A wealth of research has found that attending university will result in an individual being more economically right-wing and socially liberal (Surridge 2016), and that these differences are increasingly important for explaining vote choice (Fieldhouse et al 2020). Yet given that educational pathways are not randomly assigned, determining the causal effect of education is notoriously challenging. Previous research has suggested that where an education effect has occurred, this is attributable to peer socialisation (Mendelberg et al., 2017): but it’s not clear whether this is due to adapting views to the immediate context, or a phenomenon which holds over the longer-term. This new analysis applies the random-effects ‘within-between’ model (REWB) (Bell, Fairbrother and Jones 2019) and spline regression techniques to British Election Study Internet Panel data to estimate the change within individuals as they go through educational transitions and disentangle the contextual and socialisation effects. In this way, it provides a low-bias causal estimate on an important political question of our time.