Researchers

Fiona Steele, Harvey Goldstein,
Heather Joshi, Constantinos Kallis
formerly Institute of Education

1 June 2003 - 31 May 2005

Context

Event histories enable the study of interesting questions about social processes. However, their analysis is often complicated because different types of event may be correlated, e.g. the risk of divorce may be correlated with birth intervals; in addition, there may be repeated events, competing risks and multiple origin states.

Aims and Objectives

• To develop methodology for the analysis of complex event history data;
• To provide means for implementing methodology in existing software;
• To study interrelationships between transitions from cohabitation and marriage, and childbearing in Britain;
• To provide social scientists with practical training in advanced event history analysis.

Methodological aspects

The project will develop new methodological tools for event history analysis, specifically for the joint modelling of correlated histories where there are repeated events, competing risks and multiple states. An important objective of the research is to provide social scientists with the software and training with which to apply these methods in their own research.

Research Design

The study will extend existing methodology for the analysis of correlated event histories, where the durations between the events of interest may be mutually dependent. The new methodology will be implemented in multilevel software, e.g. MLwiN and aML.

The methodology will be applied in a study of the interrelationships between partnership (cohabitation or marriage) durations and childbearing in Britain, using event history data from the 1958 and 1970 British birth cohorts.

Outputs

• Workshop on multilevel event history analysis;
• Research papers, software and datasets will be made available on the project web site;
• Presentations at population studies and statistics conferences in Britain and the USA;
• Papers will be submitted to social science journals.