Researchers often identify degree effects by including degree attainment (D) and years of schooling (S) in a wage model, yet the source of independent variation in these measures is not well understood. We argue that S is negatively correlated with ability among degree-holders because the most able graduate the fastest, while a positive correlation exists among dropouts because the most able benefit from increased schooling. Using data from the NLSY79, we find support for this explanation, and we reject the notion that the independent variation in S and D reflects reporting error.