Statistical Analysis of Event History Studies

  • 申立勇
  • Created: 2014-12-08
Statistical Analysis of Event History Studies

 

Course No.21506Z    

Period20     

Credits1    

Course CategoryAdvanced Course      

Aims & Requirements:
Event history studies usually refer to studies on recurrent events and are often conducted in many fields including demographical studies, medical investigations, reliability studies, sociological studies and tumorigenicity experiments. The purposes of such studies could be to investigate recurrent rates of the events, to assess the effects of time-independent or time-dependent covariates on the occurrences of the events, and/or to make inference about the gap times. A useful tool for the analysis of event history studies is counting process. This is because in general, the occurrences of recurrent events can be characterized by the process and thus one can make use of the well established theory for counting processes.
Based on the designs of the studies or the structure of observed information, event history studies usually produce two types of data, recurrent event data and panel count data. The former gives complete information about the underlying process, while the latter provides only incomplete data about the process. More specifically, for the former, study subjects are supposed to be under continuous observation and the panel count data imply that study subjects are observed only at discrete time points. In consequence, panel count data tell us only the numbers of the occurrences of recurrent events but not the occurrence times as in recurrent event data.
This short course will discuss both types of data and their analyses. We will start with the concepts and then focus on various models that are commonly used for event history studies and present available inference procedures for each model. Some real examples will be used for illustrations and also some directions for future research will be discussed. The attendees are assumed to have knowledge about basic statistical inference and likelihood principle. Some knowledge about counting processes also helps but not required.
Primary Coverage

Lecture 1. Introduction
Lecture 2. Analysis of recurrent event data I
Lecture 3. Analysis of recurrent event data II
Lecture 4. Analysis of panel count data I
Lecture 5. Analysis of panel count data II

 

                                                  AuthorJianguo Sun