Maarten J. Bijlsma imparteix la conferència "The g-formula for causal mediation and counterfactual decomposition: examples from fertility and health research" en el marc dels Colloquium CED
Organitza: Centre d’Estudis Demogràfics
Lloc: Sala Àngels Torrents, CED
Hora: 11:00 - 12:30
Previ al Col·loqui, de 10:40 a 11:00h, esteu tots convidats a l’Espai del cafè.
Maarten J. Bijlsma és investigador al Population Health Laboratory, Max Planck Institute for Demographic Research (MPIDR), a Rostock (Alemanya).
The g-formula for causal mediation and counterfactual decomposition: examples from fertility and health research
Abstract.- The g-formula is a cutting edge method for modelling and understanding time-dependent relationships, such as the interdependencies between fertility, socio-economic determinants, and health over the life course. The method was originally developed in biostatistics, but I have started applying it to social science topics. In the first part of the talk, I show an application of the g-formula to study the effect of increased higher education attendance on women’s life course fertility in the United Kingdom. Our results show that socio-economic processes play an important role in determining fertility, not only directly but also indirectly via partnership status and employment status. This study was recently accepted in Journal of the Royal Statistical Society: Series A (Statistics in Society). In the second part of the talk I show work in progress, reframing the g-formula as a method to do counterfactual decomposition analysis. Our newly developed method provides an alternative to commonly used decompositions in demography such as Oaxaca-Blinder, step-wise, and Kitagawa decomposition. The new method has advantages and disadvantages compared to the classical approaches. A disadvantage is that high quality micro-level data is needed. However, advantages are that it adds a more concrete interpretation (an ‘interventional analogue’) to decomposition results, and can account for any outcome type and summary measure including medians, modes, life expectancy, and TFR. To demonstrate this, I ask the question “What is the contribution of smoking to life expectancy differences between men and women in South Korea?”.