Abstract
Background Teaching combines intense interpersonal demands with administrative and digital workload, and in Kazakhstan more than four fifths of schoolteachers are women, which makes burnout in this group a gender-relevant occupational health and wellbeing concern. Evidence from Central Asia remains scarce, and the performance of the Russian-language Maslach Burnout Inventory-Educators Survey (MBI-ES) in this population had not been formally examined. Methods In a cross-sectional study conducted between October and December 2025, 323 women teachers recruited by convenience in general education schools in three Kazakhstani cities completed the MBI-ES, the state scale of the Spielberger State–Trait Anxiety Inventory, and a structured questionnaire covering employment, workload, social-psychological, lifestyle, and digital factors. We split the sample at random, examined the structure in one half and confirmed it in the other, compared the three-factor model against one-factor, two-factor, and bifactor alternatives, tested measurement invariance, and fitted hierarchical regression models built on the job demands-resources framework. Results The correlated three-factor model outperformed one-factor and two-factor alternatives and fitted the validation subsample well ( χ 2 /df = 1.69; CFI = 0.955; RMSEA = 0.065; SRMR = 0.080). A bifactor model fitted better on global indices but yielded a degenerate exhaustion-specific factor and negative general loadings for accomplishment items, so we retained three factors. Loadings and intercepts were invariant across the random halves; across cities loadings held but intercepts did not. Subscale reliability was high (α 0.918 to 0.963). Job demands added 35.1% of explained variance in emotional exhaustion beyond background characteristics, resources a further 6.6%; perceived work stress, state anxiety, and low physical activity remained independent correlates in the full model (adjusted R 2 = 0.630). Depersonalization was associated with conflict, work stress, isolation, and an unfavorable collegial climate (adjusted R 2 = 0.443), and personal accomplishment gave the weakest model (adjusted R 2 = 0.170). Demographic and formal workload characteristics contributed little. Conclusion The Russian-language MBI-ES performs acceptably in women schoolteachers in Kazakhstan. In this convenience sample, burnout was associated more closely with appraised psychosocial working conditions than with formal job parameters. The cross-sectional design supports no causal inference, and the findings apply to women teachers in three urban settings.