The Matthew effect in Dutch primary education, differences between school, cohorts and pupils

Secondary analysis of longitudinal data from Dutch primary education was used to assess to what extent differences between educationally disadvantaged pupils and other pupils increase during their school careers. For language and arithmetic the differences between pupils with poorly educated and well-educated parents increases, while the disadvantage of Turkish and Moroccan pupils in non-verbal intelligence slightly decreases. Most progress was made in schools and cohorts with a low starting level. Differences in language achievement between schools were greatest where pupils of poorly educated parents were in a setting with many other disadvantaged pupils. Several cohorts per school were included in our analyses so that the development of schools could be examined. However, no significant differences were found between cohorts within schools regarding the size of the Matthew effect.