If the developmental education completers are drawn from the most motivated and academically prepared among the developmental students, then it may not be surprising that they do as well as or even better than the college-ready students. Thus developmental sequences may strengthen student academic skills, but they also act to screen out many students. Students who enter directly into college-level courses do not get the benefit of the remedial instruction, but neither are they subject to the screening. Comparisons between outcomes of developmental education completers and students who enter directly into college-level courses cannot differentiate between the academic benefits (what we want to measure) and the effects of screening (which tends to exaggerate the measured positive academic effects). In contrast, the regression discontinuity analyses do not try to separate the positive academic from the negative screening effects. By starting the comparison at the point when students are assessed, they measure the net effect of these two factors. It is true that these analyses do not reveal the effect on academic skills of students who complete their sequences. But the methods used in the Attewell et al. (2006) or Bahr (2010) studies are not able to answer this question either. In any case, we think that the RD studies answer the most relevant policy question. For the students included in the analysis, they measure the effect of the policy that is available to the colleges: offering remedial services. Colleges cannot refer a student to developmental education completion in the way that they can refer students to developmental education enrollment.