Life history analysis
Stochastic demography is an emerging field in which stochastic models are developed based on probability theory, which integrates the effects of risk and uncertainty into demographic models (life table and projection).
Using life history data: PhD research
In his dissertation, Liang (2000) takes a well-known model in demography, the Coale-McNeil model of age at first marriage, and shows that the model belongs to a particular type of probability distribution (log-gamma distribution). An impressive feature of the model is that it expresses the age at event in terms of the age at a reference event and several delays. In the case of marriage, the reference event is entry into marriage, with the delays being related to different levels of commitment with respect to the partnership. In modern societies, the delays become longer and more diverse, resulting in the postponement of marriage and decisions not to marry or cohabit at all. In traditional societies, such as North India, some postponement of marriage can be observed, and in eastern Europe, marriage is postponed when restrictions, such as the housing market, are removed (Mills 2000). Using life history data from India and China, Liang shows that the delays can be revealed through observations on ages at marriage.
Monitoring change is mostly data driven. In periods of rapid change and major innovations, empirical observations are inadequate to predict change. What is needed to increase the predictive performance of demographic models and thus reduce uncertainty is insight into causal factors and processes that determine the level, sequence and timing of demographic events. Two sets of theories are considered significant in the monitoring of demographic changes - biographical (life course) theories and generation theories (Willekens, 1999a). These theories have been studied extensively by De Bruijn (1999) in his dissertation. De Bruijn develops a general theory of demographic change emphasizing process - onset, continuation and termination - and multilevel context. The perspective has been referred to as the process-context approach. Complex change processes are viewed as outcomes of interactions between various processes at different levels of analysis. The specification of the processes and the nature of the interaction remain major research challenges. That research is, however, considered a first step towards a theory-driven monitoring of demographic change. Concepts such as event (life event), risk and uncertainty, exposure, interaction, and learning are central to these biographical and generation theories and consequently to new types of projection models (Willekens, 1999b). An important step is to test biographical and generation theories empirically. These issues are addressed in a number of PhD projects, some of them concluded in 2000 (e.g. Mills, 2000). Mills found that biographical and generation theories are related to the structuration theory of Giddens, which gives additional insights into the emergence of individual biographies and cohort biographies. She also found that similar mechanisms are at work in very different socio-cultural contexts and that the demographic systems that evolve are specific, path-dependent responses to enabling and constraining factors.
Two frequently used data sets are the 1994 Micro Census of Russia and the 1992-93 National Family and Health Survey (NFHS) of India. Other life history data used by several persons in research and teaching include the 1993 and 1998 Family Formation Survey of the Netherlands, the Bangladesh Demographic and Health Surveys, the Indonesian Demographic and Health Surveys, the 1998 Dharwad Survey on Child Spacing, and the Framingham Study. In addition, PhD candidates and other researchers use a variety of micro-datasets. A first project, lead by Sergei Scherbov, carried out in cooperation with Goskomstat of Russia and funded by NWO, is the analysis of the 1994 Micro Census of Russia, a 5 percent sample. The Micro Census recorded the marital and fertility histories of a very large number of women and can be used to assess demographic consequences of historical events in Russia. The Micro Census data for one region, the Pskov oblast, were used by Melinda Mills in an international comparative study of the transformation of partnerships (Mills, 2000). In addition, some MSc students used Micro Census data. The NFHS data were used by Sabu Padmadas and others, mainly in studies of reproductive health.
The Framingham Study is a prospective panel study of about 5,000 persons in Framingham, Massachusetts, USA, and is one of the most used data sets in epidemiology. The data set is used to pave the way for a new field of research - multistate models for the monitoring and the forecasting of chronic disease processes. Current research involves the development of multistate life table models of cardiovascular disease processes and multistate hazard models for the associated risk factors. The research is carried out in cooperation with the Institute of Public Health, Faculty of Medicine and Health Sciences, Erasmus University of Rotterdam (Professors J.P. Mackenbach and P.J. van der Maas), and is funded by the NWO (Medical Research Division). The models should contribute to an improved monitoring of chronic diseases in a population and should lead to new epidemiological models for mortality forecasting. The models will be applied to new data being collected in the city of Rotterdam. In writing his dissertation, PhD candidate A.A. Mamun collaborated with researchers at Erasmus University. Similar research with a focus on lung diseases has been initiated. The research is in cooperation with the Institute of Medical Technology Assessment, Faculty of Medicine and Health Sciences, Erasmus University of Rotterdam (Professor F. Rutten and Dr L. Niessen), and is also being funded by the NWO (WOTRO). The models will be applied to data currently being collected in Nepal as part of a WHO programme.
|Last modified:||15 November 2012 2.26 p.m.|