Faith primary schools: better schools or better pupils?

We estimate the causal effect of attending a state Faith school on primary education achievement in England using administrative student-level data and implementing various strategies to control for students' selection into Faith schooling. Our regressions control for fixed effects in prior achievement and residential postcode to compare pupils who are close residential neighbors and have identical observable ability. We also use information on future school choices to control for preferences for Faith schooling. Results show that pupils progress faster in Faith primary schools, but all of this advantage is explained by sorting into Faith schools according to preexisting characteristics and preferences.