Cross-disciplinary analyses using multi-attribute utility theory.

摘要

This dissertation consists of three projects that apply multi-attribute utility theory to contexts in which there are unique structural relationships between the attributes. In the first project, we develop an initial model for maximizing expected utility in health decisions, based on the principles of quality-adjusted life years (QALY). We then extend this model to allow for adaptation to health states. We do this by maintaining a reference level for each possible health condition being considered. A particular formulation of the utility function and reference levels allows us to compute the a priori reference-dependent expected utility for any given alternative.;In the second project, we develop preference conditions for decisions made using Geographic Information Systems (GIS). These conditions allow us to simplify the elicitation of preferences over spatially-defined outcomes, and are based on standard results in multi-attribute utility theory. We then apply these tools to several specific decisions, in the contexts of urban development, irrigation, nuclear accident planning, and fire coverage.;In the third project, we formulate and develop utility structures that can effectively incorporate altruistic preferences. Specifically, we examine the utility implications of multiple individuals having altruistic tendencies toward one another. We explore utility structures with two altruistic individuals, and then expand our results to a more general model. The more general model can be greatly simplified by dividing the altruistic individuals up into groups, wherein the altruistic interactions are expressed at the group level. Our results can be expanded beyond altruism; they provide insight for any situation in which each individual outcome includes externalities that affect the outcomes experienced by others.