Growing Up in Taiwan: Counterfactual Models of Part-Time Work, Romantic Relationships, and Crime among Taiwanese Youth

Using marginal structural models for counterfactual inference and Bayesian Item Response Theory (IRT) models, this dissertation investigates the effect of two turning points in the life course, work and marriage, on deviance for Taiwanese youth between the ages of 15 and 26. The research draws on the Taiwan Youth Project (TYP) survey, a longitudinal study of youth in Taiwan that presents challenges for the analysis of crime because the survey items that measure deviance change with each wave. The dissertation overcomes this problem through a novel application of Bayesian IRT models to the measurement of crime. For the first time in criminology research, vertical scaling methods from the field of educational assessment are used to place latent IRT measures of deviance from different waves onto a common scale, allowing for the analysis of within-individual change in delinquency. A Bayesian approach is both necessary given the large numbers of respondents each wave reporting zero deviant acts (a situation that leads to violations of frequentist IRT model assumptions in the case of the TYP data), and also makes possible a new approach for minimizing measurement error during vertical scaling using Dijkstra’s algorithm. Substantively, work among Taiwanese youth of high school and college age causes an increase in deviance, while variability in work intensity and extensity does not explain variability in deviance. The findings also reinforce the importance of social control in Chinese culture: youth with low parental monitoring experience a strong, positive, causal effect of work on deviance, while for those with high parental monitoring, getting a part-time job does not cause any appreciable change in delinquency. Concerning romantic relationships, the intent to marry (either in general or one’s dating partner) has a spurious association with deviance. Limitations of the available data restrict analyses of the impact of marriage on desistence to an analysis of smoking, and while Taiwanese who are married tend to not smoke, getting married does not actually cause people to quit. Overall, the findings suggest that for Taiwanese youth, school and family bonds affect both the timing of work and marriage and their subsequent effect on deviance.