Essays on inputs, admissions and returns to education

Abstract: Paper [I] analyses the associations between computer use in schools and at home and test scores by using TIMSS (Trends in International Mathematics and Science Study) data for more than 900,000 fourth-graders in 2011 and 2015. Pupils who used computers at school, especially those who used them frequently, scored lower than students who never used computers. There is also a negative association between frequent computer use at home and test scores, but moderate monthly and weekly use at home is positively associated with pupil performance. The results also suggest that the negative association of computer use at school is greater among low-performing pupils than high-performing pupils.Paper [II] estimated the marginal achievement difference between students from the two admission groups. Swedish universities select students based on two different criteria: upper secondary school grade point average (GPA) and scores on a scholastic aptitude test (SweSAT), and each forms a separate admission group. The analysis was based on data from 9,024 university entrants in the academic year 2012/2013. Marginally accepted students in the group based on school grades on average perform better than students accepted based on their SweSAT scores, suggesting that a small reallocation of study positions towards the grade admission group may increase the overall academic achievement of university students.Paper [III] focuses on gender differences in first-year university achievement. Nearest-neighbour matching was used to compare students with similar admission scores and allowed us to analyse achievement differences between male and female students. The results show that admission scores underpredict achievement for women relative to men in both admissions groups and more so for the SweSAT. Additional analysis indicates that part of the achievement differences is related to male- female composition in different fields of education.Paper [IV] studies the effect of university education on economic outcomes among individuals who initially attained low levels of education, and then participated in adult education. It uses Swedish longitudinal population register data from 1990–2015 to estimate the return to university education among those who participated in adult education in 1994 and enrolled at a university between 1996 and 1998. Difference-in-difference propensity score matching accounts for unobserved time-invariant individual characteristics and non-random selection of university education. The results show significant gains in terms of earnings and probability of employment for those who proceeded into university.