The Predictivity of Access Tests for University Success

Abstract

In Italy, the transition from secondary school to university is a topic of debate. The transition from the first to the second year is particularly crucial, with dropout and degree course changes around 20%, with many attributed to a lack of information and guidance. Admission tests play an important role during this transition, as they should provide information for both the student and university. This chapter analyzes the predictivity of admission, with respect to the credits earned at the end of the first year. To achieve this goal, a “matching” algorithm was developed, merging the admission test database provided by CISIA and the National Student Archive of the Ministry of Education (MIUR). The CISIA database contains personal information, test scores, and the date and location of the test, while the MIUR dataset contains personal and career information for all students enrolled in any Italian university. Neither database contains a common primary key of the student. To overcome this limitation, a derived key was constructed by concatenating variables from both datasets. In this way, a mixed logit was applied to detect the students’ profiles with higher probability of earning more than 20 and 40 credits at the end of first year.