How gender and prior disadvantage predict performance in college. ESRI Working Paper 667 June 2020.

Delaney, Judith and Devereux, Paul (2020) How gender and prior disadvantage predict performance in college. ESRI Working Paper 667 June 2020. [Working Paper]

PDF - Published Version Download (930Kb)

Abstract

Much research has shown that having a better class of degree has a significant payoff in the labour market. Using administrative data from Ireland, we explore the performance in college of different types of students. We find that post-primary school achievement is an important predictor: Its relationship with college performance is concave for college completion, approximately linear for the probability of obtaining at least second class honours, upper-division, and convex for the probability of obtaining a first-class honours degree. We find that females do better in college than males, even after we account for their greater prior achievement, and this is true in both non-STEM and STEM fields. Disabled students, students from disadvantaged schools, and students who qualify for means-tested financial aid are less likely to complete and less likely to obtain first-class honours or a 2.1 degree. However, once we control for post-primary school achievement, these students actually perform better in college than others. We also find that, conditional on prior achievement, students from private exam-oriented “grind” schools and from Irish-medium schools are less likely to finish a degree and less likely to perform well in college, possibly because their school exam results are high relative to their abilities. Our results suggest that current college policies that lower entry requirements for disabled students and students from disadvantaged backgrounds may be justified on efficiency as well as equity grounds. They also suggest that college performance might be improved by increasing entry requirements for students who come from school types that convey advantages in the post-primary exams that determine college entry.

Actions (login required)