Date of Thesis

Spring 2026

Description

Legal standards of proof play a central role in structuring decision-making under uncertainty, yet they are typically communicated using qualitative language rather than precise probabilities. This raises an important question: how do individuals interpret and apply these standards when making decisions based on probabilistic evidence? This study examines whether individuals apply legal standards of proof in a manner consistent with their commonly associated probabilistic thresholds, and whether quantification and incentives improve alignment with normative decision-making.

To address these questions, I conducted a controlled behavioral experiment in which participants made legal-style decisions under uncertainty. Participants were assigned roles in a simplified environment involving a potential harmful action and a third-party decision-maker responsible for determining whether to impose punishment. The decision-maker evaluated probabilistic evidence generated through an urn-based ball drawing paradigm and applied one of two legal standards of proof: preponderance of the evidence (PoE) or beyond a reasonable doubt (BaRD). The experimental design varied three key factors in a 2 x 2 x 2 between-subjects design: the standard of proof (PoE vs. BaRD), whether the standard was presented qualitatively or with an explicit numerical threshold added (quantified vs. unquantified), and whether correct decisions were financially incentivized (reward vs. no reward).

The results show that participants respond systematically to probabilistic evidence, as conviction rates increase with the posterior probability of guilt. However, participants do not consistently apply the thresholds implied by legal standards of proof. Individuals frequently convict when posterior probabilities fall below the relevant threshold and sometimes fail to convict when those thresholds are exceeded, resulting in both false convictions and false acquittals. Providing numerical thresholds does not improve decision-making and, under the PoE standard, increases both types of error while reducing the prevalence of threshold-consistent behavior. Incentives have little effect on performance, suggesting that deviations from normative decision-making are not primarily driven by lack of effort. Additionally, participants exhibit substantial heterogeneity in decision-making strategies, with only a subset applying consistent threshold-based rules while others rely on simpler heuristics or display inconsistent responses.

These findings suggest that individuals understand the direction of probabilistic evidence but struggle to translate that information into consistent decision thresholds. More broadly, the results highlight a gap between the theoretical role of standards of proof and their practical application, with implications for how such standards are communicated and implemented in real-world legal settings.

Keywords

Behavioral economics, standards of proof, Bayesian decision-making, probabilistic reasoning, legal decision-making, experimental economics

Access Type

Honors Thesis

Degree Type

Bachelor of Arts

Major

Economics

Second Major

Psychology

Minor, Emphasis, or Concentration

Philosophy

First Advisor

Dr. Rachel Landsman

Second Advisor

Dr. Jan Knoedler

Third Advisor

Dr. Matthew Baltz

Download

Share

COinS