Tens of thousands of young people leave school with no or very few qualifications in England. This paper seeks to build a fuller picture of Key Stage 4 low achievement and its correlates than available hitherto. We focus on three aspects. Firstly, the role of students' personal characteristics, especially gender, ethnicity and past achievement, in explaining the incidence of low achievement at age 16. Secondly, we investigate the extent to which particular personal characteristics constitute direct risk factors for low achievement and the extent to which they lead to low achievement because of their correlation with unobserved school and neighborhood quality, i.e. the role of sorting into schools and neighborhoods of different quality. We suggest a method of calculating school quality (how effective a school is in helping its pupils to avoid low achievement) which is akin to the value-added concept, and examine which specific observed school characteristics predict this measure of 'school quality'. Thirdly, the paper examines the relationship between school resources - particularly per pupil expenditure - and the avoidance of low achievement, exploiting the panel nature of the National Pupil Database. Going beyond simple discrete choice models, the paper employs school fixed effects regression to reduce endogeneity problems and employs panel data at the student level to analyse school resource effects. A number of interesting findings emerge about the correlates of low achievement and of school quality, and we consider the policy implications of our findings.
