Residential Context and Childbearing: Application of Multilevel Analysis to Study Individual and Contextual Determinants of Fertility

大多数工业化国家的生育率低于更替水平，近年来又进一步下降。然而，它们在不同的国家因居住环境而有很大差异。农村地区和小城镇的生育率高，大城市的生育率低。生育率空间差异的原因尚不清楚。一些研究认为，不同地区的生育水平不同，因为不同的人生活在不同的定居点。其他人则强调与直接生活环境有关的因素的重要性。本文还讨论了选择性迁移的作用。这个博士项目将通过居住环境来调查英国的生育模式，英国是一个具有显著空间生育率变化的国家。目标是：首先，确定生育模式的空间变化在多大程度上归因于组成特征和选择性迁移，以及环境因素在其中起什么作用。第二，研究环境因素对个体生育行为的影响。第三，建立空间多层次多过程危害模型，以衡量居住环境对个体生育行为的影响。了解居住环境如何影响生育率对于提高我们对工业化国家低生育率原因的理解至关重要。该项目将对来自英国国家统计局纵向研究、苏格兰纵向研究和英国家庭小组调查的数据开发和应用空间多层次多过程危害模型，以确定环境因素如何影响个人的生育行为。将所提出的方法的性能与使用最大似然和贝叶斯方法的传统方法进行比较。该项目将为英国的空间生育率变化提供丰富可靠的信息，并将提高我们对某些地区高生育率和其他地区低生育率的原因的理解。所开发的方法可用于研究个人行为其他领域（如健康、就业、住宅搬迁）的背景决定因素。

Fertility levels are below replacement level in most industrialised countries and have further declined in recent years. However, they vary significantly within countries by residential context. Fertility is high in rural areas and small towns and low in large cities. The reasons for spatial variation in fertility remain unclear. Some studies argue that fertility levels vary between places because different people live in different settlements. Others emphasise the importance of factors related to immediate living environment. The role of selective migrations has also been discussed.This PhD project will investigate childbearing patterns by residential context in Britain, a country with a significant spatial fertility variation. The objectives are: First, to determine the extent to which spatial variation in childbearing patterns is attributed to compositional characteristics and selective migrations, and what role contextual factors play. Second, to investigate how contextual factors influence individuals' childbearing behaviour. Third, to develop a spatial multilevel multiprocess hazard model to properly measure the effect of living environment on individuals' childbearing behaviour. An understanding of how residential context influences fertility is essential to improve our understanding of the causes of low fertility in industrialised countries.The project will develop and apply a spatial multilevel multiprocess hazard model to data from the ONS Longitudinal Study, the Scottish Longitudinal Study and the British Household Panel Survey to determine how contextual factors shape individuals' childbearing behaviour. The performance of the proposed approach will be compared to that of conventional methods using maximum likelihood and Bayesian methods. The project will provide rich and reliable information on spatial fertility variation in Britain and will improve our understanding of the causes of high fertility in some areas and low fertility in others. The developed method could be applied to study contextual determinants of other domains of individuals' behaviour (e.g. health, employment, residential relocations).