Exploring the predictive power of standardized assessments and non-standardized evaluations on Italian university freshpeople’s performance

The present chapter aims to compare the ability of standardized assess-ments and non-standardized evaluations at the secondary education level in explaining and predicting students’ performance in tertiaty education. In par-ticular, standardized assessments are identified through INVALSI scores, while non-standardized evaluations through high schools’ final marks. Thanks to data drawn from the Italian national registry of tertiary education (Anagrafe nazion-ale della formazione superiore) merged with INVALSI data for the cohort sub-sequently enrolled in the Italian tertiary education system in the academic year 2019/2020, we perform different mixed-effect logit models with random inter-cepts, clustering students by high schools’ regions, in order to control for the opposite geographical results which emerge in Italy between INVALSI scores and final exams’ marks. Moreover, predictive ability of these two different kinds of evaluations is tested through a Naïve Bayes algorithm. Our analysis shows that non-standardized evaluations are slightly preferable both in terms of goodness-of-fit and predictive power for individual assessment. Nonetheless, our estimation also highlights the invaluable role that standardized assessments play for evaluating the overall Italian education system.