There are few longitudinal studies of adolescent students' choice to persist in post-compulsory education. Hence, the present study introduces a longitudinal model that describes the interplay between sociological and psychological explanations of adolescents' choice to persist in post-compulsory education in the UK. Data on parental education, early childhood self-regulation (age 5), sustained school engagement (ages 7, 11, 14) and persistence in education after the end of compulsory schooling (age 17) were utilised. The sample comprised 8333 (51.1% females, 89.5% white) children from the Millennium Cohort Study. Statistical analyses included state-trait modelling, longitudinal mediation and multigroup moderation. A trait-state-occasion model was run to disentangle the trait from state variance in school engagement. Afterwards, two hypotheses were formulated, namely the 'instilment' and the 'differential'. The 'instilment' hypothesis involved a longitudinal predictive model, whereby parental education predicted early childhood self-regulation which, in turn, predicted sustained school engagement which predicted students' choice to persist. The 'differential' hypothesis examined whether higher vs. lower parental education changed the nature of the predictive relations between self-regulation, sustained school engagement and persistence. The results were in favour of an 'instilment' hypothesis, whereby higher parental education was translated to higher levels of early self-regulation which predicted higher sustained engagement, which, in turn, predicted greater probability of persisting in post-compulsory education. The findings suggest a pathway from early childhood experiences to educational outcomes via the development of a trait of engaging with school.