Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.798006
Title: Causal inference methods and simulation approaches in observational health research within a geographical framework
Author: Berrie, Lauren
ISNI:       0000 0004 8506 0837
Awarding Body: University of Leeds
Current Institution: University of Leeds
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Statistical methods are often used habitually, perhaps without sufficient reflection on their robustness in a range of novel circumstances. Increasingly, there is a desire to unravel the complexities of humans interacting with their environments, to improve our understanding and explanation of what influences population health in the wider context of our living environment. A framework is provided for using simulation and causal inference methods to evaluate analytical approaches in health geography, to introduce the reader to some of the considerations around complexity of context and data generation that may need to be reflected upon carefully when applying such methods in their own work. These methods have the potential to aid researchers in their explanation of what factors are important for population health and well-being in the context of our geographical environment while avoiding potential pitfalls in their work and allowing for greater critical evaluation of the methods employed by themselves and others. This thesis considers the utility of simulation to investigate applied problems related to mathematical coupling and specific considerations that need to be made in relation to research on the relationship between limiting long-term illness and deprivation and the challenges encountered while investigating the relationship between population mixing and childhood leukaemia -with all such considerations examined through the lens of cause and effect. The datasets chosen are representative of many others in health geography and span the full range of outcome prevalence rates likely encountered. Methods in causal inference and simulation are demonstrated to be powerful tools in understanding potential bias in research analyses. With careful planning, forethought and reflection on the data generating processes of the context of interest, causal inference and simulation methodologies are accessible to all researchers to improve their understanding of the methods they employ to address the research questions they pose.
Supervisor: Gilthorpe, Mark S. ; Norman, Paul D. ; Baxter, Paul D. Sponsor: Medical Research Council
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.798006  DOI: Not available
Share: