We semiparametrically estimate average causal effects of different lengths of exposure to academic and vocational instruction in the Job Corps (JC) under the assumption that selection into different lengths is based on a rich set of observed covariates and time-invariant factors. We find that...