Jack Adeney

Expected Utility requires that individuals never strictly prefer a convex mixture of two lotteries over either of the two. However, recent experimental literature has found that these ``mixture effects'' are pervasive. Decision theoretic models used to rationalize such effects can broadly be split into three categories: 1) decision makers have a strict preference for mixtures, 2) mixing is the realization of strict preferences with noise, and 3) mixing is a representation of uncertainty over preferences. I develop a two-part experiment to identify the dominant mechanism underlying mixture effects. I find that less than 15% of observations are consistent with convex preferences, whereas approximately 75% are consistent with stochastic choice. These findings have important implications for policy development and behavioral welfare analysis when interpreting stochastic choice data.

This manuscript identifies a largely unrecognized connection between, and distinction in results for, Ellsberg's three-color paradox and Allais' common-consequence paradox. Both phenomena derive from reducing a common state payment to zero, but the reactions to this reduction are directionally opposed across the two problems. Using a common experimental framework, we develop intermediate tasks to identify the source of this difference. Key to the difference is the reversal of Allais' common consequence paradox when problem parameters are made more similar to those for Ellsberg.

Measures of rationality for stochastic choice datasets often lack economic interpretability. This manuscript develops a new measure of stochastic rationality by decomposing data into convex combinations of rational and irrational deterministic choice functions. In solving a constrained minimization problem, this method returns a distribution over choice functions such that irrational choice functions can be partitioned by a deterministic consistency measure, and rational choice functions can be partitioned according to nested models of decision-making. Understanding the rational decomposition of choice datasets is fundamentally important for assisting with policy design and social welfare planning, while partitioning the rational choice functions allows one to identify a latent shadow cost of strengthening the restrictiveness of the decision-making model. We test the methodology on the ``GARP for Kids'' dataset from Harbaugh et al. (2001).

Bargaining delays are costly, common, and usually attributed to hidden information: parties wait, screen, and signal because they do not know enough about the other side. But negotiations often drag on even when uncertainty about fundamentals is limited. We test whether delay arises because parties manage expectations. In our experiment, buyers bargain with privately informed sellers in an environment where standard theory predicts immediate agreement. We compare direct bargaining, where offers are observed, to mediated bargaining, where failed offers are hidden and trade occurs only when offers and minimum acceptable prices are compatible. Mediation creates a tradeoff: it prevents concessions from raising future expectations, but it may introduce coordination failure through incompatible hidden submissions. The results show that visible concessions prolong bargaining. When offers are observed, buyers make lower initial offers than theory predicts, bargaining is delayed, and efficiency falls. Belief data support the mechanism: sellers who observe higher initial offers expect higher future offers, and buyers anticipate this response. Mediation keeps failed offers private and protects concessions from becoming part of the bargaining history: buyers open higher, agreements occur sooner, and efficiency rises.

Economists have developed the revealed preference approach to learn about people’s preferences from observing their choices. This study compares two traditional elicitation methods for collecting choice data: 1) asking people to choose their preferred outcome, and 2) asking people to eliminate outcomes that are not preferred. Via an incentivised online experiment, this study replicates the findings that chosen sets tend to be larger in the elimination frame compared to the choice frame. However, this study is the first to recognise that choices within the elimination frame are less consistent with welfare maximising outcomes than those provided in the choice frame. A detailed analysis of selection order suggests that decision makers eliminate sequentially from worst to best, and that welfare loss is driven by early stopping. Motivated by the natural assumption that attention between options is shifted as a result of switching frames, these differences are interpreted through the lens of a consideration set approach. Findings highlight how subtle changes in elicitation methods can have significant impacts on decision maker welfare.