This study analyzes perceived precariousness using data from the VulnYouth project, integrating both structured survey data and an open-ended response to the question: What does precariousness mean to you? As a continuation of a previous application, we integrate the output of Structural Topic Modeling (STM), which classifies text responses without manual coding, incorporating the estimated topic proportion matrix as covariates in a Bayesian ordered logistic regression model. To manage the large number of covariates, we apply a horseshoe prior for feature selection. We have introduced three models: two incorporating topics as covariates and one omitting them. We confront the three models using the widely applicable information criterion, WAIC. The model with the best predictive performance (lowest WAIC) confirms the value of including topics as covariates, reinforcing the effectiveness of natural language processing (NLP) in social science research. Bayesian inference has been obtained through the software Stan. Health, work-life balance, having a second job, financial stability, education, and economic independence emerged as key factors influencing perceived precariousness. Additionally, among the themes identified from the open-ended responses, those who mentioned the "inability to save money" and "low wages" reported a stronger sense of precariousness. In contrast, the "inability to afford basic needs" appeared to have a positive influence on perceived precariousness.
Questo studio analizza la precarietà percepita utilizzando i dati del progetto VulnYouth, integrando sia dati strutturati da un sondaggio che la risposta aperta - cos’è la precarietà per te? -. Un’innovazione chiave è l’uso dello Structural Topic Modeling (STM) per classificare le risposte aperte senza codifica manuale, incorporando la stima della matrice delle proporzioni dei topic tra le covariate in un modello bayesiano di regressione logistica ordinata. Per gestire l’elevato numero di covariate, abbiamo applicato una prior horseshoe per la selezione delle variabili. Abbiamo introdotto tre modelli, due incorporando i topic all’interno delle covariate e uno senza. Il modello con il miglior WAIC conferma l’utilità dell’inserimento dei topics all’interno delle covariate, affermando l’efficacia dell’integrazione del linguaggio naturale (NLP) nella ricerca nell’ambito delle scienze sociali. La salute, l’equilibrio tra lavoro e vita privata, avere un secondo lavoro, la stabilità finanziaria, l’istruzione e l’indipendenza economica sono emersi come fattori fondamentali che influenzano la precarietà percepita. Inoltre, tra i temi emersi dallo studio delle domande aperte, coloro che citano l’impossibilità di risparmiare e l’avere salari bassi risultano avere un maggiore senso di precarietà. Invece, l’impossibilità di permettersi beni essenziali è un tema che sembra influenzare la precarietà in positivo.
Precariousness and mental health among spanish youth
Di LIBERATORE, ELENA
2023/2024
Abstract
This study analyzes perceived precariousness using data from the VulnYouth project, integrating both structured survey data and an open-ended response to the question: What does precariousness mean to you? As a continuation of a previous application, we integrate the output of Structural Topic Modeling (STM), which classifies text responses without manual coding, incorporating the estimated topic proportion matrix as covariates in a Bayesian ordered logistic regression model. To manage the large number of covariates, we apply a horseshoe prior for feature selection. We have introduced three models: two incorporating topics as covariates and one omitting them. We confront the three models using the widely applicable information criterion, WAIC. The model with the best predictive performance (lowest WAIC) confirms the value of including topics as covariates, reinforcing the effectiveness of natural language processing (NLP) in social science research. Bayesian inference has been obtained through the software Stan. Health, work-life balance, having a second job, financial stability, education, and economic independence emerged as key factors influencing perceived precariousness. Additionally, among the themes identified from the open-ended responses, those who mentioned the "inability to save money" and "low wages" reported a stronger sense of precariousness. In contrast, the "inability to afford basic needs" appeared to have a positive influence on perceived precariousness.File | Dimensione | Formato | |
---|---|---|---|
Elena_Di_Liberatore___Thesis.pdf
accessibile in internet per tutti
Descrizione: tesi
Dimensione
2.24 MB
Formato
Adobe PDF
|
2.24 MB | Adobe PDF | Visualizza/Apri |
Elena_Di_Liberatore___Executive_Summary.pdf
accessibile in internet per tutti
Descrizione: executive summary
Dimensione
396.53 kB
Formato
Adobe PDF
|
396.53 kB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/235490