Objectives:
This course introduces basic statistical methods for the analysis of time-to-event data. First, we will cover classical methods for the analysis of life events, e. g. marriage, birth of a child, moving from town to city, becoming employed or unemployed, becoming sick. We will then introduce methods for the analysis of life events in the regression framework. Thereafter, we will show how time-varying explanatory variables can be included in the model. Finally, we will discuss a selection of advanced topics. The emphasis will be on continuous-time methods, although a brief overview of discrete-time methods will be provided on request. Illustrations are taken from the social science literature. Practical skills are developed through computer exercises (based on the software R and Stata) with longitudinal data. The course is mainly designed for PhD students in population and health studies, sociology, economics, criminology, political science, social/applied statistics and human geography who are familiar with basic regression analysis. The course will run over four days; each course day will consist of a theory session (2h) and a computer lab session (2h).
Learning Outcomes:
An understanding of basic concepts of survival and event history analysis. A competence in applying descriptive methods for duration data. A competence in applying regression techniques for event history data. The ability to use statistical software R and Stata for time-to-event data analysis.
Syllabus:
Day 1
- Survival and event history analysis: basic concepts and functions
- Descriptive methods: life table method and product-limit estimation
Day 2
- Regression for event history data: non-parametric and parametric models; exponential and piecewise constant exponential model
Day 3
- Regression models with time-varying covariates
- Modelling two or more time dimensions or multiple ‘clocks’
Day 4
- Advanced topics: competing risks, multiple episodes/events, nested structures, unobserved heterogeneity, endogeneity
- Examples of applications of survival and event history analysis