| This dissertation focuses on the spatio-temporal modeling of risk across different spatio-temporal scales and within different research contexts---namely health related exposures and environmental hazards. The first chapter introduces the motives for the spatio-temporal modeling of risk and outlines the methodological challenges that require attention.;Chapter 2 examined the accuracy of different interpolations of UV exposure and used the routine that produced the least error to deliver UV exposure estimates per square kilometer and by county across the continental USA. The procedure proved relatively simple and computationally efficient, and generated reproducible results for which the uncertainty is known.;Chapter 3 utilized these UV exposure estimates to generate lifetime cumulative UV exposure models, which facilitated, in a case-control setting, the objective assessment of UV exposure and melanoma risk. This research demonstrates that the method goes some way to reducing bias towards the null in effect estimates of UV exposure in melanoma risk.;Chapter 4 examined three alternative approaches for defining contextual environments and their characteristics for use in multilevel analysis. The results suggest that the choice of boundaries influences the neighborhood profile and the relationships between its components. These outcomes suggest, in turn, that the inferences about the effects of contextual environments on health outcomes may be affected by the way the contextual unit is defined.;Chapter 5 focused on the spatio-temporal modeling of social vulnerability in Los Angeles, 1940-2000, and addressed some of the key methodological issues related to assessment of social vulnerability. The results indicate that historically, the urban core experienced increases in social vulnerability, which coincides both spatially and temporally with the social changes that have transformed the central city. The research also demonstrates that different combinations of indicators may result in substantially different spatio-temporal patterns of social vulnerability, and highlights the uncertainties that remain when understanding risk assessments.;The outcome of the initiative to approach the spatial-temporal modeling of risk in a variety of research contexts and across different spatio-temporal scales is a demonstration that the spatio-temporal context matters and drives the handling of uncertainties, as well as the extent to which they can be controlled. |
