Date/Time: Monday 7 April 2014, 09:00 - 15:00
Tuesday 8 April 2014, 09:00 - 15:00
Location: Sala del Capitolo, Badia Fiesolana
This two-day workshop initially reviews briefly fundamental notions of event history analysis (EHA) and of multilevel modeling (MLM), to set up the stage of their merger into multilevel EHA (MEHA). The topic of what to take care of before beginning MEHA is then attended to. Issues pertaining to manifest approaches to MEHA, which do not use the concept of latent variable, are then focused on. These approaches assume also that explanatory variables are not measured with error, and include an important re-conceptualization of the MEHA setting within the framework of the popular logistic regression analysis. Their discussion is followed by such of MEHA utilizing latent variables. To this end, first a brief review of latent variable modeling (LVM) is provided, which sets the grounds for the following developments. MEHA conducted within the framework of LVM, and in particular EHA with fallible covariates, is then dealt with. This approach is subsequently extended to cover the case of two-level EHA within the LVM framework. Throughout the workshop multiple empirical examples are used and the increasingly popular software Stata is utilized.
Main Topics Covered:
- Resources and what this workshop is about.
- A brief review of needed fundamental concepts and relationships from event history analysis (EHA) and multilevel modeling (MLM).
- In the beginning: What to take care of before one gets started with multilevel EHA (MEHA).
- A manifest variable modeling approach to multilevel EHA (Day 1).
- A latent variable modeling approach to multilevel EHA (Day 2).
- Analysis of time-to-event (TTE) data from nationally representative samples.
- Conclusion.
Requirements for credits: submission of computer exercises
Contact: Alina Maria Vlad (EUI) - Send a mail
Organiser: Hans Peter Blossfeld (EUI)
Organized and financed by: the European Research Council within the framework of eduLIFE project
