Skip to Main content Skip to Navigation

Prise en compte de la surdispersion par des modèles à mélange de Poisson

Abstract : This thesis proposes a pragmatic approach in the frame of environmental data which are overdispersed. This overdispersion, which could be explained by a mispecification of the model or by an incomplete data collection, leads to an important bias in the estimation of parameters. This research work proposes an approachbased on Strict Arcsine regression as an alternative as Binomial-Negative regression. The second point which is proposing concerns the presentation of an hierarchical model, relatively recent in epidemiology, and an potential extension to spatial correlations, which could extend information available in ecological studies by individual factors. Both of these aspects will be detailled theorically and by simulation studies. In conclusion, we'll try to precise the different caracteristics of cardiovascular mortality among elderly by demographic study. In a second part, we'll detaill risk factors of this cause of death, and specially the effect of calcium and magnesium contained in drinking water.
Document type :
Complete list of metadata

Cited literature [146 references]  Display  Hide  Download
Contributor : Sebastien Marque <>
Submitted on : Monday, August 1, 2005 - 5:22:25 PM
Last modification on : Monday, October 19, 2020 - 11:06:03 AM
Long-term archiving on: : Friday, April 2, 2010 - 10:13:31 PM


  • HAL Id : tel-00009885, version 1


Sebastien Marque. Prise en compte de la surdispersion par des modèles à mélange de Poisson. Sciences du Vivant [q-bio]. Université Victor Segalen - Bordeaux II, 2003. Français. ⟨tel-00009885⟩



Record views


Files downloads