Event History Analysis
Welcome to Soc 213B!
This course is being offered Winter 2026 at UCLA. See the syllabus.
Who should take this course?
The course is a good fit for PhD students in sociology, statistics, political science, economics, and other social sciences.
Schedule of topics
Below is a tentative schedule of topics. This is my first time teaching the course, and the sequence of topics may change over the course of the quarter.
Part 1: Descriptive survival analysis.
- Week 1: Basics of survival analysis and maximum likelihood. Exponential model in math.
- Week 2: Review of Maximum Likelihood Estimation with the Bernoulli. Exponential model in software.
- Week 3: Beyond the Exponential: Weibull for hazards that vary with time. Generalizing ideas to proportional hazards and the Cox model. We will contrast with nonparametric survival curve estimation by Kaplan-Meier.
- Week 4: Unmodeled heterogeneity, competing risks, non-ignorable censoring, and other complications.
Part 2: Causal inference in event history settings.
- Week 5: Causal inference for survival outcomes
- Week 6: Event history treatments with non-survival outcomes (longitudinal inverse probability weighting, marginal structural models)
- Week 7: Event history treatments with survival outcomes
- Week 8: Causal inference in staggered adoption panels
Learning goals
Students will learn to
- define key components of survival analysis (e.g., censoring)
- state the assumptions required for a survival model
- translate from survival models to predicted quantities of interest
- apply causal inference methods where events are treatments that unfold over time
Course description
Introduction to regression-like analyses in which outcome is time to event. Topics include logit models for discrete-time event history models; piecewise exponential hazards models; proportional hazards; nonproportional hazards; parametric survival models. We will also cover events that unfold over time as causal treatment variables.