Work

Essays on sequence analysis for life course trajectories

Year: 2010
Type: dissertation
Abstract: The thesis is articulated in three chapters in which I explore methodological aspects of sequence analysis for life course studies and I present some empirical analyses. In the first chapter, I study the reliability of two holistic methods used in life-course methodology. Using simulated data, I compare the goodness of classification of Latent Class Analysis and Sequence Analysis techniques. I first compare the consistency of the classification obtained via the two techniques using an actual dataset on the life course trajectories of young adults. Then, I adopt a simulation approach to measure the ability of these two methods to correctly classify groups of life course trajectories when specific forms of “random” variability are introduced within pre-specified classes in an artificial datasets. In order to do so, I introduce simulation operators that have a life course and/or obser- vational meaning. In the second chapter, I propose a method to study the heterogeneity in life course trajectories. Using a non parametric approach, I evaluate the association between Optimal Matching distances and a set of categorical variables. Using data from the National Longitudinal Study of Adolescent Health (Add-Health), I study the hetero- geneity of early family trajectories in young women. In particular, I investigate if the OM distances can be partially explained by family characteristics and geographical context experienced during adolescence. The statistical methodology is a generalization of the analysis of variance (ANOVA) to any metric measure. In the last chapter, I present an application of sequence analysis. Using family transitions from Wave I to Wave IV of Add-health, I investigate the association between life trajectories and health outcomes at Wave IV. In particular, I am interested in exploring how differences in timing, quantum and order of family formation transitions are connected to self-reported health, depres- sion and risky behaviors in young women. Using lagged-value regression models, I take into account selection and the effect of confounding variables.
Language: English
Subfield: Demography
Open Access status: green
OpenAlex ID: W2278498300
Cites: 109
Cited by: 2
Related to: 9