We design and implement a field experiment that provides students from less advantaged backgrounds with individualized feedback on academic performance during the transition from middle to high school. The intervention reduces the gap between expected and actual performance, as well as shrinks the variance of the individual belief distributions. Guided by a simple Bayesian model, we empirically document the interplay between variance reductions and mean changes of beliefs about students' own academic ability in shaping curricular choices. The shift in revealed preferences over high school tracks enabled by the intervention affects schooling trajectories, with better performing students being assigned into more academically oriented options.