ABSTRACT
Remedial courses may support under-prepared candidates for higher education, but their effectiveness is still questioned especially in European countries, where their introduction is comparably recent. This paper implements a doubly robust estimator to account for heterogeneity between remedial and nonremedial students and possible noncompliance with the assigned remediation. Data on five cohorts of undergraduates in industrial engineering from an Italian university show average worse performances of remedial students. However, remedial students who complete the remedial path catch up in two years with the dropout rate of average nonremedial students and with the credits earned by the weakest nonremedial students.
Acknowledgements
I thank the faculty and staff of the university that provided access to the data used in this study for valuable help.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Administered test are designed and evaluated by an inter-university consortium (Consorzio Interuniversitario Sistemi Integrati per l’Accesso, CISIA) that groups 48 Italian universities.
2 Prior to 2012 students’ career files did not record the earning of the OFA credits.
3 This study employed the IPWRA estimator implemented in the teffects program in Stata 15 (StataCorp Citation2017).
4 Additional estimates, not reported in the paper, show that OLS/logit regressions tend to overestimate credit losses and underestimate the probability of dropout for remedial students. Similar despite smaller biases result from calculating the effect of compliance to treatment according to the so-called per-protocol approach to noncompliance (Imbens and Rubin Citation2015), which simply discards all observations of remedial students who do not comply with the assignment of acquiring the extra OFA credits. In this case the ATET parameter directly assesses the treatment effect on compliers, who are the only treated individuals included in the estimates.