| The value of event history techniques for understanding social processes: modelling womens employment behaviour |
| JANE ELLIOTT (Centre for Longitudinal Studies, Institute of Education) |
| This paper aims to demonstrate some of the advantages of using event history data to underpin our understanding of social processes. In particular it focuses on the importance of estimating models that allow for the control of unobserved heterogeneity in order to be able to understand the temporal dependencies in the social process under investigation. Analyses are based on data from womens life histories collected as part of the fifth sweep of the National Child Development Study when women were 33 years old. |
| |
| Random effects and latent variables in event history analysis: measurement error, multivariate events and endogeneity |
| ANDREW PICKLES (University of Manchester) |
| The somewhat specialised nature of survival and event history analysis has meant that a number of methodological problems commonly addressed in other fields have received inadequate attention. These include errors of recall, multivariate events and endogeneity. We provide an overview of the role and potential of random effects and latent variables in this area, and illustrate a number of ways in which these may be operationalised. Examples are drawn from the Virginia Twin Study of Adolescent Behavioral Development and elsewhere. |
| |
| Modelling multiple event histories |
| ARNSTEIN AASSVE (ISER, University of Essex) |
| Most applications of event history analysis tend to focus on a single outcome, such as the timing of births, duration until employment, or individuals entering or exiting poverty. Event history models allow the researcher to investigate the determinants of these outcomes by controlling for a set of explanatory variables. However, in many settings it is more natural to consider event histories in a joint modelling framework. This is particularly the case when several single events tend to be influenced simultaneously by the same factors. For instance, the decisions a couple make in terms of their job careers are not made independently of their childbearing decisions, and vice versa. In such a case it might be more appropriate to model childbearing and labour market outcomes in a joint modelling framework. This presentation provides an outline of how such processes can be estimated jointly, an outline of benefits and disadvantages of pursuing such a modelling strategy, a brief outline of available software packages, and examples of applications based on the BHPS and other data sources. |
| |
| Meeting Contact: Fiona Steele [fiona.steele@bristol.ac.uk] |
| Organising Group(s): Social Statistics Section |
Unified Solutions.