School systems around the world struggle to retain effective teachers (See et al., 2020). Changes in society and the role of schools, such as increasing diversity, changing expectations, digitalization, have magnified teacher retention problems. Research has historically focused on identifying general correlations between teacher job satisfaction and intention to quit and potentially explanatory variables (e.g., Madigan & Kim, 2021). But a decision to quit the profession is personal and highly complex. For example, beginning teachers may struggle to adjust to the praxis shock (Gallant & Riley, 2017) while experienced teachers’ intention to quit may be driven by stresses associated with shifting student populations. This paper seeks to try to build a deeper understanding of the different potential motivating factors that might lead teachers to quit the profession. I use data from TALIS 2018 in the Nordic countries to explore the predictors of teacher’s intention to quit. TALIS contains many potentially explanatory variables, such as the quality of teacher education, the distribution of teachers’ workload (i.e., the number of hours spent on different tasks), sources of teacher stress, teachers’ access to professional development, and teacher’s career stage and current position. I explore the association of these predictors with teachers’ intention to quit using Bayesian additive regression trees (BART; Kapelner & Bleich, 2016), a machine learning algorithm. BART allows for complex interactions while avoiding over-fitting, allowing explorations of how teachers’ school and work contexts relate to their intention to quit for different subgroups of teachers. Results are explored using variable importance measures and through graphical displays that subset different groups of teachers. Besides aspects of teacher’s career status (e.g., age, year of initial teacher education), overall job satisfaction, teachers’ initial reason for becoming a teacher, experienced stress, and teachers’ receipt of impactful feedback showed the strongest association with intention to quit. Interestingly, though, job satisfaction, experienced stress, and having received impactful feedback were much stronger predictors of the amount of time teachers will remain teachers for younger and less experienced teachers than for older teachers, which might suggest the importance of helping young teachers feel like they are managing the demands of their job. Teachers in long-term, but non-permanent positions, appeared to benefit the most from receiving impactful feedback, which might suggest that teachers in unstable positions could benefit from skill development. Overall, these patterns suggest that teacher retention efforts will need to account for teacher’s career stage

Kapelner, A. & Bleich, J. (2016). bartMachine: Machine Learning with Bayesian Additive Regression Trees. Journal of Statistical Software, 70(4), 1-40. doi:10.18637/jss.v070.i04 Gallant, A., & Riley, P. (2017). Early career teacher attrition in Australia: Inconvenient truths about new public management. Teachers and Teaching, 23(8), 896-913. Madigan, D. J., & Kim, L. E. (2021). Towards an understanding of teacher attrition: A meta-analysis of burnout, job satisfaction, and teachers’ intentions to quit. Teaching and teacher education, 105. See, B. H., Morris, R., Gorard, S., Kokotsaki, D., & Abdi, S. (2020). Teacher recruitment and retention: A critical review of international evidence of most promising interventions. Education Sciences, 10(10), 262