined by Maslach and her colleagues (2016)? 4. What are the implications of the findings for the treatment for burnout in teachers?
Some research has indicated that burnout is distinct from depression (Leiter & Durup, 1994; Bakker et al., 2000). Other research shows significant overlap between burnout and depression (Bianchi et al., 2013, Bianchi et al., 2014; Bianchi et al., 2015; Bianchi et al., 2016; Meier, 1984; Schonfeld, 1991; Schonfeld & Bianchi, 2016; Schonfeld, 2019a, b). Meta-analytic evidence (Bianchi et al., 2021; Koutsimani et al., 2019; Meier & Kim, 2021; Schonfeld et al., 2019a) regarding burnout-depression overlap has been interpreted as reflecting both the distinctness of burnout from depression and the overlap. To conserve space, literature cited is available upon request.
To be included in the meta-analysis, studies had to involve educators who completed at least one of the MBI subscales and a validated depression scale. The study also had to be published in English and include at least one correlation between an MBI subscale and a valid depression scale. Exclusion criteria were (a) a non-English publication, (b) duplicates, (c) non-educator samples, (d) use a non-MBI burnout scale, (e) absence of a valid depression scale, or (f) non-reporting of the correlation between any MBI subscale and a valid depression scale. The keywords guiding the search included educators (teacher, educator, faculty, or school staff) as well as depress*, burnout, Maslach, and MBI. Databases searched included APA PsycInfo, Academic Search Complete, APA PsycArticles, APA PsycBooks, ERIC, Health and Psychosocial Instruments, MEDLINE Complete, SocINDEX with Full Test, Teacher Reference Center, and Urban Studies Abstracts. Dr. Robert Boudreau, Amanda Mauthe, and Wyatt Boudreau from the University of Lethbridge searched their databases for teachers, and MBI, and BDI, CES-D and/or PHQ-9 and forwarded to us relevant references and abstracts. Analyses: Using the metafor package (Viechtbauer, 2010) in R version 3.5.2, random-effects meta-analyses were applied to the correlations of the subscale scores with depression scores and the intercorrelations among the MBI subscales. The conservative Sidik-Jonkman estimator (Harrer et al., 2019) was used to determine the heterogeneity between the studies, ?^2. Each correlation was transformed into a Fisher?s z-score. The z-scores were pooled and averaged. The averages were then back-transformed to average rs. To examine publication bias, Duval and Tweedie?s (2000) trim-and-fill procedure was applied.
Of 278 studies found, eleven have been eligible so far (n = 11,729). In the coming months, we expect to include two or three additional studies. The results thus far show that the average correlation between EE and depression, r = 0.68 (disattenuated r = 0.76), was notably higher than that of DP with depression, r = 0.45 (disattenuated r = 0.56), rPA with depression, r = 0.40 (disattenuated r = 0.47), EE with DP, r = 0.52 (disattenuated r = 0.64), EE with rPA, r = 0.41 (disattenuated r = 0.48), and DP with rPA was r = 0.43 (disattenuated r = 0.56). All correlations were significant p < .01. We found no evidence of publication bias.
The EE-depression correlation was stronger than any other relationship. The three-part conceptualization (EE, DP, and rPA) of burnout does not hold up in educators because of the close relationship between EE and depression. Limitations include the relatively small number of studies and overlap in authorship in several studies. Strengths include the meta-analysis our limiting the meta-analysis to studies that employed validated depression scales.
An implication of our finding significant overlap in burnout?s exhaustion core with depression is that teachers suffering from burnout could be helped by clinicians who specialize in treating depression. We believe this approach can help reduce teacher attrition.