This paper estimates the impact of peer achievement and variance on math achievement growth. It exploits exogenous variation in peer characteristics generated at the transition to upper-secondary school in a sample of Berlin fifth graders. Parents and schools are barely able to condition their decisions on peer characteristics since classes are newly built up from a large pool of elementary school pupils. I find positive peer effects on achievement growth and no effects for peer variance. Lower-achieving pupils benefit more from abler peers. Results from simulations suggest that pupils are slightly better off in comprehensive than in ability-tracked school systems.
peer effects in secondary school comparison between ability-tracked and comprehensive school natural experiment