Work

Predictive modelling of Australian school principals’ turnover intentions using machine learning with random effects

Year: 2026

Type: article

Abstract: This study identifies key predictors of Australian school principals' turnover intention, using a machine learning approach by integrating extreme gradient boosting (XGBoost) with random effects. Drawing on data from an annual survey across two waves, the study analyses responses from 1,630 Australian principals. The prediction model includes 89 psychosocial, health-related, and demographic factors. Results identified commitment to the workplace, job satisfaction, and cognitive demands as the strongest predictors of school principals’ turnover intention. The proposed model with random effects outperforms baseline methods, demonstrating effectiveness in capturing hierarchical and non-linear relationships. The findings provide actionable insights for policymakers and educational stakeholders to improve principal retention and enhance the sustainability of school leadership roles. This research underscores the potential of advanced analytics to inform targeted interventions and policy frameworks in educational leadership.

Source: Discover Computing

DOI: 10.1007/s10791-025-09838-1

Authors: Jiesi Guo, Danling Huang, Jinran Wu, Hamed Mogouie, Theresa Dicke

Institution: Australian Catholic University

Language: English

Topic: School Leadership and Teacher Performance

Subfield: Education

Field: Social Sciences

Domain: Social Sciences

SDG: Quality education

Open Access status: gold

Estimated APC paid: $1,520