An individual’s level of education is increasingly significant in explaining their political attitudes and behaviour, but why education should have this effect is not well understood. Competing explanations include selection effects (confounding by pre-adult characteristics such as family background) and allocation effects (confounding by post-treatment covariates such as socio-economic position). This new analysis of 1970 British Cohort Study data uses genetic matching and sequential g-estimation to estimate the absolute controlled direct effect (ACDE) of higher education on racial prejudice, social authoritarianism and economic Left-Right values, adjusting for these selection and allocation effects. It finds that attending university leads to significant reductions in racial prejudice and social authoritarianism, while the effects on economic values are attributed to economic allocation. It further extends knowledge by examining the mechanism through which the change happens: whether this is attributable to cognitive change or educational socialisation, by modelling variation in educational experience through inverse probability weighting techniques. Given the impossibility of operating a randomised field experiment to assign individuals to different educational pathways, the methods employed in this study likely produce the best, most unbiased, estimate of the effect of university on racial prejudice and political values.