Event History Analysis
- Sukriti ISSAR, Assistant Professor of Sociology at Sciences Po
Free of charge - open to all masters and doctoral students with working knowledge of STATA, R or OLS
What is Event History Analysis ?
A method widely used across social science disciplines. A class of statistical techniques that analyses the probability that an event occurs, how that probability changes over time, and how it is mediated by other factors. Essential for thinking about how social, political and economic processes evolve over time : the survival of firms or persons, the duration of unemployment, the lengths of wars or legal disputes, the failure of cabinets over time, the longevity of alliances or dictatorships.
A method widely used across social science disciplines. A class of statistical techniques that analyses the probability that an event occurs, how that probability changes over time, and how it is mediated by other factors.
Event history, or duration analysis, is a method that is widely used across social science disciplines – particularly in sociology and political science, and also of interest in history, social policy, demography and economics. Beyond the practicalities of its statistical use, the method is essential for thinking about how social, political and economic processes evolve over time. I have used this method in my research in sociology, social policy, and political science, and the course will draw on interdisciplinary examples. Many important social science questions focus on time, duration, and the probability of event occurrence - the lengths of wars or legal disputes, the failure of cabinets over time, the longevity of alliances or dictatorships, the survival of firms or persons. Event history or survival analysis is a class of statistical techniques that analyzes the probability that an event occurs, how that probability changes over time, and how it is mediated by other factors. Event history analysis is a very coherent method, with a clear progression in how students acquire understanding. We start with the research logic of event history, data structure, nonparametric techniques, and introduction to multivariate survival models. The course will end with a discussion of special topics [competing risks, unobserved heterogeneity]. It is possible for students to master this method at an advanced level within an intensive one-week course. This course provides students with the key concepts and competence to pursue further statistical expertise of this method on their own.
A working knowledge of STATA / R and OLS regression is a prerequisite. Event history analysis works with a particular type of data structure – therefore some competence in data management and understanding of data structure is a plus. The course will be taught using STATA (even earlier versions like STATA 13 are good enough), and students are free to use R if they wish.
The course aims to introduce students to the key concepts of event history, data structure, an overview of parametric, semi-parametric and non-parametric approaches, and a focus on application of these concepts on practice datasets and students’ own data. Each class will consist of a lecture component where we will discuss key concepts, and an applied component where we will analyze data, and implement the concepts from the lecture. There will also be a time for lab-work in the afternoon, where students can work together on problem sets, with aid from the instructor.
Assessment and credits
Problem sets will be handed out for the first three days of class; practice data and code will be provided and the aim of the problem sets is to allow students to interpret data output. If students have their own datasets, they are welcome to use those (advance discussion with the instructor is recommended if you intend to use your own data). For the final assessment, students will put together the problem sets into a short essay (4-5 pages).
Masters students who will successfully complete the course will earn 4 credits.
The course structure below is a guideline for reading before class. The readings are kept to the minimum, with a focus on the applied component – students should read the entries from the STATA manual listed under each lecture. The manual is a very useful and practical resource, combining theory and practice in a readable and useful way. The student can choose what interests them from the rest of the readings; recommended readings are useful especially for special topics (unobserved heterogeneity, time dependence), while the substantive readings provide empirical exemplars.·
Lecture 1: We will cover basic concepts of event history, research logic, when to use event history, and the concept of temporal dependence. For the applied component, we will explore the typical data structure of an event history dataset [read the entry on stset in the STATA Manual]
Lecture 2: We will cover non-parametric methods of analysis including Kaplan-Meier curves, life tables, and hazard and survival curves. For the applied component, we will implement these non-parametric methods [read especially the entry on ltable, sts set of commands in STATA Manual]
Lecture 3: We will cover semi-parametric and parametric multivariate models for analyzing event history data. We will cover the logic behind the Cox model, the exponential model and the Weibull model. For the applied component, we will implement these methods with a focus on the Weibull method [read the entry on streg and stcox in the STATA Manual]
Lecture 4: We will continue with semi-parametric and parametric multivariate models and briefly cover special topics of competing risks and unobserved heterogeneity. For the applied component we will continue with our focus on the Weibull method [read the entry on stcrreg and the sections on frailty in streg in the STATA Manual]
Lecture 5: Optional attendance to work on homework and ask questions
The course prerequisite is some experience with any statistical software (STATA, R), and that you have studied basic regression in some previous class. Since most of the students enrolled in the course will not have access to STATA on their personal computer, the instructor will make available various STATA outputs for you to analyze in homework sets. There will be a short homework assignment everyday - it will center on interpretation of data output based on the concepts discussed in class. At the end of class, you will put the homework assignments together into a 5 page analysis. If you have any questions about the course or pre-requisites, please contact the instructor email@example.com
STATA. (n.d.). STATA Survival Analysis and Epidemiological Tables: Reference Manual
Release 12. [available online]
Cleves, Mario, William W. Gould, and Roberto Gutierrez. 2008. An Introduction to
Survival Analysis Using Stata, Revised Edition. Stata Press.
