University drop-out is one of the most important problems occurring in degree courses. The aim of this study is to analyze the first year university drop-out at the University of Pisa (Italy). In particular, we are focused on identifying the covariates affecting the response variable (if the student dropped out or not), and on detecting unobserved subgroups of students, if they exist, having different probabilities of dropping out. To perform this analysis we propose the use of a finite mixture logit model that allow us to consider a methodological framework where the population is made up by an unknown but finite number of subpopulations (latent classes). A dataset formed by administrative data of the University of Pisa, collected at enrolment time for the academic year 2009-2010, is used for this purpose. The analysis is limited to students of the first cycle degree courses. The characteristics detected of subgroups and the influential covariates, should represent useful information for the implementation of academic policy changes that could affect the drop-out rate.
L’abbandono degli studi rappresenta uno dei problemi più delicati che si verificano durante il primo anno di università. Lo scopo di questa ricerca è quello di capire se esistono sottogruppi non osservabili di studenti caratterizzati da diverse probabilità di abbandono. L’analisi è stata effettuata utilizzando i dati delle coorti di immatricolati nei corsi di laurea triennale e ciclo unico dell’Università di Pisa negli anni 2009 e 2010. I risultati ottenuti potrebbero essere utili per programmare azioni di intervento finalizzate alla riduzione del tasso di abbandono.
A finite mixture model approach on the first year university drop-out probability
Bini M;
2014-01-01
Abstract
University drop-out is one of the most important problems occurring in degree courses. The aim of this study is to analyze the first year university drop-out at the University of Pisa (Italy). In particular, we are focused on identifying the covariates affecting the response variable (if the student dropped out or not), and on detecting unobserved subgroups of students, if they exist, having different probabilities of dropping out. To perform this analysis we propose the use of a finite mixture logit model that allow us to consider a methodological framework where the population is made up by an unknown but finite number of subpopulations (latent classes). A dataset formed by administrative data of the University of Pisa, collected at enrolment time for the academic year 2009-2010, is used for this purpose. The analysis is limited to students of the first cycle degree courses. The characteristics detected of subgroups and the influential covariates, should represent useful information for the implementation of academic policy changes that could affect the drop-out rate.File | Dimensione | Formato | |
---|---|---|---|
SIS_2014_BM_revised version.pdf
non disponibili
Dimensione
357.41 kB
Formato
Adobe PDF
|
357.41 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.