Non-Cognitive Skills, Gender, and Educational Trajectories: Evidence from Italian Administrative Data

This dissertation investigates how early non-cognitive characteristics shape academic achievement, educational trajectories, and gender differences in educational outcomes. In particular, it examines whether students' attitudes toward learning and test anxiety, measured during primary school, are associated with subsequent academic achievement and educational choices within the Italian educational system. The empirical analyses exploit longitudinal administrative data from the Italian National Institute for the Evaluation of the Education System (INVALSI), which allow students to be followed across multiple educational stages by linking standardised assessments, questionnaire information, and upper-secondary school enrolment records. The Italian institutional setting provides a particularly relevant context because students are required to choose among academic, technical, and vocational tracks at the end of lower-secondary education, making educational sorting an important determinant of subsequent educational opportunities. The dissertation consists of three essays. The first examines the relationship between early domain-specific non-cognitive characteristics and test anxiety, with later academic achievement, focusing on the persistence of these associations throughout compulsory education and on gender heterogeneity. The second investigates whether early attitudes toward learning predict upper-secondary educational paths, conditional on prior achievement and other observed characteristics. The third essay documents the development of Italy's longitudinal educational administrative data infrastructure, presents the construction of indicators for a national dashboard of educational outcomes, and discusses the methodological challenges involved in linking administrative sources and using large-scale assessment data to measure non-cognitive skills. Taken together, the three essays contribute to the Economics of Education in three respects. First, they provide evidence on the role of early non-cognitive characteristics in shaping both academic achievement and educational sorting. Second, they offer new evidence on the extent to which these mechanisms differ by gender, contributing to the literature on gender disparities in education. Third, they illustrate the potential and limitations of Italian administrative data for longitudinal research, providing a framework for future empirical work based on linked educational records.