Abstract
School suspensions are associated with adverse educational and psychosocial outcomes, yet little is known about how structural disadvantage, relational factors, health behaviours, well-being indicators and school-level factors jointly predict suspension risk in England. To our knowledge, this is the first study in England to model structural, relational, health-behaviour, well-being and school-level predictors of suspension simultaneously within a single ecological framework. We draw on linked psychosocial survey and administrative data and separately estimate predictors of initial suspension and of repeated suspension. The study extends a literature dominated by demographic analyses of national administrative data and provides evidence for both established and previously underexamined associations. We analysed #BeeWell survey and linked administrative suspension data from pupils aged 12–15 across Hampshire and Southampton (N = 19,064, attending 69 secondary schools). Bayesian zero-inflated negative binomial regression was used to model both the likelihood of receiving at least one suspension and the frequency of suspensions among those suspended. Predictor domains included structural disadvantage (e.g., familial socio-economic disadvantage), relational factors (e.g., peer pressure), health behaviours (e.g., substance use), well-being indicators (e.g., stress), and school-level factors (e.g., school belonging). Pupils eligible for free school meals, with special educational needs, and living in more deprived areas had markedly higher suspension risk. Greater school belonging, positive relationships with staff, academic satisfaction and emotion regulation were protective. Unexpectedly, higher perceived school pressure and greater emotional difficulties were associated with lower suspension risk. Receipt of school-based mental health support was linked to increased suspension risk, even after adjustment for covariates. School-level suspension rates independently predicted an individual pupil’s likelihood of being suspended, over and above their own characteristics. No significant differences by ethnic group were observed after covariate adjustment. Predictors of initial suspension (specifically, free school meal eligibility and school-level suspension rates) also predicted multiple suspensions. A variety of factors jointly shape suspension risk. Findings highlight the need for early, preventative interventions.