Blossfeld, H., & Hamerle, A. 1989. Unobserved heterogeneity in hazard rate models: a test
and an illustration from a study of career mobility. Quality and Quantity, 23, 129–141.
Box-Steffensmeier, Janet M. and Bradford Jones. 2004. Event History Modeling: A Guide
for Social Scientists. Cambridge, UK: Cambridge University Press.
Box-Steffensmeier, Janet M., & Christopher J. W. Zorn. 2001. Duration Models and
Proportional Hazards in Political Science, American Journal of Political Science, Vol. 45 (4): 972-988.
Box-Steffensmeier, Janet M., Dan Reiter, & Christopher Zorn. 2003. Nonproportional
Hazards and Event History Analysis in International Relations. Journal of Conflict
Resolution, Vol. 47 (1): 33-53.
Christopher J. W. Zorn. 2000. Modeling Duration Dependence. Political Analysis, 8:3; 367-
Gordon, S. 2002. Stochastic Dependence in Competing Risks. American Journal of Political Science, 46(1), 200–217
Alt, J. E., & King, G. 1994. Transfer of Governmental Power: The Meaning of Time
Dependence. Comparative Political Studies, 27(2), 190–210.
Bennett, D. Scott, & Allan C. Stam III. 1992. The Duration of Interstate Wars,1816-1985.
APSR, Vol. 90, 2: 239-257.
Besedes, Tibor and Thomas J. Prusa. 2006. Ins, outs, and the duration of trade. Canadian
Journal of Economics, Vol. 39, No. 1, 266-295.
Gasiorowski, Mark J. 1995. Economic Crisis and Political Regime Change: An Event
History Analysis, American Political Science Review, Vol. 89, 4. 882-897.
Golub, Jonathan. 2008. The Study of Decision-Making Speed in the European Union
Methods, Data and Theory. European Union Politics, Vol. 9 (1): 167–179.
Kiefer, Nicholas M. 1988. Economic Duration Data and Hazard Functions. Journal of
Economic Literature 26: 646-679.
Kroft, K., Lange, F., & Notowidigdo, M. J. 2013. Duration Dependence and Labor Market
Conditions: Evidence from a Field Experiment. The Quarterly Journal of Economics, 1123–
Marco, Alan C. 2007. The Dynamics of Patent Citations. Economics Letters, 94: 290-296.
Quackenbush, Stephen L. and Paul D. Senese. 2003. Sowing the Seeds of Conflict: The
Effect of Dispute Settlements on Durations of Peace. The Journal of Politics, Vol. 65, No. 3, August 2003, Pp. 696–717.
Schleiter, P., & Morgan-jones, E. 2009. Constitutional Power and Competing Risks:
Monarchs, Presidents, Prime Ministers, and the Termination of East and West European
Cabinets. American Political Science Review, 103(3), 496–512.
South, Scott J. and Glenna Spitze. 1986. Determinants of Divorce over the Marital Life
Course. American Sociological Review 51: 583-590.
Teachman, Jay D., and Mark D. Hayward. 1993. Interpreting Hazard Rate Models.
Sociological Methods & Research 21: 340-371.
Teacher: Sukriti ISSAR
Sukriti ISSAR studies how urban policy transforms cities, with a particular focus on low-income housing in Mumbai over the last hundred years. Her research interests are in urban sociology, cities in the developing world, urban governance, comparative policy, and research methods. Her published work can be found in World Politics, Social Service Review, and the Journal of Historical Sociology.
Before Sciences Po, she completed a PhD at Brown University and was a postdoctoral fellow at the University of Oxford.
CV and Publications