Three Essays on Individual and Collective Choice Theory from a Revealed Preference Approach

Date of Award

May 2015

Degree Type


Degree Name

Doctor of Philosophy (PhD)




Jerry S. Kelly

Subject Categories

Social and Behavioral Sciences


This dissertation consists of three chapters.

The first chapter develops an individual choice model to incorporate the extremeness aversion phenomenon, which refers to people's tendency to avoid extreme options. I follow a revealed preference approach and instead of involving any exogenous structures (e.g. attributes) to describe options or identify extreme ones, I deduce both preferences and extremeness endogenously from observed choices. My first result employs extremeness aversion to explain choice situations with revealed context effects and is also applicable to choice situations where the standard revealed preference theory applies. Without referring to further structures, I provide a choice-theoretical justification for extremeness. The second result sharpens this characterization by imposing a more concrete structure, a linear order, to identify extreme options. In particular, I strengthen the previous choice-theoretical justification and derive a linear order defined on the universal set of options; we describe an individual as if she consistently exhibits extremeness aversion with respect to the linear order. A concrete situation addressed by the second result is the experimental design applied to test the extremeness aversion phenomenon: options are portrayed by two attributes. To formally incorporate extremeness aversion with options represented by two attributes, I propose a two-stage process. My third result characterizes the two-stage process by obtaining the attributes endogenously. Accordingly, I interpret it as extremeness aversion with revealed attributes.

Part of the study of the two-dimensional result of the first chapter is related to a more general problem. Although the first chapter embedded the question into the context of individual decision-making, the same question can be also rephrased in terms of collective decision-making, which is formally presented and fully answered in my second chapter. Intuitively, suppose that we can observe the relation aggregated from preferences of a group of two agents but we are ignorant about the individual agents' preferences. Under what kind of conditions the aggregated relation is consistent with the Pareto dominance relation? The Pareto dominance relation of a preference profile is (the asymmetric part of) a partial order. In the second chapter, I use two independent conditions to characterize when (the asymmetric part of) a partial order defined on a finite set of options can be represented as the Pareto dominance relation of a two-agent preference profile.

More generally, for any integer n, the problem of the existence of an n-agent preference profile that generates the given Pareto dominance relation is to investigate the dimension of the partial order. In the third chapter, I provide a general characterization for an n-dimensional partial order.


Surface provides description only. Full text is available to ProQuest subscribers. Ask your Librarian for assistance.

This document is currently not available here.