Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
We study ability peer effects in English secondary schools using data on four cohorts of students taking age-14 national tests and measuring peers' ability by prior achievements at age 11. Our identification is based on within-pupil regressions exploiting variation in achievements across three compulsory subjects tested at age 14 and age 11. Using this novel strategy, we find significant and sizable negative effects arising from bad peers at the bottom of the ability distribution but little evidence that average peer quality and good peers matter. However, these results are heterogeneous, with girls benefiting from academically bright peers and boys not.